A tRNA modification balances carbon and nitrogen metabolism by regulating phosphate homeostasis
- Institute for Stem Cell Science and Regenerative Medicine (inStem), India
- Rutgers University, United States
- Manipal Academy of Higher Education, India
Abstract
Cells must appropriately sense and integrate multiple metabolic resources to commit to proliferation. Here, we report that S. cerevisiae cells regulate carbon and nitrogen metabolic homeostasis through tRNA U34-thiolation. Despite amino acid sufficiency, tRNA-thiolation deficient cells appear amino acid starved. In these cells, carbon flux towards nucleotide synthesis decreases, and trehalose synthesis increases, resulting in a starvation-like metabolic signature. Thiolation mutants have only minor translation defects. However, in these cells phosphate homeostasis genes are strongly down-regulated, resulting in an effectively phosphate-limited state. Reduced phosphate enforces a metabolic switch, where glucose-6-phosphate is routed towards storage carbohydrates. Notably, trehalose synthesis, which releases phosphate and thereby restores phosphate availability, is central to this metabolic rewiring. Thus, cells use thiolated tRNAs to perceive amino acid sufficiency, balance carbon and amino acid metabolic flux and grow optimally, by controlling phosphate availability. These results further biochemically explain how phosphate availability determines a switch to a ‘starvation-state’.
https://doi.org/10.7554/eLife.44795.001eLife digest
The building blocks of all cells are made from a handful of chemical elements, including carbon, nitrogen, sulfur and phosphorus. To grow optimally, cells need to regulate their metabolism – in other words, the biochemical reactions that keep them alive – based on the availability of these elements. As a result, cells have evolved various mechanisms to sense when usable forms of these elements are present.
Proteins are chains of building blocks known as amino acids, which are assembled with the help of molecules called transfer ribonucleic acids, or tRNAs for short. Some of these molecules can be modified by attaching sulfur-containing chemical tags known as thiol groups to make “thiolated tRNAs”. Research has shown that, when there was more of an amino acid known as methionine around, the cells made more thiolated tRNA. These previous studies also suggested that mutant cells lacking thiolated tRNAs might have altered carbon and nitrogen metabolism. Yet, it remained unclear what exactly was leading to this metabolic rewiring.
Now, Gupta et al. have combined several biochemical and genetics approaches to study the role of thiolated tRNAs in yeast. The experiments revealed that mutant cells lacking thiolated tRNAs were unable to properly sense the levels of methionine and other amino acids, which are the cell’s major source of nitrogen. These mutant cells were also found to have a reduced level of phosphorous-containing compounds known as phosphates, which are involved in numerous biological processes.
Gupta et al. showed that reducing the level of phosphates caused carbon that is normally used to make chemicals required for growth to be re-routed towards making carbohydrates to store energy instead. This is similar to what happens when the cells are starving, showing that a ‘squeeze’ on internal phosphates metabolically rewires cells into a state that is like starvation.
These findings show how modified tRNAs can use the availability of amino acids to alter the cell’s metabolism by altering how much phosphate is present. In doing so, the thiolated tRNAs essentially allow the cell to decide whether it has enough of the right nutrients to grow. These findings may also have implications for human health, since errors in coordinating metabolism are responsible for certain medical conditions including several cancers.
Finally, technical challenges mean many questions remain unanswered about how phosphate levels are regulated within cells. These new findings point to a pressing need to understand phosphate metabolism as a prerequisite to better understand how cells regulate their overall metabolism.
https://doi.org/10.7554/eLife.44795.002Introduction
Cells utilize multiple mechanisms to sense available nutrients, and appropriately alter their internal metabolic state. Such nutrient-sensing systems assess internal resources, relay this information to interconnected biochemical networks, and control global responses that collectively reset the metabolic state of the cell, thereby determining eventual cell fate outcomes (Jeong et al., 2000; Förster et al., 2003; Zaman et al., 2008; Broach, 2012; Cai and Tu, 2012; Ljungdahl and Daignan-Fornier, 2012). However, much remains unknown about how cells sense and integrate information from multiple nutrient inputs, to coordinately regulate the metabolic state of the cell and commit to different fates.
In this context, the metabolic state of the cell is also closely coupled with mRNA translation. Protein synthesis is enormously energy consuming, and therefore must be carefully regulated in tune with nutrient availability (Warner, 2001). Generally, overall translational capacity and output increases during growth and proliferation (Jorgensen et al., 2004), and decreases during nutrient limitation (Wullschleger et al., 2006). Signalling processes that regulate translational outputs (such as the TORC1 and PKA pathways) are well studied (Wullschleger et al., 2006; Zaman et al., 2008; Broach, 2012; González and Hall, 2017). Notwithstanding this, little is known about how core components of the translation machinery might directly control metabolic outputs, and thus couple metabolic states with physiological cellular outcomes.
tRNAs are core components of the translation machinery, and are extensively modified post-transcriptionally (Björk et al., 1987; Phizicky and Hopper, 2010). Some tRNA modifications are required for tRNA folding, stability, or the accuracy and efficiency of translation (Phizicky and Hopper, 2010). However, the roles of many of these highly conserved modifications remain unclear. One such modification is a thiolation of uridine residue present at the wobble-anticodon (U34) position of specifically glu-, gln- and lys- tRNAs (s2U34) (Gustilo et al., 2008; Phizicky and Hopper, 2010). In yeast, this is mediated by a group of six enzymes- Nfs1, Tum1, Uba4, Urm1, Ncs2 and Ncs6, which are evolutionarily conserved (Nakai et al., 2008; Leidel et al., 2009; Noma et al., 2009). These enzymes incorporate a thiol group derived directly from an amino acid (cysteine), and replace the oxygen present at the 2-position of U34 with sulfur (Schmitz et al., 2008; Leidel et al., 2009; Noma et al., 2009). Surprisingly, these thiolated tRNAs appear to have a relatively minor role in general translation, as seen in multiple studies (Rezgui et al., 2013; Zinshteyn and Gilbert, 2013; Klassen et al., 2016; Chou et al., 2017) with modest roles in enhancing the efficiency of wobble base codon-anticodon pairing (Yarian et al., 2002; Rezgui et al., 2013).
Contrastingly, tRNA thiolation appears to directly alter cellular metabolism, but this connection has remained largely unexplored. The suggestive connections to metabolism come from disparate studies. The loss of tRNA thiolation results in hypersensitivity to oxidative agents, and the TORC1 inhibitor rapamycin (Fichtner et al., 2003; Goehring et al., 2003a; Goehring et al., 2003b; Laxman and Tu, 2011; Scheidt et al., 2014), suggesting a role for thiolated tRNAs in maintaining metabolic homeostasis. More pertinently, several studies have observed that a loss of this modification alters amino acid homeostasis, and the amino-acid starvation regulator Gcn4 is induced (Laxman et al., 2013; Zinshteyn and Gilbert, 2013; Nedialkova and Leidel, 2015). Further, thiolated tRNAs are required to maintain metabolic cycles in yeast (Laxman et al., 2013). Finally, the amounts of thiolated tRNAs reflect the intracellular availability of sulfur-containing amino acids (cysteine and methionine), and couple the sensing of amino acid sufficiency with growth (Laxman et al., 2013). These studies all hint that a core function of this tRNA modification may be to integrate the sensing of amino acid availability (primarily methionine/cysteine), with the coordinate regulation of overall metabolic state, in order for the cell to appropriately commit to growth. Yet, how thiolated tRNAs regulate metabolism, and the extent to which this may control cellular outcomes remains entirely unaddressed.
In this study, by directly analyzing different metabolic outputs, we identify the metabolic nodes that are altered in tRNA thiolation deficient cells. We find that tRNA thiolation regulates central carbon and nitrogen (amino acid) metabolic outputs, by controlling flux towards storage carbohydrates. In tRNA thiolation deficient cells, overall metabolic homeostasis is altered, with carbon flux diverted away from the pentose-phosphate pathway and nucleotide synthesis axis, and towards storage carbohydrates trehalose and glycogen. This thereby alters cellular commitments towards growth and cell cycle progression. Counter-intuitively, we discover that this metabolic-state switch in cells lacking tRNA thiolation is achieved by down-regulating a distant metabolic arm of phosphate homeostasis. We biochemically elucidate how regulating phosphate balance can couple amino acid and carbon utilization towards or away from nucleotide synthesis, and identify trehalose synthesis as a pivotal control point for this metabolic switch. Through these findings we show how tRNA thiol-modifications regulate overall metabolic homeostasis, integrating nutrient inputs to enable optimal growth. We further present a general biochemical explanation for how inorganic phosphate homeostasis regulates commitments to different arms of carbon and nitrogen metabolism, thereby determining how cells commit to a ‘growth’ or ‘starvation’ state.
Results
Amino acid and nucleotide metabolism are decoupled in tRNA thiolation deficient cells
Figure 1 with 2 supplements see all
Amino acid and nucleotide metabolism are decoupled in tRNA thiolation deficient cells.
(A) Intracellular pools of amino acids are increased in tRNA thiolation mutants. Steady-state amino acid amounts were measured in wild-type (WT) and tRNA thiolation mutant cells (uba4Δ and ncs2Δ) grown in minimal media by targeted liquid chromatography/mass spectrometry (LC-MS/MS). Amino acid levels in tRNA thiolation mutant cells relative to WT are plotted, where levels in WT were set to 1. * denotes statistical significance (Student’s t-test), comparing tRNA thiolation mutant cells (uba4Δ and ncs2Δ) to WT. Data are displayed as mean ± SD, n >= 3. *p<0.05, **p<0.01, ***p<0.001. (B) A schematic representation illustrating the induction of Gcn4 translational upon amino acid starvation, as mediated by the Gcn2 kinase, and phosphorylation of the eIF2α initiation factor. (C) Gcn4 protein is increased in tRNA thiolation mutants. Western blots indicating Gcn4 protein levels (Gcn4 tagged with HA epitope at the endogenous locus) in WT and tRNA thiolation mutant cells (uba4Δ, ncs2Δ and ncs6Δ) grown in minimal media, as detected using an anti-HA antibody. A representative blot obtained from three biological replicates (n = 3) is shown. Also see Figure 1—figure supplement 1. (D) A schematic representation of de novo nucleotide (purine and pyrimidine) biosynthesis from its precursors- amino acids, the pentose phosphate pathway and PRPP (5-Phopsphoribosyl-1-Pyrophosphate), and the folate/one-carbon pathway. (E) Nucleotide synthesis is decreased in tRNA thiolation mutants (nitrogen label). WT and tRNA thiolation mutant cells (uba4Δ and ncs2Δ) grown in minimal media were pulse-labelled with 15N2-labelled ammonium sulfate for 90 min to measure newly synthesized nucleotides (GMP, AMP and CMP) using targeted LC-MS/MS. The incorporation of 15N atoms from 15N2-labelled ammonium sulfate into nucleotides is represented as M + n, where n is the number of 15N-labelled atoms. Label incorporation in tRNA thiolation mutant cells relative to WT is plotted, where label incorporation in WT was set to 100. * denotes statistical significance (Student’s t-test), comparing tRNA thiolation mutant cells (uba4Δ and ncs2Δ) to WT. Data are displayed as means ± SD, n = 3 for AMP and CMP, n = 2 for GMP. *p<0.05, **p<0.01, ***p<0.001. Also see Figure 1—figure supplement 2A,B and D.
Earlier studies had observed an increased expression of amino acid biosynthetic genes, and an activation of the amino acid starvation responsive transcription factor Gcn4, in cells lacking tRNA thiolation (Laxman et al., 2013; Zinshteyn and Gilbert, 2013; Nedialkova and Leidel, 2015). These studies therefore suggested that tRNA thiolation-deficient cells were amino-acid starved. We investigated this surmise, by directly measuring free intracellular amino acids in wild-type (WT) and tRNA thiolation mutant cells, using two distinct thiolation pathway mutants (uba4Δ and ncs2Δ). In our studies, we used a prototrophic yeast strain grown in synthetic minimal medium without supplemented amino acids, in order to minimize any confounding interpretations coming from supplemented amino acids/nucleobases in the medium. Using quantitative, targeted LC-MS/MS approaches, we compared relative amounts of amino acids in WT and thiolation mutants. Unexpectedly, we observed a substantial increase in the intracellular pools of free amino acids in thiolation mutants (Figure 1A). This shows that the thiolation deficient cells are not amino acid starved, but instead accumulate amino acids. We next correlated these actual amino acid amounts with the abundance of Gcn4. Gcn4 is the major amino acid starvation responsive transcription factor, and is induced upon amino acid starvation (Hinnebusch, 1984; Hinnebusch, 2005) (Figure 1B). We measured Gcn4 protein in WT and thiolation deficient cells, and contrarily observed increased Gcn4 protein in thiolation mutants (Figure 1C). Further, GCN4 translation was correspondingly higher in thiolation mutants (Figure 1—figure supplement 1A) (as also seen earlier in Zinshteyn and Gilbert, 2013; Nedialkova and Leidel, 2015). This increased GCN4 translation in the thiolation mutants was also Gcn2- and eIF2α phosphorylation-dependent (Figure 1—figure supplement 1B and C). These observations comparing actual amino acid amounts in cells with the activity of Gcn4 therefore present a striking paradox. As canonically understood, Gcn4 is induced upon amino acid starvation, while Gcn4 translation and protein decrease when intracellular amino acid amounts are restored (Hinnebusch, 1984; Hinnebusch, 2005). Contrastingly, in the results observed here, despite the high amino acid amounts present in the tRNA thiolation mutants, the Gcn2-Gcn4 pathway remains induced. We therefore concluded that the metabolic node regulated by tRNA thiolation, resulting in an apparent amino acid starvation signature, cannot be at the level of amino acid biosynthesis and availability.
We therefore considered the possible metabolic outcomes of amino acid utilization, and hypothesized that an alteration in amino acid utilization could be a source of this metabolic rewiring. In particular, amino acids are the sole nitrogen donors for de novo nucleotide synthesis (Figure 1D). Since amino acids accumulated in thiolation mutants, we explored the possibility that this was due to reduced de novo nucleotide synthesis. We first measured steady-state nucleotides in WT and thiolation mutants, and observed decreased steady-state levels of nucleotides (NMPs, as well as ATP) in thiolation mutants (Figure 1—figure supplement 2A and B). To unambiguously determine if these decreased nucleotide amounts in the thiolation mutants were due to reduced nucleotide synthesis, we adopted a metabolic flux-based approach we had developed earlier (Walvekar et al., 2018b). In such an approach, 15N-labelled ammonium sulfate can be provided as a pulse to cells growing with ammonium sulfate as a sole nitrogen source, and label incorporation via glutamine and aspartate into newly formed nucleotides can be measured. Notably, the incorporation of 15N-label into nucleotides (GMP, AMP and CMP) decreased in thiolation mutants relative to WT cells (Figure 1E), indicating reduced flux towards nucleotide synthesis. As an internal control, the 15N-label incorporation into amino acids (aspartate and glutamine) themselves were not affected in thiolation mutants (Figure 1—figure supplement 2C), ruling out amino acid synthesis defects. These data therefore show that nitrogen incorporation from amino acids to nucleotides decrease in the thiolation mutants, resulting in decreased nucleotides. We further addressed this pharmacologically, using a purine-analog, 8-azaadenine, which acts as a pseudo-feedback inhibitor of nucleotide biosynthesis. Consistent with the decreased nucleotide levels observed, thiolation mutants exhibited increased sensitivity to 8-azaadenine (Figure 1—figure supplement 2D). Collectively, these data show that the loss of tRNA thiolation decreases nucleotide biosynthesis, with a corresponding accumulation of amino acids. Notably, early studies have shown that nucleotide limitation can itself directly induce Gcn4 (Rolfes and Hinnebusch, 1993), suggesting that the increased Gcn4 amounts could be due to this. We also asked if the decreased nucleotide synthesis was due to reduced expression of nucleotide biosynthetic genes. We observed that the expression of candidate genes in this pathway were increased in thiolation mutants (Figure 1—figure supplement 2E), diminishing the possibility of reduced nucleotide biosynthetic capacity as a reason for decreased nucleotides. Indeed, increased mRNA levels of nucleotide biosynthetic genes observed in thiolation mutants may be due to feedback upregulation in response to reduced nucleotides, which is also a well-established phenomenon (Davis and Ares, 2006; Kwapisz et al., 2008).
Collectively, despite increased intracellular pools of amino acids, tRNA thiolation deficient cells exhibit signatures of amino acid starvation, including decreased nucleotide biosynthesis. These data therefore suggest that the tRNA thiolation pathway is important for cells to appropriately balance amino acid utilization for nucleotide synthesis.
Carbon flux is routed towards storage carbohydrates in thiolation mutants
Despite amino acids being non-limiting in thiolation deficient cells, flux towards nucleotide synthesis was decreased. This observation was in itself puzzling, and the reason was not obvious. We therefore asked if carbon metabolism was instead rewired in the thiolation mutants. Our reasoning was as follows: while amino acids are the sole nitrogen donors for nucleotide synthesis, the carbon backbone for nucleotides is derived from central carbon metabolism (Figure 2A). We reasoned that since the decreased nucleotide synthesis was not due to amino acid limitation, this could instead be due to a metabolic shift where carbon flux is routed away from nucleotide synthesis. Carbon derived from glucose is converted to glucose-6-phosphate, and then is typically directed towards glycolysis and the pentose phosphate pathway (PPP). The PPP, along with one-carbon folate metabolism provides the necessary carbon precursors (including ribose sugars) for nucleotide synthesis (Figure 2A) (Boyle, 2005; Hosios and Vander Heiden, 2018). However, if glucose-6-phosphate is instead diverted towards storage carbohydrates trehalose and glycogen (Figure 2A), this can result in reduced flux into the arm of glucose metabolism leading towards nucleotide synthesis. To assess if such a metabolic rewiring might happen in tRNA thiolation mutants, we pulsed [U-13C6]-labelled glucose to growing WT or thiolation deficient cells, and measured label incorporation into nucleotides as the end-point readout. Here, labelled carbons will only be present in newly synthesized nucleotides, and the label can only come from the pulsed labelled glucose, through the PPP and one-carbon metabolic pathways (Figure 2A). We observed significantly decreased carbon label incorporation towards new nucleotide synthesis, as shown for GMP and AMP, as well as label incorporation into ADP and ATP in the thiolation mutants (Figure 2B, Figure 2C, and Figure 2—figure supplement 1A). This result is also consistent with the decreased nucleotide synthesis based on amino acid derived nitrogen assimilation, observed earlier (Figure 1E). Experimental note: Given how rapidly the pulsed carbon label saturates in glucose medium for early glycolytic and PPP intermediates (Heerden et al., 2014), using nucleotide synthesis as a read-out of this arm of carbon metabolism is a more reliable, quantitative indicator of carbon flux. Here, we reliably obtained nucleotide information (ensuring that label incorporation was not saturated) by pulsing cells with 13C-glucose and quenching/processing metabolites within 5 min. By this time, the 13C-glucose label incorporation into glycolytic and pentose phosphate pathway intermediates was already near-saturation, and hence differences in these metabolites could not be seen (Figure 2—figure supplement 1B). We therefore carried out experiments where cells were quenched and metabolites extracted within 2 min of 13C-glucose addition. Here, a significant decrease in 13C-label incorporation into newly synthesized ribose-5-phosphate (a late PPP metabolite) was observed (Figure 2D). This is entirely consistent with results obtained with nucleotides in Figure 2B and C. Summarizing, these data show that carbon (glucose) flux towards nucleotide synthesis was reduced in the thiolation mutants.
Figure 2 with 2 supplements see all
Carbon flux is routed towards storage carbohydrates in thiolation mutants.
(A) Schematic representation depicting nucleotide and storage carbohydrates (trehalose and glycogen) biosynthesis, starting from a common precursor, glucose-6-phosphate. In normal conditions where carbon and nitrogen are not limiting, flux is typically higher towards the pentose phosphate pathway (ribose-5-phosphate), and eventually nucleotides. (B) Nucleotide synthesis is decreased in tRNA thiolation mutant (carbon label). WT and tRNA thiolation mutant cells (ncs2Δ) grown in minimal media were pulse-labelled with [U-13C6]-labelled glucose for 10 min, and quenched, and newly synthesized nucleotides (GMP, AMP) were measured, using LC-MS/MS. Percent label incorporation in WT and tRNA thiolation mutant cells was plotted. The incorporation of 13C atoms from [U-13C6]-labelled glucose into nucleotides is represented as M + n, where n is the number of 13C-labelled atoms (with all five carbons labelled in the ribose sugar). * denotes statistical significance (Student’s t-test) of relevant data comparing tRNA thiolation mutant cells (ncs2Δ) to WT. Data are displayed as means ± SD, n = 3 for GMP and n = 2 for AMP. *p<0.05, **p<0.01. Also see Figure 2—figure supplement 1A. (C) ADP and ATP synthesis is decreased in the thiolation mutant. WT and tRNA thiolation mutant cells (uba4Δ) grown in minimal media were pulse-labelled with [U-13C6]-labelled glucose for 10 min, quenched, and newly synthesized ADP and ATP were measured, using LC-MS/MS. Relative label incorporation in WT and tRNA thiolation mutant cells are shown, where label amounts in WT cells is set to 1. The incorporation of 13C atoms from [U-13C6]-labelled glucose into nucleotides is represented as M + n, where n is the number of 13C-labelled atoms (with all five carbons labelled in the ribose sugar). * denotes statistical significance (Student’s t-test) of relevant data comparing tRNA thiolation mutant cells (ncs2Δ) to WT. Data are displayed as means ± SD, n = 3. *p<0.05, **p<0.01. (D) PPP flux is decreased in the tRNA thiolation mutant. WT and tRNA thiolation mutant cells (uba4Δ) grown in minimal media were pulse-labelled with [U-13C6]-labelled glucose for 2 min, quenched and collected, and early (glucose-6-phosphate) and late (ribose-5-phosphate) PPP intermediates were measured by LC-MS/MS. Relative 13C6-label incorporation into these metabolites are shown. Data are displayed as means ± SD, n = 2. *p<0.05, **p<0.01, Student’s t-test. Also see Figure 2—figure supplement 1B. (E) Steady-state trehalose and glycogen amounts are increased in tRNA thiolation mutants. Trehalose and glycogen content of WT and tRNA thiolation mutant cells (uba4Δ and ncs2Δ) grown in minimal media was plotted. * denotes statistical significance (Student’s t-test), comparing tRNA thiolation mutant cells (uba4Δ and ncs2Δ) to WT. Data are displayed as means ± SD, n = 4 biological replicates with two technical replicates each for trehalose, and n = 3 biological replicates with two technical replicates each for glycogen. ***p<0.001, ****p<0.0001. Also see Figure 2—figure supplement 1C. (F) Trehalose synthesis is increased in tRNA thiolation mutant. WT and tRNA thiolation mutant cells (ncs2Δ) grown in minimal media were pulse-labelled with [U-13C6]-labelled glucose for 5 min, quenched, and label incorporation into newly synthesized trehalose was measured using targeted LC-MS/MS. Relative label incorporation into trehalose in WT and tRNA thiolation mutant cells is shown, where label in WT M + 6 is set to 1. The incorporation of 13C atoms from [U-13C6]-labelled glucose into trehalose is represented as M + n, where n is the number of 13C-labelled atoms. * denotes statistical significance (Student’s t-test), comparing tRNA thiolation mutant cells (ncs2Δ) to WT. Data are displayed as means ± SD, n = 3. *p<0.05, **p<0.01. (G) Thiolation mutants exhibit reduced methionine-induced carbon incorporation into nucleotides. WT and tRNA thiolation mutant cells (uba4Δ) were grown in standard minimal media, 2% glucose, with or without 2 mM methionine (met) supplemented. Cells were pulse-labelled with [U-13C6]-labelled glucose for 15 min, and metabolites extracted. Here, carbon incorporation into newly synthesized nucleotides (AMP) was measured, using LC-MS/MS. The relative label incorporation in WT and tRNA thiolation mutant cells with or without methionine is shown, where label in WT in the absence of methionine is set to 1. Two independent biological replicates (replicate 1 and 2) are shown. The relative label incorporation in WT and tRNA thiolation mutant cells growing in standard minimum medium (without methionine supplementation) is shown as an inset, to suitably illustrate the strong induction in nucleotide synthesis due to methionine supplementation. The incorporation of 13C atoms from [U-13C6]-labelled glucose into nucleotides is represented as M + n, where n is the number of 13C-labelled atoms (with all five carbons labelled in the ribose sugar). Also see Figure 2—figure supplement 2D.
The end-point readouts of the alternative metabolic arm where glucose-6-phosphate is diverted away from the PPP are trehalose and glycogen (as shown in Figure 2A). To examine if this arm of metabolism is altered in the thiolation mutants, thereby resulting in the decreased carbon flux towards nucleotides (shown earlier), we first estimated the steady-state amounts of trehalose and glycogen with a biochemical assay. Here, we observed a marked increase in these metabolites in the thiolation mutants (Figure 2E and Figure 2—figure supplement 1C). Subsequently, we directly estimated flux towards trehalose synthesis. For this, we used a similar experiment as described earlier, where [U-13C6]-labelled glucose was pulsed and newly formed labelled trehalose was measured in a metabolic flux experiment, to test if this arm of the pathway is altered. Here, we observed a strong increase in the synthesis of trehalose (as measured by label incorporation into M + 6 and M + 12 mass isotopomers of trehalose) in the tRNA thiolation mutants (Figure 2F). Collectively, these results show that cells lacking tRNA thiolation rewire metabolic outputs towards the synthesis of storage carbohydrates, and away from nucleotide synthesis, suggesting a ‘starvation-like’ metabolic state. Notably, this occurs despite the absence of glucose (carbon) or amino acid (nitrogen) limitation in the thiolation mutants.
Sulfur starvation phenocopies the metabolic state of thiolation mutants
Earlier studies have noted a coupling of tRNA thiolation with methionine/sulfur amino acid availability (Laxman et al., 2013). Here, the amount of thiolated tRNAs increase with increasing sulfur amino acids and vice versa. Further, in thiolation mutants, proteins involved in methionine salvage and biosynthesis increase even when methionine was in excess, suggesting a mis-sensing of this amino acid in these mutants. Finally, Uba4 is itself regulated by methionine amounts, decreasing sharply with slight methionine-limitation (Laxman et al., 2013). Separately, studies (Tu et al., 2005; Laxman et al., 2013) show that the amounts of thiolated tRNAs are highest in cells entering a ‘growth state’ (with high ribosomal biosynthesis). In this context, we also recently defined a methionine induced anabolic program, where abundant methionine triggers increased carbon flux (particularly PPP flux) towards nucleotide synthesis (Walvekar et al., 2018b). Given our observations here in standard medium, where thiolation mutants showed decreased carbon flux from glucose towards nucleotides, we further investigated this coupling of tRNA thiolation with methionine availability. The prediction is that if tRNA thiolation enables cells to fully respond to abundant methionine, then thiolation mutants will exhibit a reduced methionine response. As a result, when methionine is supplemented, thiolation mutants will show reduced carbon incorporation into nucleotides (i.e. nucleotide synthesis), compared to WT cells. To test this, WT or thiolation mutant (uba4Δ) cells grown in standard glucose minimal medium were supplemented with 2 mM methionine, and then pulsed with 13C-glucose (as described earlier) for 15 min. Subsequently, we compared carbon-label incorporation into newly synthesized nucleotides. Here, we expectedly observed a sharp increase in carbon-label incorporation into nucleotides in WT cells supplemented with methionine (Figure 2G, and Figure 2—figure supplement 1D). Notably, while the extent of this methionine-dependent induction of carbon flux into nucleotides was significantly reduced, it was not completely abolished in the thiolation mutant (uba4Δ), indicating the involvement of additional as yet unknown pathways (Figure 2G, and Figure 2—figure supplement 1D). Together, these data reiterate that tRNA thiolation allows cells to fully integrate amino acid (methionine) sensing, with a routing of carbon (glucose) flux towards new nucleotide synthesis, collectively indicative of a growth state. Furthermore, in a converse experiment, we subjected WT cells to brief inorganic sulfur/sulfur amino acid limitation, to determine the metabolic signature of cells in this condition (experimental design shown in Figure 2—figure supplement 2A). Cells subjected to a brief shift to sulfur starved medium showed a sharp decrease in sulfur amino acid-related metabolites (Figure 2—figure supplement 2B). We measured the steady-state amounts of other amino acids, nucleotides (AMP), trehalose and the level of Gcn4 translation. We observed increased steady-state amino acid amounts (Figure 2—figure supplement 2C), reduced nucleotides (Figure 2—figure supplement 2D), increased trehalose levels (Figure 2—figure supplement 2E), and strong Gcn4 induction (Figure 2—figure supplement 2F) upon sulfur limitation. These data strikingly resembled the metabolic state of the thiolation mutants. Thus, WT cells subject to sulfur amino acid limitation phenocopied the metabolic signature of tRNA thiolation mutants.
tRNA thiolation couples cellular metabolic state with normal cell cycle progression
Dissecting physiological roles of such a fine-tuning of metabolic outputs can be challenging, and this has been the case for tRNA thiolation mutants. However, a simple yeast system, termed ‘yeast metabolic cycles’ or metabolic oscillations, has been effective in identifying regulators that couple metabolism with cell growth/cell-division (Tu et al., 2005; Slavov and Botstein, 2011). In continuous, glucose-limited cultures, yeast cells exhibit robust metabolic oscillations, which are tightly coupled to the cell division cycle, and where DNA replication and cell division are restricted to a single temporal phase (Tu et al., 2005; Chen et al., 2007). In an earlier study we had observed that tRNA thiolation mutants exhibit abnormal metabolic cycles (Laxman et al., 2013). This was reminiscent of phenotypes exhibited by mutants of cell division cycle regulators (Chen et al., 2007). We therefore asked if tRNA thiolation coupled metabolic and cell division cycles. To test this, we sampled the cells at regular intervals of time during the metabolic cycles, and determined their budding index. While WT cells showed synchronized cell cycle progression, tRNA thiolation mutants showed asynchronous cell division (Figure 3A), suggesting a de-coupling of metabolic and cell division cycles. Given our earlier data showing a metabolic rewiring away from nucleotide synthesis in thiolation mutants, we hypothesized that tRNA thiolation controlled normal cell cycle progression by regulating the balance between nucleotide synthesis, and storage carbohydrate synthesis.
Figure 3 with 1 supplement see all
tRNA thiolation couples cellular metabolic state with normal cell cycle progression.
(A) tRNA thiolation mutants exhibit asynchronous cell division and disrupted yeast metabolic cycles. WT and tRNA thiolation mutant cells (uba4Δ) growing in chemostat cultures under conditions of normal yeast metabolic cycles, were sampled at 20 min time intervals. Percentage of budded cells represents the fraction of cells in S/G2/M phases of the cell cycle. At least 200 cells were analysed for each time point, for the respective strains. Inset: the oxygen consumption profiles of WT and tRNA thiolation mutant metabolic cycles are represented. Note that the metabolic cycles of thiolation mutants are disrupted. (B) tRNA thiolation mutants show delayed cell cycle progression. The DNA content of WT and tRNA thiolation mutant cells (uba4Δ and ncs2Δ) grown in minimal media, during G1 arrest (0 min) and after release from G1 arrest (30, 60 and 90 min) was determined by flow cytometry analysis, and is presented. (C and D) Cell cycle duration, distribution of G1 phase duration (red) and S/G2/M phase durations (green) for WT and tRNA thiolation mutant cells (uba4Δ and ncs2Δ) were measured in single cells using time lapse live-cell microscopy. The G1 duration (starting from the time of complete division of mother and daughter cells to bud emergence), and S/G2/M durations (from the time of bud emergence to complete division of mother and daughter cells) were determined for only first generation mother cells. At least 100 cells were analysed for each strain. * denotes statistical significance (Student’s t-test), comparing tRNA thiolation mutant cells (uba4Δ and ncs2Δ) to WT. ns denotes non-significant difference. ****p<0.0001. (E) tRNA thiolation mutants exhibit increased HU sensitivity. WT and tRNA thiolation mutant cells (uba4Δ and ncs2Δ) grown in minimal media were spotted on minimal media agar plates containing 150 mM HU. Also see Figure 3—figure supplement 1A and B.
To test this directly, we arrested cells in G1-phase using alpha factor, synchronously released them into the cell cycle by washing away the alpha factor, and monitored cell cycle progression by flow cytometry (Figure 3B). 30 min post-release from G1-arrest, we observed delayed cell cycle progression and accumulation of cells in the S-phase in tRNA thiolation deficient cells. Further, using time lapse live-cell microscopy we found that the duration of the S-G2/M phase was longer in thiolation mutants (Figure 3C and Figure 3D). Since nucleotides are required for DNA replication during the S-phase of the cell cycle, we reasoned that that this S-phase delay is due to decreased flux towards nucleotide synthesis in thiolation mutants (shown earlier). To investigate this, we examined the sensitivity of WT and thiolation deficient cells to hydroxyurea (HU), which inhibits the ribonucleotide reductase (RNR) enzyme and arrests cells in the S-phase. Thiolation mutants exhibited increased sensitivity to HU (Figure 3E and Figure 3—figure supplement 1A). As a control, to rule out any abnormal activation of DNA damage readouts in the thiolation mutants as a cause for this phenotype, we also examined activity of the Rad53 checkpoint pathway. Indeed, the observed HU sensitivity was not due to any defect in the activation of the Rad53 checkpoint pathway in thiolation mutants (Figure 3—figure supplement 1B).
Summarizing, these results show that tRNA thiolation-mediated regulation of metabolic homeostasis, leading towards regulated nucleotide synthesis, is required for appropriately coupling metabolic state with normal cell cycle progression.
Loss of tRNA thiolation results in reduced phosphate homeostasis (PHO)-related transcripts and ribosome-footprints
Thus far, it remains unclear why the loss of tRNA thiolation results in this distinct metabolic switch, where carbon and amino acid flux is diverted away from nucleotide synthesis and into storage carbohydrates. This therefore suggests a deeper, non-intuitive regulatory check-point underpinning the overall metabolic rewiring towards a ‘starvation-like’ state in tRNA thiolation mutants. In order to identify what this controlling bottleneck might be, we identified transcriptional and translational changes in thiolation mutants relative to WT by performing RNA-seq and Ribo-seq (Ingolia et al., 2009) based on methods described earlier (Weinberg et al., 2016; McGlincy and Ingolia, 2017). Notably, as detailed in the Materials and methods section, while collecting cells for RNA and Ribo-seq, we avoided the use of cycloheximide, as described earlier (Weinberg et al., 2016), to minimize biases in our interpretations of ribosome profiling datasets due to the use of this translational inhibitor as described elsewhere (Hussmann et al., 2015). Instead, cells were rapidly harvested by filtration and lysed as described. Ribosome profiling datasets were generated for WT cells as well as two distinct thiolation pathway mutants (uba4Δ and ncs2Δ), in biological triplicates. The correlations among the biological replicates of ribosome footprint reads, and also among RNA-seq reads were all excellent (R > 0.97), as shown in Figure 4—figure supplement 1A). Further, figures (Figure 4—figure supplement 1B; Figure 4—figure supplement 1C) show transcript and ribosome footprint read correlations (R2 in all comparisons > 0.8), as well as read-length distributions.
Using these datasets, we compared global gene expression, as well as ribosome footprints of WT cells with the uba4Δ and ncs2Δ thiolation mutants (Figure 4A). Notably, comparing WT cells with the thiolation mutants (uba4Δ and ncs2Δ), we surprisingly found exceptional correlation for transcripts, as well as ribosome footprints (R2 >0.97, and p<=2.2×10−16 for all datasets) (Figure 4A and B). These data surprisingly revealed that there are very little gene expression or translation changes observed in the thiolation mutants (complete data in Supplementary file 4). Fewer than ~30 genes were up or downregulated at a two-fold change cutoff (arbitrarily used to illustrate the point), compared to WT cells (Figure 4A and B), with a false discovery rate (FDR) of 0.05. Furthermore, we observe only modest increases in ribosome-densities at codons recognized by thiolated tRNAs – AAA, CAA and GAA in the uba4Δ and ncs2Δ cells (Figure 4—figure supplement 2A). Collectively, these extensive analysis show that the loss of tRNA thiolation has minimal effects on translational outputs in vivo, and any of the changes observed in the translation rates largely corresponded to changes at the transcriptional level.
Figure 4 with 3 supplements see all
Loss of tRNA thiolation results in reduced phosphate homeostasis (PHO)-related transcripts and ribosome-footprints.
(A) Correlation plots are shown, for WT and tRNA thiolation mutant cells (ncs2Δ and uba4Δ), for both gene expression (transcript) and ribosome footprint (translation) changes. The coefficients of determination (R2) values are indicated, along with their significance (p values). Note: Very few differentially regulated genes were observed, and these are indicated as red points. Also see Figure 4—figure supplement 1. (B) Correlation plots comparing ribosome densities (RPF TPM/RNA TPM) for the thiolation mutants (ncs2Δ or uba4Δ respectively) with WT cells. Note: overall ribosome densities for thiolation mutants correlate exceptionally well with ribosome densities in WT cells, with the correlation coefficients (R) > 0.88 in both comparisons. Also see Figure 4—figure supplement 1 for correlations between biological replicates of the same genetic background. (C) A density plot, representing changes in expression (transcript and ribosome footprints) of genes associated with amino acid biosynthetic pathways. In general, amino acid biosynthesis genes were upregulated in tRNA thiolation mutant cells (uba4Δ and ncs2Δ). Also see Figure 4—figure supplement 3C. (D) A density plot, representing changes in expression (transcript and ribosome footprints) of genes associated with trehalose and glycogen biosynthetic pathways, and the pentose phosphate pathway (PPP). In general, trehalose, glycogen biosynthesis and PPP genes were unchanged in tRNA thiolation mutant cells (uba4Δ and ncs2Δ), compared to WT cells. Also see Figure 4—figure supplement 3C (E) Heat map depicting changes in expression (transcript and ribosome footprints) of genes associated with the phosphate (PHO) regulon, in the tRNA thiolation mutant cells (uba4Δ and ncs2Δ), compared to WT cells, at the transcript and ribosome-footprint levels (log2 2-fold changes). (F) A density plot with a statistical comparison between WT and tRNA thiolation mutants, for changes in transcript or ribosome footprints of all genes in the genome, or only the PHO regulon. Note: p<10−7 for all the comparison datasets. Also see Figure 4—figure supplement 3C.
Given the lack of large-scale changes at the transcriptional and translational levels, but robust changes in cellular metabolism in the thiolation mutants, we focused our analysis on changes in expression levels of genes involved in metabolic pathways. We first examined several general amino acid control (GAAC) response genes, including the Gcn4 targets in amino acid biosynthetic pathways. Expectedly, we found these to be transcriptionally upregulated in the thiolation deficient cells (Figure 4C, and Figure 4—figure supplement 3C). Additionally, as expected, we observed increase in GCN4 translation itself in the thiolation mutants, with no change in GCN4 transcripts (Figure 4—figure supplement 2B and C). These data collectively corroborate our earlier data from Figure 1, and agrees with previous reports (Zinshteyn and Gilbert, 2013; Nedialkova and Leidel, 2015). Also consistent with our earlier data, most nucleotide biosynthesis genes showed an increase in mRNA and ribosome-footprint abundances in the thiolation deficient cells (Figure 4—figure supplement 3A; Figure 4—figure supplement 3C). In these datasets, there were also no obvious changes in central carbon metabolism genes in thiolation mutants. Notably, genes related to either trehalose/glycogen biosynthesis, or the PPP, showed negligible changes in either mRNA or ribosome-footprint abundances in the thiolation mutants (Figure 4D, and Figure 4—figure supplement 3C). This further reiterates that the rewiring observed in thiolation deficient cells was largely driven by metabolic flux. Finally, we also found small decreases in the transcription and translation rates of all large and small subunit genes of ribosomes in uba4Δ and ncs2Δ cells at the translational level (Figure 4—figure supplement 3B and C), consistent with earlier observations (Laxman et al., 2013; Nedialkova and Leidel, 2015).
However, in the course of this extensive functional analysis, we observed that an unusual group of ~20 genes were strongly downregulated in the thiolation mutants, both at the transcript and ribosome-footprint levels (Figure 4E). Although these do not obviously group into a single category based on gene-ontology (GO), we noted that these were functionally related to an important metabolic node. These genes are all part of the PHO regulon, which regulates phosphate homeostasis in cells (Ljungdahl and Daignan-Fornier, 2012; Secco et al., 2012). This downregulation of these PHO-related genes was exceptionally significant (p<10−7), compared to other genes across the genome, both at the level of transcript abundances and ribosome-footprints (Figure 4E, Figure 4F, and Figure 4—figure supplement 3C). Also notably, the only unaltered gene transcripts/ribosome footprints in the PHO regulon in the thiolation mutants, viz. PHO2, PHO4, PHO80, PHO85, PHO87, PHO90 and PHO91, are transcription factors, cyclins/cyclin-dependent kinases or low affinity phosphate transporters that are not transcriptionally/translationally regulated, but are regulated at the level of their activity (Lemire et al., 1985; Toh-e and Shimauchi, 1986; Madden et al., 1988; Yoshida et al., 1989; Madden et al., 1990; Schneider et al., 1994; Ogawa et al., 1995; Lenburg and O'Shea, 1996; Auesukaree et al., 2003). Finally, we also observed that some genes related to phospholipid metabolism were downregulated in the thiolation mutants (Figure 4—figure supplement 3C). Collectively, these data unexpectedly revealed a strong downregulation of genes related to phosphate homeostasis in the tRNA thiolation mutants.
tRNA thiolation mutants exhibit a dampened PHO response
Inorganic phosphate (Pi) homeostasis is complex, but critical for overall nutrient homeostasis (Ljungdahl and Daignan-Fornier, 2012; Secco et al., 2012). The PHO regulon comprises of several genes that respond to phosphate starvation, and maintains internal phosphate levels by balancing transport of Pi from the external environment, from within vacuolar stores, and the nucleus (Figure 5A) (Ljungdahl and Daignan-Fornier, 2012; Secco et al., 2012). Extensive studies have defined global cellular responses to phosphate limitation (Ogawa et al., 2000; Wykoff and O'Shea, 2001; Boer et al., 2003; Boer et al., 2010; Saldanha et al., 2004; Gresham et al., 2011; Levy et al., 2011; Choi et al., 2017; Gurvich et al., 2017). In general, the PHO response is very sensitive to phosphate limitation, and is induced rapidly to restore internal phosphate levels. Extended phosphate starvation switches cells to an overall metabolically starved state. Our observed reduction in PHO-related transcripts and ribosome-footprints in the tRNA thiolation mutants was striking. We therefore first biochemically validated our results from the ribosome-profiling data, to more systematically investigate the extent of PHO downregulation. For this, we first measured protein amounts of Pho12 and Pho84 (two arbitrarily selected PHO genes which are downregulated in the thiolation mutants, as shown in Figure 4E), in WT cells and thiolation mutants grown in the same conditions used for ribosome profiling analysis. Amounts of these proteins were substantially reduced in uba4Δ and ncs2Δ cells (Figure 5B). We further compared Pho12 and Pho84 protein levels in WT cells and thiolation mutants under conditions of phosphate-limitation. Interestingly, we observed that even under these PHO inducible conditions, thiolation mutants exhibited substantially reduced levels of these PHO proteins compared to WT cells (Figure 5C and Figure 5—figure supplement 1), although the PHO response was induced. This suggested a constitutive dampening (but not shut-down and absence) of the PHO response in the thiolation mutants.
Figure 5 with 1 supplement see all
tRNA thiolation mutants exhibit a dampened PHO response.
(A) A schematic representation of PHO regulon-related genes, and their roles. Pho84 and Pho89 are high affinity phosphate transporters, Pho87 and Pho90 are low affinity phosphate transporters, Spl2 is a negative regulator of low affinity phosphate transporters, Pho3, Pho5, Pho11 and Pho12 are secreted acid phosphatases, Pho80-Pho85 is a cyclin-dependent kinase (CDK) complex, Pho81 is a CDK inhibitor, Pho2 and Pho4 are transcription factors, Vtc1, Vtc3 and Vtc4 are involved in vacuolar polyphosphate accumulation, Pho91 is a vacuolar phosphate transporter, Ddp1 and Ppn1 are polyphosphatases. Proteins highlighted in red are down-regulated in tRNA thiolation mutant cells (uba4Δ and ncs2Δ). (B) Pho12 and Pho84 proteins are decreased in tRNA thiolation mutants. Pho12 and Pho84 protein levels (Pho12 and Pho84 tagged with FLAG epitope at their endogenous loci) in WT and tRNA thiolation mutant cells (uba4Δ and ncs2Δ) grown in minimal media were detected by Western blot analysis using an anti- FLAG antibody. A representative blot is shown (n = 3). (C) Pho84 protein is decreased in tRNA thiolation mutants in both high and low Pi conditions. Pho84 protein levels (the carboxy-terminus of Pho84 tagged with a FLAG epitope at its endogenous locus) in WT and tRNA thiolation mutant cells (uba4Δ and ncs2Δ) grown in high and low Pi media were detected and compared by Western blot analysis using an anti- FLAG antibody. For high and low Pi condition comparisons, the same amount of total protein was loaded in each lane of the SDS-PAGE gel. For low Pi, two different concentrations of total protein (undiluted and 1:3 diluted) were loaded, since the PHO-related proteins are strongly induced in this condition, and hence a dilution is required to visualize a non-saturated image. A representative blot is shown (n = 2). Also see Figure 5—figure supplement 1A. (D) Acid phosphatase activity is decreased in tRNA thiolation mutants. A schematic representation of the acid phosphatase reaction, and the colorimetric assay used to study this reaction, is shown. This quantitatively measures the collective intracellular activity of the PHO acid phosphatases Pho5, Pho11, Pho12 and Pho3. Acid phosphatase activity was determined in WT, tRNA thiolation mutant (uba4Δ), pho4Δ and pho80Δ in both high and low Pi media conditions, using this assay. Acid phosphatase activity was plotted relative to WT grown in high Pi, which was set to 1. * denotes statistical significance (Student’s t-test), comparing all samples to WT in both high and low Pi condition. Data are displayed as means ± SD, n = 2 biological replicates with three technical replicates each. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.
In order to more quantitatively estimate this dampening of the PHO response in the thiolation mutants, we utilized a robust assay to measure acid phosphatase activity. This represents the enzymatic activity of the PHO acid phosphatases, including Pho5, Pho11, Pho12 and the more constitutively expressed Pho3. Using this assay, we observed significantly reduced PHO activity in thiolation mutants (Figure 5D). Notably, this also quantitatively revealed that the thiolation mutants are unlike cells lacking Pho4 (PHO induction absent), or lacking Pho80 (constitutively extremely high PHO induction). These data reveal that the thiolation mutants are effectively in a constitutively phosphate-limited state, due to a dampened PHO response. These cells will therefore have a constitutively altered phosphate homeostasis, with reduced phosphate availability.
Phosphate depletion in wild-type cells phenocopies tRNA thiolation mutants
We therefore asked if phosphate starvation in WT cells itself resembles the metabolic hallmarks of the thiolation mutants. We first biochemically estimated amounts of trehalose in WT cells with or without phosphate starvation, and found a robust increase in trehalose upon phosphate starvation (Figure 6A), much like the thiolation mutants. We next measured Gcn4 (protein) in WT cells, with or without phosphate starvation. Here, we observed a strong induction in Gcn4 protein upon phosphate starvation (Figure 6B). Further, like the Gcn4 response in the thiolation mutants, the Gcn4 induction in phosphate-starved WT cells was Gcn2-dependent (Figure 6—figure supplement 1). In addition to these data shown here in WT cells with phosphate starvation, earlier studies in response to phosphate limitation, have also observed high amino acid and low nucleotide levels under these conditions (Boer et al., 2010; Klosinska et al., 2011). These data are strikingly consistent with the metabolic profile observed in thiolation mutants, and suggests that the induction of Gcn4 observed in response to phosphate limitation might be due to reduced nucleotide levels. Notably, a metabolic signature of phosphate starvation is a depletion in ATP levels (Boer et al., 2010). We have already demonstrated not just reduced ATP levels in the thiolation mutants, but carbon flux through nucleotide synthesis, including ATP synthesis was also reduced in thiolation mutants (shown earlier in Figure 2). Finally, prior studies with phosphate limitation have shown that ribosomal genes are slightly repressed with phosphate limitation (Saldanha et al., 2004). In the thiolation mutants, where phosphate homeostasis is affected due to the dampened PHO response, we also observed a small decrease in ribosomal genes (as shown earlier in Figure 4—figure supplement 3B; Figure 4—figure supplement 3C). Thus, these data from WT cells starved of phosphate strikingly phenocopied the tRNA thiolation mutants.
Figure 6 with 1 supplement see all
Phosphate depletion in wild-type cells phenocopies tRNA thiolation mutants.
(A) Trehalose amounts are increased upon phosphate starvation. Trehalose content of WT cells grown in high and no Pi media was plotted. Data are displayed as means ± SD, n = 3. ****p<0.0001. (B) Gcn4 protein is increased upon phosphate starvation. Gcn4 protein levels (Gcn4 tagged with HA epitope at the endogenous locus) in WT grown in high and no Pi media were detected by Western blot analysis using an anti-HA antibody. A representative blot is shown (n = 3). (C) Cells lacking Uba4 and Gcn4 (uba4Δ gcn4Δ) also show decreased acid phosphatase activity. Acid phosphatase activity was determined in WT, tRNA thiolation mutant (uba4Δ), gcn4Δ, uba4Δ gcn4Δ, pho4Δ and pho80Δ grown in high Pi media conditions supplemented with amino acids, by a colorimetric assay. Note that loss of Gcn4 alone has no effect on acid phosphatase activity. * denotes statistical significance (Student’s t-test), comparing all samples to WT. Data are displayed as means ± SD, n = 3 biological replicates (with three technical replicates each). ns denotes non-significant difference. *p<0.05, ***p<0.001.
Finally, we used the acid phosphatase activity assay (described earlier) as a robust read-out for the extent of the PHO response, to compare activities in WT cells, cells lacking tRNA thiolation, and cells lacking Gcn4 (individually or in combination). This was done to test if the Gcn4 induction was upstream or downstream of the PHO response. Notably, the loss of GCN4 in cells lacking thiolation (uba4Δ gcn4Δ) also resulted in significantly decreased acid phosphatase activity (Figure 6C), similar to the thiolation mutants. However, the loss of GCN4 alone had no effect on acid phosphatase activity. This suggests that the Gcn4 induction is downstream of the dampened PHO response in the thiolation mutants.
Collectively, these results strongly suggest that effective phosphate limitation is responsible for the metabolic state switch exhibited by the thiolation deficient cells.
Trehalose synthesis associated phosphate release enables cells to maintain phosphate balance
This observed downregulation of phosphate metabolism in the thiolation deficient cells is striking. Nonetheless, it is not immediately obvious biochemically how this relates to re-routing carbon towards storage carbohydrates, and decoupling amino acid metabolism from nucleotide synthesis. Perplexingly, in our transcript and translation analysis, no other metabolic arms were similarly decreased in the thiolation deficient cells, and only the amino acid biosynthesis arm (dependent on Gcn4) increases, which we have addressed earlier. Notably, while earlier studies have hinted that phosphate limitation results in a shift towards storage carbohydrates (Lillie and Pringle, 1980; Boer et al., 2003; Boer et al., 2010), this more extensive metabolic rewiring has not been carefully analyzed, and a biochemical explanation for this is missing. We wondered if some overlooked aspect within this biochemical process could explain why a perturbation in phosphate homeostasis connects to the synthesis of storage carbohydrates trehalose and glycogen, as is also seen in tRNA thiolation mutants. To address this, we carefully examined all the metabolic nodes altered in the tRNA thiolation mutants, evaluating necessary co-factors and products of each pathway, and looking for possible connections to phosphate. Here, we noted an apparently minor, largely ignored output in the arm of carbon metabolism, where glucose-6-phosphate is routed towards trehalose synthesis. The first step of trehalose synthesis is the formation trehalose-6-phosphate (T-6-P), carried out by trehalose-6-phosphate synthase (Tps1). This is followed by the dephosphorylation of T-6-P by Tps2 (De Virgilio et al., 1993), forming trehalose (Figure 7A). We noted that this Tps2-dependent second step is accompanied by the release of free, inorganic phosphate (Pi) (Figure 7A). Canonically, these two steps are viewed as an apparently futile trehalose cycle during glycolysis, regenerating glucose, in order to maintain balanced glycolytic flux (Heerden et al., 2014; van Heerden et al., 2014). However, we reasoned that if the availability of inorganic phosphate is limiting, a shift to trehalose synthesis can be a way by which cells can liberate Pi, and restore phosphate levels. For this to be generally true, the prediction is that during phosphate starvation, WT cells must accumulate trehalose in order to recover phosphate. As shown earlier, this is exactly what is observed in WT cells limited for phosphate (Figure 6A), and in the tRNA thiolation mutants (Figure 2E; Figure 2F) which are effectively phosphate limited due to a reduction in the PHO genes.
Figure 7 with 1 supplement see all
Trehalose synthesis associated phosphate release enables cells to maintain phosphate balance.
(A) A schematic representation of the trehalose cycle, showing the routing of glucose-6-phospate towards trehalose and glycogen biosynthesis. Trehalose synthesis requires two glucose molecules, catalysed by the Tps1 and Tps2 enzymes. The Tps2-mediated reaction synthesizes trehalose, and notably releases free Pi. (B) Intracellular Pi levels are maintained in tRNA thiolation mutant cells (uba4Δ) by Tps2 activity. Free intracellular Pi levels were determined in WT, tRNA thiolation mutant (uba4Δ), tps2Δ, uba4Δ tps2Δ and pho85Δ cells grown in minimal media, by a colorimetric assay. Intracellular Pi levels in mutant cells relative to WT are plotted, where WT was set to 100. Data are displayed as means ± SD, at least n = 2 biological replicates with three technical replicates each. ****p<0.0001, Student’s t-test, comparing all samples to WT. Also see Figure 7—figure supplement 1A. (C) Synthetic genetic interaction between TPS2 and tRNA thiolation genes (UBA4 and NCS2) at 37° C. WT, tRNA thiolation mutants (uba4Δ and ncs2Δ), tps2Δ, uba4Δ tps2Δ and ncs2Δ tps2Δ double mutant cells grown in minimal media were spotted on low Pi media agar plates with 2% glucose and incubated at 30° C and 37° C. Also see Figure 7—figure supplement 1B. (D) Synthetic genetic interaction between PHO80 and tRNA thiolation genes (UBA4 and NCS2). WT, tRNA thiolation mutants (uba4Δ and ncs2Δ), pho80Δ, uba4Δ pho80Δ and ncs2Δ pho80Δ double mutant cells grown in minimal media (0.1% glucose) were spotted on minimal media agar plates (0.1% and 2% glucose) and incubated at 30° C.
Given the central role of phosphate, cells utilize all means possible to restore internal phosphate (Ljungdahl and Daignan-Fornier, 2012). Therefore it is experimentally challenging to study changes in phosphate homeostasis in cells. However, we directly tested the hypothesis that trehalose synthesis is a direct way for cells to restore internal phosphate in tRNA thiolation mutants, by utilizing cells lacking TPS2. These cells cannot complete trehalose synthesis, and importantly cannot release phosphate (Figure 7A). We first measured the intracellular Pi levels in WT cells, thiolation mutants (uba4Δ), cells lacking Tps2p (tps2Δ), and cells lacking tRNA thiolation as well as Tps2p (uba4Δ tps2Δ). pho85Δ cells were used as a control, since they exhibit intrinsically higher intracellular Pi levels (Liu et al., 2017). We first observed that in cells lacking TPS2 (tps2Δ) intracellular Pi levels were lower (~75–80%) relative to WT cells (Figure 7B and Figure 7—figure supplement 1A). This suggests that while other pathways (phosphate uptake, glycerol production and vacuolar phosphate export) remain relevant, Tps2p-mediated Pi release by dephosphorylation of trehalose-6-P is itself important for maintaining internal phosphate levels. Importantly, uba4Δ cells had only slightly reduced intracellular Pi levels (~90%) relative to WT cells (Figure 7B and Figure 7—figure supplement 1A). This is consistent with the prediction that due to reduced PHO expression in these cells, phosphate homeostasis is altered, but they can compensate for phosphate availability through increased trehalose synthesis. Contrastingly, the cells lacking both thiolation and Tps2 (uba4Δ tps2Δ) showed a dramatic reduction in Pi levels (~65%), compared to either of their single mutants. This striking reduction in Pi levels in these cells is consistent with the predicted outcome, where an inability to release phosphate from trehalose (tps2Δ) is also coupled with reduced expression of phosphate assimilation genes (uba4Δ). Next, we tested possible genetic interactions between tps2Δ and thiolation mutants (uba4Δ and ncs2Δ) by assessing relative growth. In our genetic background, tps2Δ cells exhibit slightly slower growth at 37°C. Notably, in cells lacking both TPS2 and tRNA thiolation (tps2Δ uba4Δ or tps2Δ ncs2Δ), we observed a strong synthetic growth defect, in conditions of low phosphate as well as normal phosphate (Figure 7C and Figure 7—figure supplement 1B). This is entirely consistent with the proposed role of Tps2p in maintaining phosphate balance in thiolation mutants. Finally, if we completely imbalance phosphate homeostasis in cells, using cells lacking PHO80, individual mutants of either pho80Δ or thiolation deficient cells show minimal growth defects, but double mutants (pho80Δ uba4Δ or pho80Δ ncs2Δ) show a severe synthetic growth defect (Figure 7D).
Collectively, our results suggest that altering phosphate homeostasis by decreasing PHO activity regulates overall carbon and nitrogen flow. Cells can therefore deal with decreased phosphate availability by diverting carbon flux away from nucleotide biosynthesis, and towards Tps2-dependent trehalose synthesis and Pi release. This restores phosphate, while concurrently resulting in an accumulation of amino acids, and a reduction in nucleotide synthesis.
Discussion
In this study, we highlight two related findings- a direct role for a component of translational machinery, U34 thiolated tRNAs, in regulating cellular metabolism by controlling phosphate homeostasis; and a biochemical rationale for how phosphate availability regulates flux through carbon and nitrogen metabolism.
An integrative model emerges from our studies, explaining how high amounts of thiolated tRNAs reflect a ‘growth state’, while reduced tRNA thiolation reflect a ‘starvation state’ (Figure 8). Cells can use tRNA thiolation to sense overall nutrient sufficiency and appropriately modulate metabolic outputs, using phosphate homeostasis as the metabolic control point (Figure 8). In this model, tRNAs are thiolated in tune with methionine and cysteine availability, as has been demonstrated earlier (Laxman et al., 2013). Separately, the presence of these sulfur amino acids themselves reflect an overall amino acid sufficiency state, leading to an anabolic program capable of sustaining cell growth and proliferation (Walvekar et al., 2018b). Methionine up-regulates both amino acid synthesis and carbon flux leading towards nucleotide synthesis, and therefore growth (Walvekar et al., 2018b). Collectively therefore, in conditions of methionine sufficiency (and therefore amino acid sufficiency), where tRNAs are maximally thiolated, cells direct carbon flux towards nucleotide biosynthesis, coupled with amino acid utilization (as shown in Figures 1 and 2). Accordingly, at this level of metabolic coupling, thiolated tRNAs sense amino acids, and ensure appropriate nucleotide levels for growth and cell cycle progression (as shown in Figures 2 and 3). On the other hand, the loss of thiolated tRNAs results in an inability of cells to fully sense and integrate these nutrient cues, rewiring carbon and nitrogen flux away from nucleotide synthesis and instead towards storage carbohydrates. This switches cells to a ‘starvation-like state’. This metabolic rewiring mediated by tRNA thiolation is achieved not by directly regulating enzymes in these arms of carbon metabolism, but instead by down-regulating the PHO regulon. This constricts intracellular phosphate availability (as shown in Figures 4 and 5). A result of this dampened PHO response, and constriction in available free phosphate, is that in order to restore phosphate, cells divert glucose flux towards Tps2-mediated trehalose synthesis, which concurrently releases Pi (as shown in Figures 6 and 7). Thus, while the trehalose shunt and phosphate recycling restores phosphate levels, this is at the cost of decreased nucleotide biosynthesis, and delayed cell cycle progression. Effectively, the loss of tRNA thiolation rewires cells to a starved metabolic state. Collectively, tRNA thiolation appropriately regulates metabolic outputs by controlling phosphate homeostasis, thereby enabling cells to commit to growth (Figure 8). Intriguingly, this correlation of tRNA thiolation with growth and rewired metabolism is emerging in cancer development (McMahon and Ruggero, 2018; Rapino et al., 2018), suggesting possibly conserved metabolic roles for these modified tRNAs.
Figure 8
A model illustrating how tRNA thiolation regulates the metabolic state of the cell.
WT cells (which have a functional tRNA thiolation machinery) in amino acid (methionine and cysteine) replete conditions have high amounts of thiolated tRNAs. In these cells, carbon flux coupled with amino acid utilization is towards nucleotide biosynthesis. Sufficient nucleotide levels support growth, and proper cell cycle progression, and reflect an overall ‘growth’ metabolic state. The absence of the tRNA thiolation machinery results in non-thiolated tRNAs, and here carbon flux is driven away from nucleotide synthesis and towards storage carbohydrates, with a concurrent accumulation of amino acids. This metabolic rewiring in these mutants is due to reduced expression of genes involved in phosphate-responsive signalling pathway (a dampened PHO response), which results in a phosphate-limited state. Due to this restricted phosphate availability in the thiolation mutants, these cells attempt to restore intracellular phosphate levels through a metabolic rewiring, via Tps2-dependent trehalose synthesis, accompanied by phosphate release. Thus, despite the absence of carbon or nitrogen (amino acid) starvation, the tRNA thiolation mutants exhibit an overall metabolic signature of a ‘starved state’, with decreased nucleotide synthesis and delayed cell cycle progression. Key: Pho: phosphate transporters, Pi: inorganic phosphate, PPP: Pentose Phosphate Pathway, cys: cysteine, met: methionine.
Insight into a deeper coupling of discrete metabolic arms emerges from this study, suggesting how cells can fully integrate carbon, nitrogen, sulfur and phosphate inputs for optimal growth. Earlier studies have noted a strong correlation of phosphate starvation with decreased nucleotide synthesis, and ATP availability (Boer et al., 2010; Klosinska et al., 2011). Reduced ATP synthesis is a hallmark of phosphate starvation (Boer et al., 2010), along with increased trehalose synthesis (discussed in a subsequent paragraph). Through this metabolic rewiring, reduced flux through the PPP and one-carbon/folate cycle can be inferred. A reduction in these metabolic pathways will reduce not just nucleotide synthesis, but also the production of NADPH, and cellular reductive biosynthetic capacity (Fan et al., 2014; Hosios and Vander Heiden, 2018). Notably, the reductive costs in terms of NADPH utilized to assimilate sulfates into sulfur amino acids (methionine and cysteine), and their subsequent metabolites, are themselves extremely high (Thomas and Surdin-Kerjan, 1997; Kaleta et al., 2013; Walvekar et al., 2018b). Indeed, this coupling of NADPH production and methionine availability is also observed in the converse direction (as noted earlier in the text), where the presence of methionine increases PPP flux (Walvekar et al., 2018b), and mutants in the PPP pathway are methionine auxotrophs (Thomas et al., 1991). Therefore, an amplified regulatory response in relation to tRNA thiolation can be imagined. In such a response, in the presence of methionine, there is an increase in PPP and one-carbon flux and nucleotide synthesis, which is amplified by the maximally thiolated tRNAs by ensuring sufficient phosphate availability and an intact PHO response. Conversely, reduced sulfate assimilation in turn could limit thiolation, and thereby amplify the regulatory response, again using phosphate homeostasis as a control point. Indeed, the metabolic state of the tRNA thiolation mutants (and the phenotypes associated) are consistent with what is expected in a cell with reduced NADPH availability and reduced reductive biosynthetic capacity. Our data suggest a subtler, more integrative role for tRNA thiolation in optimally sensing overall nutrient sufficiency, and modulating overall metabolic responses leading to a growth-sustaining state.
More generally, our findings identify a biochemical reaction, the trehalose shunt, that explains how phosphate homeostasis determines the extent of carbon and nitrogen flux towards nucleotide synthesis. While it is textbook knowledge that inorganic phosphate is important for glucose homeostasis (Mason et al., 1981; Boyle, 2005; Heerden et al., 2014; van Heerden et al., 2014), the biochemical connection of phosphate balance to carbon and nitrogen flux remains poorly explained. Studies have observed that trehalose increases upon phosphate starvation (Lillie and Pringle, 1980; Klosinska et al., 2011), and TPS2 is upregulated (Ogawa et al., 2000). In these conditions, central carbon metabolism is down, and phosphate limitation is a ‘general starvation’ cue (Brauer et al., 2008; Boer et al., 2010; Gurvich et al., 2017). Why this occurs has not been immediately apparent. Here, identifying the trehalose shunt as a way to restore phosphate balance, explains these observations. These data also explain earlier observations from pathogenic fungi, showing that that trehalose synthesis determines flux through the pentose phosphate pathway, and nitrogen metabolism (Wilson et al., 2007). Additionally, phosphate starvation results in better cell survival in limited nutrient conditions, and also promotes efficient recovery when nutrients become available (Gurvich et al., 2017). Since trehalose accumulation and utilization are respectively tightly coupled with exit from and re-entry into the cell division cycle (Shi et al., 2010; Shi and Tu, 2013), we propose a dual role for trehalose synthesis during phosphate-limitation. During phosphate limitation, trehalose synthesis concurrently releases inorganic phosphate, which restores phosphate balance in the cell and diverts flux away from nucleotide synthesis and growth. When phosphate is no longer limiting, cells can liquidate trehalose to re-enter the cell division cycle, enabling rapid recovery. Thus, this study adds an important biochemical function to the many roles played by this versatile metabolite. These include cell survival during desiccation and freezing (Calahan et al., 2011; Erkut et al., 2011; Erkut et al., 2016; Tapia and Koshland, 2014; Tapia et al., 2015), the ability to act as a protein chaperone (Tapia and Koshland, 2014), and as a membrane protectant (Abusharkh et al., 2014).
A modified tRNA is an unusual but effective mechanism to coordinately integrate sensing of overall nutrient sufficiency, and regulate metabolic homeostasis. While previous studies have observed decreased phosphate-related transcripts in tRNA thiolation deficient cells (Leidel et al., 2009; Nedialkova and Leidel, 2015; Chou et al., 2017), possible roles for phosphate in tRNA thiolation mediated function have been ignored. Furthermore, tRNAs undergo other conserved modifications in the U34 position: 5-methoxycarbonylmethyluridine (mcm5 U34), 2-thiouridine (s2 U34), 5-methoxycarbonylmethyl-2-thiouridine (mcm5s2 U34), 5-methylaminomethyluridine (mnm5U34) (Phizicky and Hopper, 2010). In this study we focus only on how the s2 U34 modification (which is derived from sulfur amino acids) regulates cellular metabolic state. Interestingly, the other U34 modifications all require s-adenosyl methionine (SAM), and SAM is itself directly derived from sulfur amino acid metabolism (Thomas and Surdin-Kerjan, 1997. Mutants of all these related U34 tRNA modifications show similar metabolic phenotypes as the thiolation mutants (Zinshteyn and Gilbert, 2013; Nedialkova and Leidel, 2015; Chou et al., 2017; Han et al., 2018), and have a down-regulated PHO response (Chou et al., 2017). This raises the possibility that these U34-tRNA and other tRNA modifications derived from amino acid metabolism use similar mechanisms, of controlling phosphate availability in order to regulate metabolic homeostasis. A primordial role of such tRNA modifications might therefore be to appropriately sense overall amino acid sufficiency (with methionine as a sentinel growth signal; Walvekar et al., 2018b), and modulate metabolic states towards growth, regulating phosphate availability as a means to achieve this. Co-opting tRNAs (which are the translation components most closely linked to amino acids) to control metabolic states can therefore be an efficient means to ensure appropriate commitments to growth and proliferation, and maximize cellular fitness.
Concluding, here we discover that a sulfur amino acid-dependent tRNA modification (thiolated U34) enables cells to appropriately balance amino acid and nucleotide levels and regulate metabolic state, by controlling phosphate homeostasis. More generally, we suggest how phosphate homeostasis can impact flux through different arms of carbon and nitrogen metabolism.
Materials and methods
Key resources table
| Reagent type (species) or resource | Designation | Source or reference | Identifiers | Additional information |
|---|---|---|---|---|
| Antibody | anti-HA | Roche | Cat# 12CA5 | WB (1:2000) |
| Antibody | anti-phospho eIF2a (Ser51) | Cell Signalling Technology | Cat# 9721S | WB (1:1000) |
| Antibody | anti-FLAG | Sigma-Aldrich | Cat # F1804-5MG | WB (1:2000) |
| Antibody | anti-Rad53 yC-19 | Santa Cruz Biotechnology | Cat # sc-6749 | WB (>1:1000) |
| Antibody | HRP-conjugated secondary antibodies (anti-mouse, anti-rabbit) | Sigma-Aldrich | Cat # 7076S Cat # 7074S | WB (1:5000) |
| HPLC column | Synergi 4µ Fusion-RP 80A column (100 × 4.6 mm) | Phenomenex | Cat # 00D-4424-E0 | |
| Sequence-based reagent | Gcn4-luciferase translation reporters | this study | ||
| Peptide, recombinant protein | trehalase | Sigma-Aldrich | Cat # T8778 | |
| Peptide, recombinant protein | amyloglucosidase | Sigma-Aldrich | Cat # 10115 | |
| Peptide, recombinant protein | RNAse A | Sigma-Aldrich | Cat # R4875 | |
| Peptide, recombinant protein | protease solution | Sigma-Aldrich | Cat # P6887 | |
| Commercial assay or kit | Glucose estimation kit | Sigma-Aldrich | Cat # GAGO20 | |
| Commercial assay or kit | RiboZero | Epicenter | Cat # MRZH116 | |
| Commercial assay or kit | luciferase assay kit | Promega | Cat # E1500 | |
| Commercial assay or kit | Maxima SYBRGreen/ROX qPCR Master Mix | Thermo Scientific | Cat # K0222 | |
| Commercial assay or kit | ATP estimation kit | Thermo Scientific | Cat # A22066 | |
| Chemical compound, drug | SYTOX green | Thermo Scientific (Invitrogen) | Cat # S7020 | |
| Chemical compound, drug | p-nitrophenyl phosphate | Sigma-Aldrich | Cat # N4645 | |
| Chemical compound, drug | bicinchoninic acid assay | Thermo Scientific | Cat # 23225 | |
| Software, algorithm | Prism 7 | Graphpad |
Yeast strains, media and growth conditions
Request a detailed protocolThe prototrophic CEN.PK strain of Saccharomyces cerevisiae was used in all the experiments (van Dijken JP et al., 2000). All the strains used in this study are listed in Supplementary file 1. For all experiments, cells were grown overnight at 30°C in rich media (1% yeast extract, 2% peptone, 2% dextrose), washed once and subsequently sub-cultured in minimal media (0.67% yeast nitrogen base with ammonium sulfate, without amino acids, 0.1% glucose) unless specified. Phosphate-limited medium was prepared as described previously (Klosinska et al., 2011) except that 0.1%. glucose was used unless specified. The only source of phosphorus in phosphate-limited media (low Pi) was KH2PO4, which was present at a concentration of 0.15 mM. In high phosphate media (high Pi), KH2PO4 was present at a concentration of 7.5 mM with 0.1%. glucose unless specified. In no phosphate media (no Pi), KH2PO4 was completely absent and 0.1%. glucose was used. Complete medium was high Pi media supplemented with amino acids and 0.1% glucose. Amino acid concentrations were used as described previously (Sherman, 2002). Sulfur-rich medium was minimal media (0.67% yeast nitrogen base with ammonium sulfate, without amino acids) with 2% glucose. Sulfur-starved medium was prepared as described previously with 2% glucose (Kankipati et al., 2015) Sulfur amino acid limited medium was minimal media (0.67% yeast nitrogen base with ammonium sulfate, without amino acids) supplemented with all amino acids at a final concentration of 2 mM with methionine and cysteine being completely absent and 2% glucose.
Western blot analysis
Request a detailed protocolFor Gcn4, total eIF2α, P- eIF2α, Pho12, Pho84 and Rad53 protein levels, cells were grown overnight in rich media, washed once and subsequently sub-cultured in minimal media at an initial OD600 of 0.1 and grown till the OD600 reached 0.8–1.0. For Pho84 and Pho12 protein levels in high and low Pi media, cells were grown overnight in rich media, washed once and subsequently sub-cultured in high and low Pi media at an initial OD600 of 0.1 and incubated for 5 hr at 30°C. For Gcn4 protein levels in high and no Pi media, cells were grown overnight in rich media, washed once and subsequently sub-cultured in high and no Pi media at an initial OD600 of 0.1 and incubated for 8 hr at 30°C. Cells were harvested by centrifugation and protein was isolated by trichloroacetic acid (TCA) precipitation method. Briefly, cells were resuspended in 400 μl of 10% trichloroacetic acid and lysed by bead-beating three times. The precipitates were collected by centrifugation, resuspended in 400 μl of SDS-glycerol buffer (7.3% SDS, 29.1% glycerol and 83.3 mM Tris base) and heated at 100°C for 10 min. The lysate was cleared by centrifugation and protein concentration was determined by using a bicinchoninic acid assay (23225, Thermo Fisher). Equal amounts of samples were electrophoretically resolved on 4–12% pre-cast Bis-tris polyacrylamide gels (NP0322BOX, Invitrogen). Anti-HA (12CA5, Roche) was used to detect Gcn4-HA, anti-phospho eIF2α (Ser51) (9721S, Cell Signalling Technology) was used to detect phospho-eIF2α, anti-FLAG (F1804-5MG, Sigma-Aldrich) was used to detect total eIF2α (eIF2α-FLAG), Pho12-FLAG and Pho84-FLAG, anti-Rad53 yC-19 (sc-6749, Santa Cruz Biotechnology) was used to detect phosphorylated Rad53 protein. Horseradish peroxidase-conjugated secondary antibodies (mouse and rabbit) were obtained from Sigma-Aldrich. For Western blotting, standard enhanced chemiluminescence reagent (GE Healthcare) was used. Coomassie brilliant blue R-250 was used to stain gels for loading control.
Metabolite extraction and LC-MS/MS analysis
View detailed protocolFor steady state amino acids and nucleotides levels, cells were grown overnight in rich media, washed once and subsequently sub-cultured in minimal media at an initial OD600 of 0.1 and grown till the OD600 reached ~0.8.~10 OD600 cells were quenched with 60% methanol at −40°C, and metabolites were extracted, as explained in detail elsewhere (Walvekar et al., 2018a). For steady state amino acids, nucleotides and sulfur-containing metabolite levels in sulfur-rich and sulfurstarved conditions, wild-type cells were grown overnight in rich media, washed once and subsequently sub-cultured in sulfur-rich and sulfur-starved conditions at an initial OD600 of 0.25 and incubated for 3 hr at 30°C.~5 OD600 cells were quenched with 60% methanol at −40°C, and metabolites were extracted. For 15N-label incorporation in amino acids and nucleotides, cells were grown overnight in rich media, washed once and subsequently sub-cultured in minimal media (0.67% yeast nitrogen base without amino acids and ammonium sulfate, 0.1% glucose, 20 mM ammonium sulfate) at an initial OD600 of 0.1 and grown till the OD600 reached 0.5. At this point, 15N2-ammonium sulfate (299286, Sigma-Aldrich) was added to reach a ratio of 50% unlabeled to 50% fully labelled ammonium sulfate. Metabolites were extracted from ~6 OD600 cells. For 13C- label incorporation in nucleotides and other central carbon metabolites, cells were grown overnight in rich media, washed once and subsequently sub-cultured in minimal media at an initial OD600 of 0.1 and grown till the OD600 reached 0.5. For experiments where 13C- label incorporation into nucleotides was measured, in medium supplemented with or without additional methionine, cells were grown overnight in rich media, and subsequently sub-cultured in fresh rich media at an initial OD600 of 0.2 and grown till the OD600 reached 1, washed once and shifted to minimal media, 2% glucose with or without 2 mM methionine for 1 hr. After this time, [U-13C6] glucose (CLM-1396-PK, Cambridge Isotope Laboratories) was added to reach a ratio of 50% unlabeled to 50% fully labelled glucose. Metabolites were extracted from ~6 OD600 cells. Extensive metabolite extraction protocols are described (Walvekar et al., 2018a). Metabolites were analyzed using LC-MS/MS method as described in Walvekar et al. (2018a). Standards were used for developing multiple reaction monitoring (MRM) methods on Thermo Scientific TSQ Vantage Triple Stage Quadrupole Mass Spectrometer or Sciex QTRAP 6500. All the parent/product masses relevant to this study are listed in Supplementary file 3. Amino acids were detected in the positive polarity mode. For nucleotide measurements, nitrogen base release was monitored in the positive polarity mode. Trehalose was detected in the negative polarity mode. For PPP metabolites and other triose phosphates, phosphate release was monitored in the negative polarity mode.
Metabolites were separated using a Synergi 4µ Fusion-RP 80A column (100 × 4.6 mm, Phenomenex) on Agilent’s 1290 infinity series UHPLC system coupled to the mass spectrometer. For positive polarity mode, buffers used for separation were- buffer A: 99.9% H2O/0.1% formic acid and buffer B: 99.9% methanol/0.1% formic acid (Column temperature, 40°C; Flow rate, 0.4 ml/min; T = 0 min, 0% B; T = 3 min, 5% B; T = 10 min, 60% B; T = 11 min, 95% B; T = 14 min, 95% B; T = 15 min, 5% B; T = 16 min, 0% B; T = 21 min, stop). For ADP and ATP separation and detection, buffers used for separation were- buffer A: 5 mM ammonium acetate in H2O and buffer B: 5 mM ammonium acetate in 100% methanol, and metabolites were measured in positive polarity mode. Alternately, buffers used for separation were- buffer A: 5 mM ammonium acetate in H2O and buffer B: 100% acetonitrile (Column temperature, 25°C; Flow rate: 0.4 ml/min; T = 0 min, 0% B; T = 3 min, 5% B; T = 10 min, 60% B; T = 11 min, 95% B; T = 14 min, 95% B; T = 15 min, 5% B; T = 16 min, 0% B; T = 21 min, stop), and negative polarity mode was used. The area under each peak was calculated using Thermo Xcalibur software (Qual and Quan browsers) and AB SCIEX MultiQuant software 3.0.1.
Spotting assay for comparative cell growth estimation
Request a detailed protocolFor all spotting assays, cells were grown overnight in rich media, washed once and subsequently sub-cultured in minimal media at an initial OD600 of 0.2–0.25 and grown till the OD600 reached 0.8–1.0. Cells were harvested by centrifugation, washed once with water and 10 μl sample for each suspension was spotted in serial 10-fold dilutions. For 8-aza adenine and hydroxyurea sensitivity assays, cells were spotted onto minimal media plates containing 250 and 300 μg/ml 8-aza adenine (A0552, TCI chemicals) or 150 mM hydroxyurea (H8627, Sigma-Aldrich) and incubated at 30°C. For control plates without drug, images were taken after 1–2 days and for drug containing plates after 4–5 days. For genetic interaction analysis with Tps2, cells were spotted onto high and low Pi media plates with 2% glucose. Plates were incubated at 30°C and 37°C. For genetic interaction analysis with Pho80, cells were spotted onto minimal media plates with 0.1% and 2% glucose. Plates were incubated at 30°C.
Trehalose and glycogen measurements
View detailed protocolFor trehalose and glycogen measurements in wild-type and thiolation mutants, cells were grown overnight in rich media, washed once and subsequently sub-cultured in minimal media at an initial OD600 of 0.1 and grown till the OD600 reached 0.8–1.0. For trehalose measurement in high and no Pi media, cells were grown overnight in rich media, washed once and subsequently sub-cultured in either high or no Pi media at an initial OD600 of 0.1 and incubated for 8 hr at 30°C. For trehalose measurement in sulfur-rich and sulfur-starved conditions, wild-type cells were grown overnight in rich media, washed once and subsequently sub-cultured in sulfur-rich and sulfur-starved conditions at an initial OD600 of 0.25 and incubated for 5 hr at 30°C. Cells were harvested by centrifugation and washed with ice-cold water. Cells were lysed in 0.25 M sodium carbonate by incubating at 95–98°C for 4 hr. Subsequently, added 0.15 ml 1M acetic acid and 0.6 ml of 0.2 M sodium acetate to bring the solution to pH 5.2. Trehalose and glycogen were digested overnight using trehalase (T8778, Sigma-Aldrich) and amyloglucosidase (10115, Sigma-Aldrich) respectively. Glucose released from these digestions was measured using a Glucose (GO) Assay Kit (GAGO20, Sigma-Aldrich). The concentration of released glucose (μg/ml) was determined from the standard curve and plotted. Statistical significance was determined using Student T-test (GraphPad Prism 7).
Continuous chemostat culture growth to study yeast metabolic cycles, and microscopic analysis
Request a detailed protocolContinuous chemostat cultures to establish the YMC were performed as described previously (Tu et al., 2005). An overnight batch culture of prototrophic CEN.PK strain (van Dijken JP et al., 2000) grown in rich medium was used to inoculate working volume of 1L in the chemostat. At 20 min time-intervals, cells were fixed with 2% paraformaldehyde, and imaged under a bright-field microscope. ~200 cells from each time point were sampled, and budding cells were counted manually.
Cell cycle synchronization and flow cytometry analysis bar1Δ::Hyg, uba4Δ::NAT bar1Δ::Hyg and ncs2Δ::NAT bar1Δ::Hyg cells were grown overnight in minimal media and subsequently sub-cultured in minimal media at an initial OD600 of 0.05 and grown till the OD600 reached 0.2. Cells were harvested by centrifugation, washed with water and resuspended in the same medium containing 10 μg/ml of α-factor (GenScript). Cells were kept at 30°C for 3 hr till complete G1 arrest was observed by light microscopy. Subsequently, 5 ml culture was harvested by centrifugation, washed with water and fixed with 70% ethanol for G1-arrested population. Remaining culture was synchronously released into the cell cycle by washing away the α-factor. Cells were collected at different intervals of time post G1 release, fixed with ethanol, treated with RNaseA (R4875, Sigma-Aldrich) and a protease solution (P6887, Sigma-Aldrich) as described (Haase and Reed, Cell cycle, 2002). Cells were stained with SYTOX green (S7020, Invitrogen) and analyzed on BD FACS Verse flow cytometer.
Time-lapse live cell microscopy
Request a detailed protocolCells were grown overnight in rich media, washed once and subsequently sub-cultured in minimal media at an initial OD600 of 0.1 and grown till the OD600 reached 0.4–0.5. 1.5% agar pads (50081, Lonza) were prepared containing minimal media. The pad was cut into small pieces after it solidified. 2 μl of the cell suspension was placed on the agar pad, which was inverted and placed in a glass bottom confocal dish (101350, SPL Life Sciences) for imaging. Phase-contrast images were captured after every 3 min’ interval for total of 360 mins on ECLIPSE Ti2 inverted microscope (NIKON) and 60X oil-immersion objective. Images were stacked and analyzed using ImageJ software. Statistical significance was determined using a Student T-test (GraphPad Prism 7).
Transcriptome and ribosome profiling analyses
Request a detailed protocolCells were grown overnight in rich media and subsequently shifted to minimal media till the OD600 reached 0.5–0.8. Cells were rapidly harvested by filtration and lysed, as described in detail (Weinberg et al., 2016; McGlincy and Ingolia, 2017). For both transcriptome and ribosome profiling analyses, three biological replicates each for WT and tRNA thiolation mutant cells (uba4Δ and ncs2Δ) were included. Total RNA and ribosome-protected fragments were isolated from the cell lysates and RNA-seq and ribosome profiling were performed, as described (Weinberg et al., 2016; McGlincy and Ingolia, 2017), with minor modifications. Separate 5’ and 3’ linkers were ligated to the RNA- fragment instead of 3’ linker followed by circularization (Subtelny et al., 2014). 5’ linkers contained four random nt unique molecular identifier (UMI) similar to a five nt UMI in 3’ linkers. During size-selection, we restricted the footprint lengths to 18–34 nts. Matched RNA-seq libraries were prepared using RNA that was randomly fragmentation by incubating for 14 min at 950C with in 1 mM EDTA, 6 mM Na2CO3, 44 mM NaHCO3, pH 9.3. RNA-seq fragments were restricted to 18–50 nts. Ribosomal rRNA were removed from pooled RNA-seq and footprinting samples using RiboZero (Epicentre MRZH116). cDNA for the pooled libraries were PCR amplified for 16 cycles.
Ribosome profiling data processing and analysis
Request a detailed protocolRNA-seq and footprinting reads were mapped to the yeast transcriptome using the riboviz pipeline (Carja et al., 2017). Sequencing adapters were trimmed from reads using Cutadapt 1.14 (Martin, 2011) (--trim-n -e 0.2 –minimum-length 24). The reads from different samples were separated based on the barcodes in their 3’ linkers using fastx_barcode_splitter (FASTX toolkit, Hannon lab) with utmost one mismatch allowed. UMI and barcodes were removed from reads in each sample using Cutadapt (--trim-n -m 10 u 4 u −10). Trimmed reads that aligned to yeast rRNAs and tRNAs were removed using HISAT2 v2.1.0 (Kim et al., 2015). Remaining reads were mapped to a set of 5812 genes in the yeast genome (SGD version R64-2-1_20150113) using HISAT2. Only reads that mapped uniquely were used for all downstream analyses. Codes for generating processed fastq and gff files were obtained from riboviz package (https://github.com/shahpr/RiboViz; Carja et al., 2017). Gene-specific fold-changes in RNA and footprint abundances were estimated using DESeq2 packages in R (Love et al., 2014) using default log-fold-change shrinkage options. Changes in ribosome-densities (translation efficiencies) were estimated using the Riborex package in R (Li et al., 2017).
The complete transcript/ribosome footprint datasets are available at GEO (number GSE124428; link: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE124428).
Acid phosphatase assay for Pho5 and related enzyme activity
Request a detailed protocolCells were grown overnight in rich media, washed once and subsequently sub-cultured in high and low Pi media (for experiment related to wild-type and uba4Δ) or complete media (for experiment related to wild-type, uba4Δ, gcn4Δ and uba4Δ gcn4Δ) at an initial OD600 of 0.1 and grown till the OD600 reached 0.5–0.6. 8 OD600 cells were collected, washed once and resuspended in sterile water to final OD600 of 16. Acid phosphatase activity was assayed, as described previously with some modifications (Huang and O'Shea, 2005). Briefly, 450 μl of each cell suspension was added to 200 μl of 20 mM p-nitrophenyl phosphate (PNPP, N4645, Sigma‐Aldrich) in 0.1M sodium acetate, pH-4.2, mixed and incubated at room temperature for 30 min. To stop the reaction, 200 μl of the reaction mixture was withdrawn and added to 200 μl of ice cold 10% trichloroacetic acid. To this reaction, 400 μl of saturated sodium carbonate solution (2M, pH-11.5) was added, mixed and centrifuged at 3000 rpm for 10 min. 80 μl of the supernatant was transferred in technical triplicates to a 96-well plate and liberated p-nitrophenol was determined by measuring OD420 on a plate reader. Phosphatase activity was measured in units expressed as OD420/OD600 × 1000. Statistical significance was determined using a Student T‐test (GraphPad Prism 7).
Phosphate measurement
Request a detailed protocolCells were grown overnight in rich media, washed once and subsequently sub-cultured in minimal media at an initial OD600 of 0.1 and grown till the OD600 reached 0.8–1.0. Free intracellular phosphate levels were determined, as described previously (McNaughton et al., 2010). Briefly, cells were harvested by centrifugation and washed twice with ice-cold water. Cells were lysed by resuspending in 200 μl 0.1% triton X-100 and vortexed for 5 min with glass beads. The lysate was cleared by centrifugation and protein concentration was determined by using bicinchoninic acid assay (23225, Thermo Fisher). 30 μg of whole cell lysate was used for measurement of free intracellular phosphate levels using ammonium molybdate and ascorbic acid colorimetric assay as described (Ames, 1966). Potassium dihydrogen phosphate solution was used for standard curve (0 to 500 µM KH2PO4). The amount of phosphate was expressed as µM Pi. Statistical significance was determined using a Student T-test (GraphPad Prism 7).
Luciferase assay for GCN4 translation
Request a detailed protocolWT, uORF1* and uORF4* Gcn4-luciferase reporter plasmids (SL148, SL149 and SL150) were generated by PCR amplification of a 777 bp fragment of WT, uORF1*and uORF4* Gcn4 constructs (GeneArt, Thermo Scientific) and subsequent cloning in SL147 plasmid (Supplementary file 2). SL147 was generated by cloning luciferase cDNA from pGL3-Basic (Addgene) in pSL80 plasmid (Wu and Tu, 2011). For luciferase assay in wild-type, thiolation mutants, gcn2Δ and double mutants, cells transformed with Gcn4-luciferase reporter plasmids were grown overnight in rich media in presence of G418. Cells were subsequently sub-cultured in the same media without selection till logarithmic phase and then shifted to minimal media. After 4–5 hr of growth (OD600 of 0.8–1.0), cells were collected, washed twice with ice-cold lysis buffer (1X PBS, 1 mM PMSF). For luciferase assay in wild-type and gcn2Δ, in high and no Pi media, cells transformed with Gcn4-luciferase reporter plasmid were grown overnight in rich media in presence of G418. Cells were subsequently sub-cultured in high and no Pi media supplemented with amino acids, 2% glucose and incubated for 8 hr at 30°C, cells were collected, washed twice with ice-cold lysis buffer (1X PBS, 1 mM PMSF). For luciferase assay in sulfur amino acid limited condition, wild-type cells transformed with Gcn4-luciferase reporter plasmid were grown overnight in rich media, and subsequently sub-cultured in rich media at an initial OD600 of 0.2 and grown till the OD600 reached 0.4–0.5 and half of the cells were collected for luciferase assay. Remaining cells were washed once and subsequently shifted to sulfur amino acid limited condition and incubated for 1 hr at 30°C. Lysis was done by vortexing cells for 5 min in presence of glass beads. Lysates were cleared by centrifugation and protein concentration was determined by using a bicinchoninic acid assay (23225, Thermo Fisher). Luciferase assay was performed using luciferase assay system kit (E1500, Promega) and activity was measured using a Sirius luminometer (Tiertek Berthold). Data output provided as Relative Light Units per sec (RLU/s) was used to determine relative luciferase activity. Statistical significance was determined using a Student T-test (GraphPad Prism 7).
RNA isolation and quantitative real-time PCR (qRT-PCR) analysis
Request a detailed protocolCells were grown overnight in rich media, washed once and subsequently sub-cultured in minimal media at an initial OD600 of 0.1 and grown till the OD600 reached 0.8–1.0. Cells were harvested by centrifugation, and RNA was isolated by hot phenol beating method (Collart and Oliviero, 2001. 6 μg of total RNA was used for DNase I (AM2238, Thermo Fisher) treatment. cDNA was synthesized with random primers (48190011, Thermo Fisher) and SuperScript II reverse transcriptase (18064014, Thermo Fisher Scientific) according to the manufacturer’s protocol. cDNA quantification was done by real-time PCR on an ViiA 7 Real-Time PCR System (Thermo Fisher) using Maxima SYBR Green/ROX qPCR Master Mix (K0222, Thermo Fisher). ACT1 was used as an internal normalization control. All qRT-PCRs were performed in triplicates using at least two independent biological RNA samples. Statistical significance was determined using a Student T-test (GraphPad Prism 7).
ATP measurement
Request a detailed protocolTotal cellular ATP was measured using the ATP determination kit (A22066, Thermo Fisher) according to the manufacturer’s protocol. Briefly, 5 µl sample or 5 µl of different concentrations of ATP standard solution (0 to 5 µM ATP) was added to 50 µl of assay solution. Luminescence was measured directly after addition of the sample to assay solution using a Sirius luminometer (Tiertek Berthold). ATP concentrations in samples were calculated from the ATP standard curve and relative levels were plotted. Statistical significance was determined using a Student T-test (GraphPad Prism 7).
Data availability
Sequencing (transcript and ribosome footprint) data have been deposited in GEO under accession codes GSE124428, and are fully open.
-
NCBI Gene Expression OmnibusID GSE124428. A tRNA modification balances carbon and nitrogen metabolism by regulating phosphate homeostasis.
References
-
[10] Assay of inorganic phosphate, total phosphate and phosphatasesMethods in Enzymology 8:115–118.https://doi.org/10.1016/0076-6879(66)08014-5
-
Transcriptional regulation of phosphate-responsive genes in low-affinity phosphate-transporter-defective mutants in Saccharomyces cerevisiaeBiochemical and Biophysical Research Communications 306:843–850.https://doi.org/10.1016/S0006-291X(03)01068-4
-
Transfer RNA modificationAnnual Review of Biochemistry 56:263–285.https://doi.org/10.1146/annurev.bi.56.070187.001403
-
Growth-limiting intracellular metabolites in yeast growing under diverse nutrient limitationsMolecular Biology of the Cell 21:198–211.https://doi.org/10.1091/mbc.e09-07-0597
-
Biochemistry and Molecular Biology EducationLehninger principles of biochemistry, Biochemistry and Molecular Biology Education, Fourth Edition, W.H. Freeman.
-
Coordination of growth rate, cell cycle, stress response, and metabolic activity in yeastMolecular Biology of the Cell 19:352–367.https://doi.org/10.1091/mbc.e07-08-0779
-
Driving the cell cycle through metabolismAnnual Review of Cell and Developmental Biology 28:59–87.https://doi.org/10.1146/annurev-cellbio-092910-154010
-
Preparation of yeast RNACurrent Protocols in Molecular Biology Chapter 13:Unit13.12.https://doi.org/10.1002/0471142727.mb1312s23
-
Disruption of TPS2, the gene encoding the 100-kDa subunit of the trehalose-6-phosphate synthase/phosphatase complex in Saccharomyces cerevisiae, causes accumulation of trehalose-6-phosphate and loss of trehalose-6-phosphate phosphatase activityEuropean Journal of Biochemistry 212:315–323.https://doi.org/10.1111/j.1432-1033.1993.tb17664.x
-
Urmylation: a ubiquitin-like pathway that functions during invasive growth and budding in yeastMolecular Biology of the Cell 14:4329–4341.https://doi.org/10.1091/mbc.e03-02-0079
-
Nutrient sensing and TOR signaling in yeast and mammalsThe EMBO Journal 36:397–408.https://doi.org/10.15252/embj.201696010
-
tRNA's modifications bring order to gene expressionCurrent Opinion in Microbiology 11:134–140.https://doi.org/10.1016/j.mib.2008.02.003
-
Translational regulation of GCN4 and the general amino acid control of yeastAnnual Review of Microbiology 59:407–450.https://doi.org/10.1146/annurev.micro.59.031805.133833
-
The redox requirements of proliferating mammalian cellsJournal of Biological Chemistry 293:7490–7498.https://doi.org/10.1074/jbc.TM117.000239
-
Metabolic costs of amino acid and protein production in Escherichia coliBiotechnology Journal 8:1105–1114.https://doi.org/10.1002/biot.201200267
-
Sul1 and Sul2 sulfate transceptors signal to protein kinase A upon exit of sulfur starvationJournal of Biological Chemistry 290:10430–10446.https://doi.org/10.1074/jbc.M114.629022
-
HISAT: a fast spliced aligner with low memory requirementsNature Methods 12:357–360.https://doi.org/10.1038/nmeth.3317
-
tRNA anticodon loop modifications ensure protein homeostasis and cell morphogenesis in yeastNucleic Acids Research 44:10946–10959.https://doi.org/10.1093/nar/gkw705
-
Yeast cells can access distinct quiescent statesGenes & Development 25:336–349.https://doi.org/10.1101/gad.2011311
-
Regulation of repressible acid phosphatase gene transcription in Saccharomyces cerevisiaeMolecular and Cellular Biology 5:2131–2141.https://doi.org/10.1128/MCB.5.8.2131
-
Signaling phosphate starvationTrends in Biochemical Sciences 21:383–387.https://doi.org/10.1016/0968-0004(96)10048-7
-
Reserve carbohydrate metabolism in Saccharomyces cerevisiae: responses to nutrient limitationJournal of Bacteriology 143:1384–1394.
-
Structure and expression of the PHO80 gene of Saccharomyces cerevisiaeNucleic Acids Research 16:2625–2637.https://doi.org/10.1093/nar/16.6.2625
-
The importance of inorganic phosphate in regulation of energy metabolism of Streptococcus lactisThe Journal of Biological Chemistry 256:1861–1866.
-
Thio-modification of yeast cytosolic tRNA requires a Ubiquitin-related system that resembles bacterial sulfur transfer systemsJournal of Biological Chemistry 283:27469–27476.https://doi.org/10.1074/jbc.M804043200
-
tRNA biology charges to the frontGenes & Development 24:1832–1860.https://doi.org/10.1101/gad.1956510
-
Nutritional homeostasis in batch and steady-state culture of yeastMolecular Biology of the Cell 15:4089–4104.https://doi.org/10.1091/mbc.e04-04-0306
-
Getting started with yeastMethods in Enzymology 350:3–41.https://doi.org/10.1016/s0076-6879(02)50954-x
-
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
-
Coupling among growth rate response, metabolic cycle, and cell division cycle in yeastMolecular Biology of the Cell 22:1997–2009.https://doi.org/10.1091/mbc.e11-02-0132
-
Metabolism of sulfur amino acids in Saccharomyces cerevisiaeMicrobiology and Molecular Biology Reviews : MMBR 61:503–532.
-
An interlaboratory comparison of physiological and genetic properties of four Saccharomyces cerevisiae strainsEnzyme and Microbial Technology 26:706–714.https://doi.org/10.1016/S0141-0229(00)00162-9
-
Methionine coordinates a hierarchically organized anabolic program enabling proliferationMolecular Biology of the Cell 29:3183–3200.https://doi.org/10.1091/mbc.E18-08-0515
-
Selective regulation of autophagy by the Iml1-Npr2-Npr3 complex in the absence of nitrogen starvationMolecular Biology of the Cell 22:4124–4133.https://doi.org/10.1091/mbc.e11-06-0525
-
Phosphate transport and sensing in Saccharomyces cerevisiaeGenetics 159:1491–1499.
-
Accurate translation of the genetic code depends on tRNA modified nucleosidesJournal of Biological Chemistry 277:16391–16395.https://doi.org/10.1074/jbc.M200253200
-
Mode of expression of the positive regulatory genes PHO2 and PHO4 of the phosphatase regulon in Saccharomyces cerevisiaeMGG Molecular & General Genetics 217:31–39.https://doi.org/10.1007/BF00330939
-
How Saccharomyces responds to nutrientsAnnual Review of Genetics 42:27–81.https://doi.org/10.1146/annurev.genet.41.110306.130206
Decision letter
-
Alan G HinnebuschReviewing Editor; Eunice Kennedy Shriver National Institute of Child Health and Human Development, United States
-
Naama BarkaiSenior Editor; Weizmann Institute of Science, Israel
-
Bertrand Daignan-FornierReviewer; Université de Bordeaux IBGC UMR 5095, France
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.
Thank you for submitting your article "A tRNA modification balances carbon and nitrogen metabolism by regulating phosphate homeostasis" for consideration by eLife. Your article has been reviewed by two peer reviewers, including Alan Hinnebusch as the Reviewing Editor, and the evaluation has been overseen by Naama Barkai as the Senior Editor. The following individual involved in review of your submission has also agreed to reveal his identity: Bertrand Daignon-Fornier (Reviewer #2).
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
Summary:
This paper presents evidence that in yeast mutants incapable of thiolating tRNAs the expression of many genes of the PHO phosphate regulon is reduced, along with intracellular phosphate (Pi) levels. This apparently accounts for the decreased nucleotide synthesis and nucleotide abundance observed in these mutants, with attendant induction of Gcn2-dependent Gcn4 translation and amino acid biosynthesis (shown previously to be induced by purine limitation). The mutants also exhibit elevated trehalose, whose synthesis by Tps2 is shown to mitigate the reduction in intracellular Pi levels and to support cell viability in the thiolation mutants. Furthermore, Pi starvation was shown to induce trehalose and Gcn4 in wild-type cells (although nucleotide synthesis/levels were not measured and probably should have been). They propose a model that tRNA thiolation is an intracellular signal for methionine/cysteine availability and that when thiolation drops it reduces nucleotide synthesis and cell division as an adaptive response by down-regulating PHO gene expression and intracellular Pi levels. This in turn diverts glucose utilization from glycolysis to trehalose synthesis, liberating Pi from glucose-6-Pi to ameliorate the Pi starvation and slow down cell growth and metabolism. The study does not provide a molecular basis for the down-regulation of PHO gene transcripts in the thiolation mutants; nor was it shown that starvation for Met or Cys would elicit the key phenotypic responses observed in the thiolation mutants, which would seem to be essential evidence for their model if it is not already demonstrated in the literature. The study also provides an apparently overlooked explanation for why trehalose increases on phosphate starvation-to release Pi from Glu-6-Pi-in addition to diverting glucose to a storage carbohydrate like glycogen; although they seem to overlook the fact that increased trehalose levels is a common response to many different kinds sorts of cellular stress.
Essential revisions:
1) Perform new experiments to show that limitation for the sulfur-containing amino acids and/or inorganic sulfate, elicits the same key responses seen for the thiolation mutants, to support the claim that tRNA thiolation is used as a sensor for sulfur levels.
2) Cite the appropriate literature from the Rabinowitz and Broach labs (Boer et al., 2010 and Klosinslka et al., 2011) showing that nucleotide levels are reduced on Pi starvation, to support the claim that the reduced nucleotide levels in the thiolation mutants can be attributed to the reduced Pi levels in these strains.
3) Measure the ATP/GTP levels (or at least AXP/GXP) in the thiolation mutants rather than relying on measurements of the much less abundant AMP/GMP.
4) Expand the analysis of the RNA-Seq data of Figure 5 to show and discuss all of the transcriptional changes for genes in the PPP and trehalose and glycogen pathways.
5) Use the proper time frame for the experiment in Figure 2 in order to measure the flux in the PPP and glycolysis pathways in the thiolation mutants, to confirm the rerouting hypothesis.
6) Discuss the possibility that lower flux in the PPP might result in low NADPH and attendant diminished sulfate assimilation under conditions where sulfur is already limiting. Reduced sulfate assimilation could in turn limit thiolation and amplify the whole regulatory response.
Reviewer #1:
This paper presents evidence that in yeast mutants incapable of thiolating tRNAs the expression of many genes of the PHO phosphate regulon is reduced, along with intracellular phosphate (Pi) levels. This apparently accounts for the decreased nucleotide synthesis and nucleotide abundance observed in these mutants, with attendant induction of Gcn2-dependent Gcn4 translation and amino acid biosynthesis (shown previously to be induced by purine limitation). The mutants also exhibit elevated trehalose, whose synthesis by Tps2 is shown to mitigate the reduction in intracellular Pi levels and to support cell viability in the thiolation mutants. Furthermore, Pi starvation was shown to induce trehalose and Gcn4 in wild-type cells (although nucleotide synthesis/levels were not measured and probably should have been). They propose a model that tRNA thiolation is an intracellular signal for methionine/cysteine availability and that when thiolation drops it reduces nucleotide synthesis and cell division as an adaptive response by down-regulating PHO gene expression and intracellular Pi levels. This in turn diverts glucose utilization from glycolysis to trehalose synthesis, liberating Pi from glucose-6-Pi to ameliorate the Pi starvation and slow down cell growth and metabolism. The study does not provide a molecular basis for the down-regulation of PHO gene transcripts in the thiolation mutants; nor was it shown that starvation for Met or Cys would elicit the key phenotypic responses observed in the thiolation mutants, which would seem to be essential evidence for their model if it is not already demonstrated in the literature. The study also provides an apparently overlooked explanation for why trehalose increases on phosphate starvation-to release Pi from Glu-6-Pi-in addition to diverting glucose to a storage carbohydrate like glycogen; although they seem to overlook the fact that increased trehalose levels is a common response to many different kinds sorts of cellular stress.
- It seems crucial to determine whether starvation for Met or Cys elicits decreased expression of the PHO regulon, decreased Pi levels (in the uba4 tps2 double mutant where it can most easily be measured), and increased trehalose levels in order to provide key support for their model that tRNA thiolyation is the intracellular sensor for Met/Cys levels and used to re-wire phosphate, nucleotide, and trehalose metabolism in the cell.
- Figure 1—figure supplement 1A: It's also unexpected that the mutations would increase expression from the uORF1* and uORF4* mutant reporters. They should examine the reporter lacking both uORF1 and uORF4 to determine whether the increased expression in the mutants is eliminated by removing both uORFs.
- Can the S phase delay, as well as the induction of Gcn4, in the thiolation mutants be rescued by supplementing cultures with nucleoside precursors or ribose-5-P to confirm that these phenotypes result from low nucleotide levels?
- Subsection “Phosphate depletion in wild-type cells phenocopies tRNA thiolation mutants” and Figure 5D: It should be determined whether the increase in Gcn4 protein level elicted by Pi starvation is abolished in a gcn2 mutant, as would be expected if phosphate starvation is mimicking the thiolation mutants.
- It seems important to determine whether nucleotide levels are reduced during Pi starvation, unless shown in the literature already, to support the claim that the reduced nucleotide levels in the thiolation mutants result from their reduced Pi levels.
- Subsection “Trehalose synthesis associated phosphate release enables cells to maintain phosphate balance”, second paragraph and Figure 6D: PHO gene expression is reduced in the thiolyation mutants. Why then should constitutively activating PHO genes under the control of Po2/Pho4 in a pho80 mutant be synthetic-sick with the thiolyation mutants? This needs to be explained more precisely than just a loss of phosphate homeostasis, which is too vague.
- Given that increased trehalose synthesis is a common response to different kinds of stress e.g. trehalose is an osmoprotectant – isn't it dangerous to ascribe to increased trehalose synthesis a specific role in mitigating the effects of reduced Pi levels in the thiolation mutants, and by extension, conditions of Met/Cys limitation? Wouldn't it make more sense to emphasize the role of increased trehalose synthesis in depleting Glucose-6-P levels, with attendant reduction in nucleotide synthesis and cell division when cells are limited for Met/Cys?
Reviewer #2:
The manuscript entitled « A tRNA modification balances carbon and nitrogen metabolism by regulating phosphate homeostasis. » by Gupta and co-workers reveals a very interesting crosstalk between metabolic pathways in yeast. The authors take advantage of mutants lacking a specific tRNA-modification to reveal and decipher complex interactions between metabolic pathways. This work hence nicely illustrates how cells integrate signals resulting from various nutrient inputs. The general scheme emerging from the work is that lack of tRNA-thiolation somehow affects the regulation of phosphate metabolism and results in lower intracellular inorganic-phosphate. As a consequence, glucose-6P would be rerouted from the pentose phosphate pathway (PPP) to trehalose synthesis which releases inorganic phosphate. Lower flux in the PPP would limit nucleotide synthesis and thereby activate a Gcn4-dependent response.
Each of the metabolic and expression steps is well documented, however most of the mechanisms connecting the regulation processes are not elucidated nor discussed. Addressing (when possible) and/or discussing the following questions would improve and strengthen the manuscript:
1) What is the physiological signal mimicked by thiolation defective mutants? Since tRNA-thiolation reflects intracellular availability of sulfur containing amino-acids (Introduction, fourth paragraph) and since sulfur is mostly used for synthesis of methionine, cysteine and S-adenosylmethionine (Thomas and Kerjan, 1997, p. 503-532), does thiolation mirror sulfur availability? Would sulfur limitation recapitulate the thiolation-defect? this would strongly increase the overall interest of the paper.
2) What is the connection between tRNA-thiolation and PHO-genes down-regulation? Simple experiments would allow to genetically position the tRNA-thiolation mutants within the phosphate regulation scheme. For example, do the ncs2/uba4 mutants affect Pho4 nuclear localization? Systematic additivity and epistasis studies with PHO regulation-mutants could be carried out. Unexpectedly the double ncs2/uba4 mutants with pho80 (subsection “Trehalose synthesis associated phosphate release enables cells to maintain phosphate balance”, second paragraph) reveal a synthetic growth phenotype while these mutants should antagonize each other. As such this experiment does not support the model.
3) It is not clear how the rerouting from the PPP to trehalose synthesis takes place. The authors should show (Figure 4) and discuss the transcriptional response for all the genes of the PPP as well as the trehalose and glycogen pathways. Is transcriptional regulation responsible for the rerouting? Importantly the chosen time frame (Figure 2 and subsection “Carbon flux is routed towards storage carbohydrates in thiolation mutants”, first paragraph) did not allow to directly measure the flux in the PPP and glycolysis and therefore the rerouting hypothesis relies only on trehalose, AMP and GMP synthesis.
4) The proposed PPP to trehalose rerouting allows to recycle inorganic phosphate when carbon is plentiful and phosphate is scarce, however the lower flux in the PPP might result in low NADPH which is crucial for sulfate assimilation (the zwf1/met19 mutant is a strict methionine-auxotroph). Hence a decreased flux in the PPP could limit sulfur assimilation under conditions where sulfur might already be limiting (see point 1 above).
5) In the proposed model, activation of the Gcn4-dependent pathway is merely a consequence of the rerouting and is not responsible for lower phosphate or higher trehalose. This assumption could easily be verified by combining ncs2/uba4 and gcn4 mutations and measuring phosphate and trehalose in the double mutants.
6) The effect of tRNA-thiolation on nucleotides synthesis should also be documented at the level of tri-phosphates (not only AMP and GMP) since ATP and GTP are by far the most abundant nucleotides and more directly reflect intracellular nucleotide content. At least AXP and GXP could be shown. Similarly, measurements of intracellular dNTPs would be relevant to confirm cell cycle analyses/hypotheses.
7) I do not understand how the authors can normalize on "cell number as well as biomass". I could not find the information in the Materials and methods section nor in the figure legend. This is a very important point that should be clarified. Cell number imperfectly reflects biomass if not corrected by the cell volume.
[Editors' note: further revisions were requested prior to acceptance, as described below.]
Thank you for resubmitting your work entitled "A tRNA modification balances carbon and nitrogen metabolism by regulating phosphate homeostasis" for further consideration at eLife. Your revised article has been favorably evaluated by Naama Barkai as the Senior Editor, and two reviewers, one of whom is a member of our Board of Reviewing Editors.
The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined in each of the two reviews shown below. Please make the appropriate revisions and provide a brief point-by-point explanation in response to the remaining reviewers' comments.
Reviewer #1:
I am generally satisfied that the addition of new experimental data and revisions of text address the major concerns with the previous version of the paper, except that the GCN4 mRNA reads still were not added to Figure 4—figure supplement 2.
Reviewer #2:
(Partially rephrased by Deputy Editor Detlef Weigel for clarity.)
General comment: the essential response points by the authors are difficult to extract from pages of justification which in my opinion turbidify rather than clarify the issues. Several of the original concerns have been satisfactorily addressed, but the following remain:
1) Perform new experiments to show that limitation for the sulfur-containing amino acids and/or inorganic sulfate, elicits the same key responses seen for the thiolation mutants, to support the claim that tRNA thiolation is used as a sensor for sulfur levels.
Figure 2G shows that absence of methionine elicits a much stronger response than the thiolation mutant does. Indeed, the increased flux toward AMP in the presence of methionine is diminished in the uba4 mutant but stays much higher than in the absence of methionine (same for either WT or uba4) and similar partial effects were observed for GMP (Figure 2—figure supplement 1D). These results indicate that tRNA thiolation contributes (very) partially to the methionine (sulfur) signal (when purine synthesis is used as a readout) suggesting the existence of other (yet unknown) mechanisms connecting sulfur and carbon metabolism. This should be discussed.
4) Expand the analysis of the RNA-Seq data of Figure 5 to show and discuss all of the transcriptional changes for genes in the PPP and trehalose and glycogen pathways.
As now shown (Figure 4C and Figure 4—figure supplement 3C) and stated by the authors "there are very minor changes in either transcript or RPF for any of the genes in these pathways". Consequently, the molecular mechanism resulting in rerouting the carbon flux away from the PPP toward trehalose synthesis is clearly not at the RNA or protein expression levels and is not yet elucidated. Hence the last sentence of the Discussion "we biochemically explain how phosphate homeostasis determines flux through different arms of carbon and nitrogen metabolism" appears overstated to me.
https://doi.org/10.7554/eLife.44795.030Author response
Essential revisions:
1) Perform new experiments to show that limitation for the sulfur-containing amino acids and/or inorganic sulfate, elicits the same key responses seen for the thiolation mutants, to support the claim that tRNA thiolation is used as a sensor for sulfur levels.
This is an important point that requires a substantial response, which also allows us to more clearly explain inferences made from earlier studies. Multiple lines of evidence, based on published work, along with this study, suggest that tRNA thiolation is not a starvation response, but a growth response, i.e. tRNA thiolation enables cells to fully respond to the presence of sulfur amino acids, and not starvation /absence of sulfur/sulfur amino acids. The loss of tRNA thiolation results in a more subtle metabolic phenotype, and understanding this is the primary point of this manuscript. To clarify further, the connection of tRNA thiolation (and related modifications) with sulfur amino acids and sulfur comes from our earlier work that demonstrate the following: (i) when comparing the amounts of thiolated tRNAs in methionine rich medium vs. medium with less methionine, the amount of thiolated tRNAs increased with increasing methionine availability and vice versa (see Laxman et al., 2013, Figures 1C, 1D, 1F, and supplementary figures). (ii) This same study quantified the absolute amounts of sulfur amino acid metabolites, and compared it with amounts of thiolated tRNAs. Both methionine and thiolated tRNAs (and closely related modified tRNAs) were present at low μM amounts in cells in normal medium, i.e. the amount of tRNA thiolation reflected the availability of methionine. (iii) The amounts of thiolated tRNAs are highest in cells entering a ‘growth state’ (high ribosomal biosynthesis, entry into the cell cycle etc.), as suggested from that study, as well as earlier gene expression studies (e.g. Tu et al., 2005). (iv) in thiolation mutants, the proteins involved in methionine salvage and biosynthesis increase in amounts over time (Laxman et al., 2013), further suggesting a coupling between the thiolation pathway and methionine. (v) the key enzyme in the thiolation pathway, Uba4, is strongly regulated by methionine availability, decreasing in methionine limited medium. These together show the coupling of tRNA thiolation with methionine/sulfur amino acids, pointing towards a role for tRNA thiolation when sulfur amino acids are abundantly present. Separately, several recent studies have suggested a multi-component growth program when methionine is abundant (including Sutter et al. Cell 2013, Yi et al. Mol Cell 2017). Pertinently, our recent study defined a methionine induced anabolic program, which centred around two arms: an increase in carbon metabolism (particularly PPP flux) towards nucleotide synthesis, and an increased utilization of amino acids for the same, when methionine is abundant (Walvekar et al., 2018). Therefore, to demonstrate this coupling of tRNA thiolation with the presence of sulfur amino acids, and the full response of sulfur amino acids, the more appropriate experiment will be to compare carbon flux towards nucleotides (via the PPP, as also elaborated in response to point #5 below) when abundant methionine is provided. Here, the prediction would be that if tRNA thiolation aided the methionine mediated metabolic program, this increase in nucleotide synthesis due to methionine would be less in thiolation mutants, compared to WT cells. We therefore carried out the following experiment: in WT cells growing in minimal glucose medium, or in this medium supplemented with methionine, using labeled 13C glucose pulsed into the system, we measured flux of carbon incorporation into nucleotides (also see response to point #5 below). Here, methionine strongly increases carbon flux into nucleotides in WT cells. Notably, this methionine dependent, increased flux through this pathway is strikingly lower in the thiolation mutants. This shows that for the full metabolic response due to methionine, leading to increased carbon flux towards nucleotide synthesis, tRNA thiolation is required. These data are now included in the new Figures 2G, Figure 2—figure supplement 1D), and are explained in detail in a new Results subsection (“Methionine induced carbon flux to nucleotide synthesis is dampened in thiolation mutants”), and Discussion. This data therefore more directly substantiates the coupling of tRNA thiolation with increased sulfur amino acid (methionine) availability, and in regulating the overall metabolic output mediated by methionine.
We note that sulfur starvation is a separate, far more complex phenomenon. Given the central nature of sulfur for a variety of essential biochemical functions, cells have a complex, coordinated response to this. This includes a downregulation of all methylation reactions due to reduced SAM, a strong induction of the MET regulon, and sulfur salvage/assimilation pathways, and a large reduction in growth (including low carbon metabolism), as has been demonstrated in different studies. When there is sulfur starvation, there is almost no thiolated tRNAs, or the enzyme required to make thiolated tRNAs, as has been demonstrated (Laxman et al., 2013). So a clear experiment in this condition to directly probe the coupling of tRNA thiolation with the absence sulfur amino acids, is not obvious, nor is it likely to clarify the question at hand.
We will separately also note that an earlier study (Boer VM et al., 2010) has made a connection between phosphate starvation (the experiment performed), and sulfur metabolism, where (we quote): “In phosphorus limitation, one of the most striking transcriptional patterns was, intriguingly, not directly related to phosphate metabolism, but instead to sulfate. SUL1, MMP1, MHT1, CYS3, MUP1, and SAM1 were strongly repressed by phosphorus limitation, with SUL1 in phosphorus limitation showing the most positive growth rate slope of any gene in any condition. This suggests a strong relationship between phosphate limitation and sulfur metabolism.”These data parallel our data (with sulfur metabolism as a starting point) where we note a coupling between sulfur metabolism and phosphate metabolism. If tRNA thiolation is indeed coupled with sulfur metabolism, then such a down-regulation of phosphate metabolism when tRNA thiolation is absent would therefore be a logical prediction.
Finally, in relation to this, and for the reference of the reviewers, we include some additional data. We and others have worked with conditions where reduced sulfur (methionine and related metabolites) are limiting (e.g. Xi and Tu BP MBoC 2011, Sutter et al. Cell 2013, Walvekar et al., 2018 etc.). We analyzed some transcription data sets from Walvekar et al., 2018, as well as some unpublished datasets (unrelated to this manuscript) from an ongoing study. Here, we find that in cells present in medium supplemented with methionine (compared to cells in somewhat methionine-limited conditions), several genes related to phosphate metabolism are induced. Here, in these condition phosphate is not limiting. While this correlates with observations made in this study, w.r.t the thiolation mutants, this also suggests a deeper coupling between sulfur and phosphate metabolism. This certainly deserves a more substantial investigation in future studies, well beyond the scope of this manuscript. We include this preliminary data in Author response image 1. We feel it is premature (and distracting) to include these data in this manuscript, since it takes away from the primary point of tRNA thiolation integrating methionine/sulfur sensing with overall metabolic outputs, via phosphate regulation. We hope collectively, we have sufficiently clarified and addressed all concerns raised.
Author response image 1
2) Cite the appropriate literature from the Rabinowitz and Broach labs (Boer et al., 2010 and Klosinslka et al., 2011) showing that nucleotide levels are reduced on Pi starvation, to support the claim that the reduced nucleotide levels in the thiolation mutants can be attributed to the reduced Pi levels in these strains.
In our original submission, we had cited the Boer et al., 2010 and Klosinska et al., 2011 references, but had not fully described this important hallmark of Pi starvation, while describing how phosphate starvation phenocopied the metabolic switch observed in thiolation mutants. We have now done so, substantially clarifying the text corresponding to the section related to Figure 6, as well as in the Discussion section.
We also now include an additional citation (Saldanha et al., 2004). This study shows that phosphate starvation itself results in a decrease in ribosomal genes. In the thiolation mutants, we also see slightly reduced ribosomal genes (RNA and ribo-seq), as shown in Figure 4—figure supplement 2C. Thus, this is an additional converging datapoint, where phosphate starvation phenocopies tRNA thiolation mutants. This is added to the Results subsection (“Phosphate depletion in wild-type cells phenocopies tRNA thiolation mutants”).
Summarizing additions in the text; in the result section where we show that phosphate starvation in the WT cells phenocopies thiolation mutants, we state that reduced nucleotides are a feature of phosphate starvation in WT cells (citing Boer et al., 2010, Klosinska et al., 2011), and that reduced ATP is a signature feature of phosphate starvation. We summarize our presented data where we show reduced de novo nucleotide synthesis (using both nitrogen and carbon labeling), as well as reduced steady state nucleotides, with an emphasis on the reduced ATP amounts observed in the thiolation mutants. This is summarized in the aforementioned Results subsection, and in the revised Discussion section.
3) Measure the ATP/GTP levels (or at least AXP/GXP) in the thiolation mutants rather than relying on measurements of the much less abundant AMP/GMP.
This is an important point raised by the reviewers, therefore we would like to elaborate and clarify this point in detail. The reviewers are correct that at steady-state, the amounts of ATP/GTP are much higher than AMP/GMP. Indeed, nucleotides are synthesized as mono phosphates, and then eventually form di and triphosphates, and cycle between these states. Therefore, if measuring only steady state changes, it is essential to show levels of all these. However (in addition to steady-state levels of nucleoside monophosphates), we had included data from stable-isotope based flux analysis of nucleotide synthesis, using pulse-labels (of both nitrogen and carbon), using LC-MS/MS. Unlike changes observed in steady-state amounts of a metabolite (which can come due to decreased synthesis or increased utilization), this more directly addresses changes in rates of nucleotide synthesis, and since the di and tri phosphates are made from the monophosphates, presenting monophosphates alone is usually sufficient. Given this difference between steady-state measurements, and stable-isotope based flux experiments, it is now well accepted to show the synthesis of the nucleoside monophosphates, and to indicate carbon incorporation into nucleotides. This is commonly used in studies from cancer metabolism, where nucleotide synthesis is the focus (examples from the de Berardinis, Locasale, Vander Heiden and Maddocks groups: Huang F et al., Cell Metabolism 2018, 28(3):369-382.e5, Reid MA et al., Nature Communications 2018, 9: 5442, Lunt et al. Mol Cell 2015, Labuschagne et al. Cell Reports 2014). We have ourselves used this approach in other studies (e.g. descriptive methods paperWalvekar et al., 2018, Laxman et al. Sci Signal 2014, Walvekar et al., 2018 etc.). One reason for measuring NMPs alone; while using LC-mass spectrometry, resolving and detecting NXPs have different challenges. NMPs are easily detected in positive-polarity modes (MS/MS), and separated in aqueous-organic phases like 0.1% formic acid/methanol. NTPs however ionize and fragment best in negative polarity modes, and require different reverse phase columns/solvents to efficiently separate them. So it is difficult to get both NMP and NTP information within the same MS run.
However, since this was a particular point raised by the reviewers, we have carried out separate experiments and included ADP and ATP measurements. We compared steady-state ATP levels in WT and thiolation mutants using standard biochemical assays, and saw a similar reduction in steady state ATP as we did for steady state NMPs. This is included as a supporting figure (Figure 1—figure supplement 2A and B). Next, we developed/optimized a new LC-MS/MS method where AMP, ADP and ATP could all be reliably, quantitatively measured in the same run. This is now described in the Materials and methods section. Using this, we repeated a 13C-glucose stable-isotope based flux experiment (as we had done for NMPs), and measured relative label incorporation into ADP and ATP as well. Here, we clearly observed a significant decrease in the relative label incorporation into ADP and ATP in the thiolation mutants, largely mirrorring the data shown earlier for AMP and GMP. These data substantially strengthen the observation that the thiolation mutants have reduced nucleotide synthesis, and are now included as a new figure panel, now Figure 2C, D, and collectively summarized in the Results section primarily related to Figure 2.
4) Expand the analysis of the RNA-Seq data of Figure 5 to show and discuss all of the transcriptional changes for genes in the PPP and trehalose and glycogen pathways.
We thank the reviewers for this suggestion. We have included the analysis of transcriptional and ribosome-footprint changes in the genes related to the PPP, as well as trehalose/glycogen synthesis. This is now included as a new figure (Figure 4D), and also described in the text in the relevant Results subsection (“Loss of tRNA thiolation results in reduced phosphate homeostasis (PHO) related transcripts and ribosome-footprints”, third paragraph). As can be seen, there are very minor changes in either transcript or RPF for any of the genes in these pathways, in the thiolation mutants compared to WT cells. Indeed, including this data as a figure reiterates that the metabolic rewiring, in central carbon metabolism, is largely dictated by flux and mass-action. This strengthens our assertion that the causal control point in the tRNA thiolation mutants is the dampened PHO response. For summarizing and illustrating this overall point, we also now include a mega-violin plot, comparing both RNA and RPF (in WT vs. thiolation mutants), for genes related to PHO, the PPP, trehalose/gly synthesis, amino acid biosynthesis, nucleotide biosynthesis and ribosomal subunits (Figure 4—figure supplement 3C). This should address any concerns the reviewers might have.
5) Use the proper time frame for the experiment in Figure 2 in order to measure the flux in the PPP and glycolysis pathways in the thiolation mutants, to confirm the rerouting hypothesis.
This is a fair point raised by the reviewers, which requires some clarification. As explained earlier for point 1, this experiment itself was very challenging to carry out, for the following reason: Yeast cells have very high rates of glycolysis and the PPP in standard medium containing glucose as a carbon source. In yeast growing in glucose, because of the very high rate of glycolysis, label saturation is almost instantaneous (in a couple of minutes), immediately after addition of label, for most intermediates in glycolysis and the PPP. Hence, in yeast experiments to measure relative rates of newly synthesized early intermediates of the PPP, using pulse-labels of stable isotopes of glucose (and following labelled carbon incorporation) are challenging. This is due to the fact that it is difficult to label and properly quench metabolites/process samples in time-frames of <5 min, and because most reliable measurements of metabolites require ~10 min post label addition for collection/processing. Indeed, even in more slowly growing, glucose addicted cancer cells, where flux is slower than yeast cells, and where longer time frames of sample collection after stable-isotope addition are used, changes in carbon label incorporation are usually seen with late PPP intermediates, e.g. ribose-5-phosphate. Therefore, even in these studies (performed in timescales of 30min-1 hour post label addition), carbon label incorporation into nucleotides (which come after ribose-5-phosphate) is accepted as an indicator of carbon flux via the PPP and one-carbon metabolism towards nucleotides (e.g., Reid MA et al., Nature Communications2018, 9: 5442, or Lunt et al. Mol Cell 2015). In yeast, given the even faster rates of glycolysis/PPP (compared to typical mammalian cells), it is therefore experimentally reliably possible to process samples only after a slightly longer time interval post label addition, which is why relative label incorporation into nucleotides is preferred. This was the reason we had initially presented data with carbon-label incorporation into NMPs, done ~5-10 minutes after labelled glucose addition. With the new data now included for carbon-label (from glucose) incorporation into ADP and ATP, as well as NMPs in Figures 2B, 2C and related text (also see response to point 3), we fully address this concern.
However, to directly address this point, we optimized even shorter time-points of metabolite extraction after 13C-glucose addition. In ~5 minutes, there already is substantial label saturation, so differences cannot be observed (shown inFigure 2—figure supplement 1B). Therefore, we attempted and reproducibly carried out experiments where metabolites were extracted/processed in <2 minutes after 13C glucose addition. Here, we also observe a significant reduction in relative label incorporation into the late PPP intermediate (and nucleotide precursor) ribose-5-phosphate, in the thiolation mutants (new Figure 2D). This therefore nicely corroborates our conclusions made from the data shown reduced carbon label incorporation into nucleotides (AMP/ADP/ATP) (Figures 2B, C). Experimental note: This difference may be even larger if anyone manages to do such a label addition-quench-process experiment in <1 minute, where label incorporation will be ideal, at <20%. The only reference we found where such a flux experiment has been done in yeast in glucose medium, with such a rapid time-frame is (van Heerden et al., Science 2014). We hope all concerns are addressed now with these new data.
6) Discuss the possibility that lower flux in the PPP might result in low NADPH and attendant diminished sulfate assimilation under conditions where sulfur is already limiting. Reduced sulfate assimilation could in turn limit thiolation and amplify the whole regulatory response.
We thank the reviewers for drawing this broader connection to NADPH, and allowing us to include a more nuanced discussion in this regard. We completely agree that nearly all of the thiolation mutant phenotypes can be explained based on NADPH flux and availability, and indeed believe this is so. We had not discussed this in the original submission, because any direct measurement and analysis of NADPH flux and pools in a rigorous, definitive way, is very difficult, and therefore it is difficult to obtain direct data to support this hypothesis. Demonstrated examples of reduction in NADPH flux, coming from decrease flux through the PPP and 1-carbon/folate cycles, have been shown only in rare studies like Fan J et al., Nature 2014, or Li Chen et al. Nature Metabolism (2019). Such studies require a combination of hydrogen-deuterium exchange and possibly mathematical modeling, and is beyond the capability of most labs in the world. Hence, we had cautiously avoided speculation and discussion in this regard. We have now included a nuanced discussion related to this point. As summarized earlier, the thiolation pathway allows a subtler, deeper coupling and integration of various nutrient cues, and modulation of metabolism, for enabling optimal growth. The collective result of the metabolic ‘squeeze’ exerted by the thiolation mutants, by restricting phosphate availability, is a reduction in flux through the PPP and one-carbon cycles, resulting in decreased nucleotide synthesis (as demonstrated). This will also decrease the production of NADPH, and reduce overall reductive biosynthesis capacity, and affect the assimilation of sulfates into sulfur amino acids. Conversely, in the presence of methionine, where there will be maximally thiolated tRNAs, we can connect two distinct observations. In an earlier study, we have shown that the presence of methionine itself increases flux through the PPP and carbon metabolism leading to nucleotide synthesis (Walvekar et al., 2018). We also show this more directly in this study (through measurements of carbon incorporation into nucleotide synthesis in the presence of methionine) (Figure 2G). Here, in the presence of methionine, the loss of thiolation results in a significantly smaller increase in this flux through this arm of carbon metabolism. Hence, the inferred amplified response in relation to tRNA thiolation would be as follows: in the presence of methionine, there is increased PPP and one-carbon flux towards nucleotide synthesis, which is amplified by the maximally thiolated tRNAs by ensuring sufficient phosphate availability. Conversely, reduced sulfate assimilation in turn limits thiolation, and thereby amplify the regulatory response, again using phosphate homeostasis as a control point. This is now an extended paragraph in the Discussion section (third paragraph).
For the information of the reviewers, we include a piece of data consistent with this interpretation: thiolation mutants show hydroxyurea (HU) sensitivity as shown in Figure 3E, and this would be as expected when there are decreased pools of nucleotides, and also explained by decreased NADPH pools available. Here, if this were true, one would predict a growth rescue in the presence of HU if excess reduced glutathione (GSH) is supplemented. This is because oxidized glutathione (GSSG) is reduced to GSH consuming two molecules of NADPH, and hence is a major sink of NADPH consumption. Further, GSH is itself made from sulfur amino acids, and is a major sink of reduced sulfur. By providing excess GSH, this sink can be relieved, and hence free up pools of NADPH. Notably, we now observe a rescue of HU sensitivity in the thiolation mutants upon supplementing GSH (see Author response image 2).
Author response image 2
We however feel that including such data without directly showing changes in NADPH flux is premature. However, we are happy to include a measured Discussion section in this regard, to make a cautious, complete interpretation of our data.
[Editors' note: further revisions were requested prior to acceptance, as described below.]
The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined in each of the two reviews shown below. Please make the appropriate revisions and provide a brief point-by-point explanation in response to the remaining reviewers' comments.
Reviewer #1:
I am generally satisfied that the addition of new experimental data and revisions of text address the major concerns with the previous version of the paper, except that the GCN4 mRNA reads still were not added to Figure 4—figure supplement 2.
We apologize for inadvertently not including this earlier. We have now included the GCN4 mRNA reads as an additional panel (Figure 4—figure supplement 2C), and text (subsection “Loss of tRNA thiolation results in reduced phosphate homeostasis (PHO) related transcripts and ribosome-footprints”, third paragraph). As shown, GCN4 mRNA reads are very similar in WT and thiolation mutant cells, and GCN4 transcript amounts do not change in the mutants.
Reviewer #2:
(Partially rephrased by Deputy Editor Detlef Weigel for clarity.)
General comment: the essential response points by the authors are difficult to extract from pages of justification which in my opinion turbidify rather than clarify the issues. Several of the original concerns have been satisfactorily addressed, but the following remain:
1) Perform new experiments to show that limitation for the sulfur-containing amino acids and/or inorganic sulfate, elicits the same key responses seen for the thiolation mutants, to support the claim that tRNA thiolation is used as a sensor for sulfur levels.
We optimized a simple sulfur-starvation experiment, and performed it as follows: We shifted WT cells growing in standard rich glucose medium, to either minimal glucose medium containing sufficient ammonium sulfate and other trace elements with inorganic sulfur, or to the same medium with sulfur limitation (where chloride salts replace the sulfate salts, as is described in Kankipati et al., 2015). Cells remained in this medium for ~2-4 hours (as described in the Materials and methods). At this time, we collected cells and measured sulfur amino acid related metabolites, and observed substantial decreases in the same, confirming sulfur amino acid limitation (now included as Figure 2—figure supplement 2A). Using this experimental set-up, we compared the key metabolic features of WT cells in this short-term sulfur limitation, with the thiolation mutants as a reference point. The key metabolic features of the thiolation mutants are (i) amino acid accumulation, (ii) decreased nucleotide amounts, and (iii) increased trehalose amounts (due to the re-routing of glucose towards storage carbohydrates). This is also accompanied by an induction of Gcn4 in the thiolation mutants. In WT cells subjected to short-term sulfur limitation, we observe (i) strong accumulation of amino acids (now included as Figure 2—figure supplement 2C), (ii) decreased nucleotides (decreased AMP is shown, now included as Figure 2—figure supplement 2D), and (iii) a strong increase in trehalose synthesis (now included as Figure 2—figure supplement 2). Finally, in a simple sulfur amino acid limitation shift experiment, we also see a strong induction of Gcn4 in the WT cells (now included as Figure 2—figure supplement 2F). For easy comparison, we show the results from the thiolation mutants side-by-side in these figures.
These data compellingly show that sulfur starvation phenocopies tRNA thiolation mutants.
After including these data, we make the following additions to the Results, stating:
“..Furthermore, in a converse experiment, we subjected WT cells to brief inorganic sulfur/sulfur amino acid limitation, to determine the metabolic signature of cells in this condition (experimental design shown in Figure 2—figure supplement 2A).[…] Thus, WT cells subject to sulfur amino acid limitation phenocopied the metabolic signature tRNA thiolation mutants.”
Appropriate additions have also been made to the Materials and methods, and the supplementary figure legends.
We hope these data address all remaining concerns of the reviewer.
Figure 2G shows that absence of methionine elicits a much stronger response than the thiolation mutant does. Indeed, the increased flux toward AMP in the presence of methionine is diminished in the uba4 mutant but stays much higher than in the absence of methionine (same for either WT or uba4) and similar partial effects were observed for GMP (Figure 2—figure supplement 1D). These results indicate that tRNA thiolation contributes (very) partially to the methionine (sulfur) signal (when purine synthesis is used as a readout) suggesting the existence of other (yet unknown) mechanisms connecting sulfur and carbon metabolism. This should be discussed.
We completely agree that the tRNA thiolation does not regulate all methionine dependent responses. In fact we do not suggest these modifications as the sole regulators of methionine dependent responses. For absolute clarity, we have now included the following text stating “Notably, while the extent of this methionine-dependent induction of carbon flux into nucleotides was significantly reduced, it was not completely abolished in the thiolation mutant (Δuba4), indicating the involvement of additional as yet unknown pathways (Figure 2G, and Figure 2—figure supplement 1D).”
4) Expand the analysis of the RNA-Seq data of Figure 5 to show and discuss all of the transcriptional changes for genes in the PPP and trehalose and glycogen pathways.
As now shown (Figure 4C and Figure 4—figure supplement 3C) and stated by the authors "there are very minor changes in either transcript or RPF for any of the genes in these pathways". Consequently, the molecular mechanism resulting in rerouting the carbon flux away from the PPP toward trehalose synthesis is clearly not at the RNA or protein expression levels and is not yet elucidated. Hence the last sentence of the Discussion "we biochemically explain how phosphate homeostasis determines flux through different arms of carbon and nitrogen metabolism" appears overstated to me.
We apologize for this overstatement. We have now changed the text, and the final lines read: “Concluding, here we discover that a sulfur amino acid dependent tRNA modification (thiolated U34) enables cells to appropriately balance amino acid and nucleotide levels and regulate metabolic state, by controlling phosphate homeostasis. More generally, we suggest how phosphate homeostasis can impact flux through different arms of carbon and nitrogen metabolism.”
https://doi.org/10.7554/eLife.44795.031Article and author information
Author details
Funding
Science and Engineering Research Board, DST (PDF/2016/000416)
- Ritu Gupta
Science and Engineering Research Board, DST (PDF/2015/000225)
- Adhish S Walvekar
Wellcome Trust/DBT India Alliance (IA/I/14/2/501523)
- Sunil Laxman
Department of Biotechnology, Govt. of India
- Sunil Laxman
inStem
- Sunil Laxman
National Institutes of Health (R35 GM124976)
- Premal Shah
National Institutes of Health (R01 DK056645)
- Premal Shah
National Institutes of Health (R01 DK109714)
- Premal Shah
Human Genetics Institute of New Jersey at Rutgers University (Start-up funds)
- Premal Shah
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Acknowledgements
We acknowledge the use of the NCBS/inStem/CCAMP mass spectrometry facility for LC-MS/MS instrument support. We thank Dr. Anjana Badrinarayanan for the use of her live-cell imaging microscope system. We thank Claudio de Virgilio, Nikolai Slavov, Sider Penkov, and Sriram Varahan for critical comments on this manuscript. RG and ASW acknowledge the Department of Science and Technology and Science and Engineering Research Board (DST-SERB) national postdoctoral fellowships (PDF/2016/000416 and PDF/2015/000225 respectively). SLA is supported by an Intermediate Fellowship from the Wellcome Trust-DBT India Alliance (grant number IA/I/14/2/501523), as well as institutional support from inStem and the Dept. of Biotechnology (Govt. of India). PS is supported by grants NIH R35 GM124976, and subcontracts from NIH R01 DK056645 and NIH R01 DK109714 as well as start-up funds from the Human Genetics Institute of New Jersey at Rutgers University.
Senior Editor
- Naama Barkai, Weizmann Institute of Science, Israel
Reviewing Editor
- Alan G Hinnebusch, Eunice Kennedy Shriver National Institute of Child Health and Human Development, United States
Reviewer
- Bertrand Daignan-Fornier, Université de Bordeaux IBGC UMR 5095, France
Version history
- Received: December 29, 2018
- Accepted: June 30, 2019
- Accepted Manuscript published: July 1, 2019 (version 1)
- Version of Record published: August 9, 2019 (version 2)
Copyright
© 2019, Gupta 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.
Metrics
-
- 4,507
- Page views
-
- 659
- Downloads
-
- 38
- Citations
Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
A tRNA modification balances carbon and nitrogen metabolism by regulating phosphate homeostasis
eLife 8:e44795.
https://doi.org/10.7554/eLife.44795
Further reading
-
- Biochemistry and Chemical Biology
- Cell Biology
Pancreatic a-cells secrete glucagon, an insulin counter-regulatory peptide hormone critical for the maintenance of glucose homeostasis. Investigation of the function of human a-cells remains a challenge due to the lack of cost-effective purification methods to isolate high-quality a-cells from islets. Here, we use the reaction-based probe diacetylated Zinpyr1 (DA-ZP1) to introduce a novel and simple method for enriching live a-cells from dissociated human islet cells with ~ 95% purity. The a-cells, confirmed by sorting and immunostaining for glucagon, were cultured up to 10 days to form a-pseudoislets. The a-pseudoislets could be maintained in culture without significant loss of viability, and responded to glucose challenge by secreting appropriate levels of glucagon. RNA-sequencing analyses (RNA-seq) revealed that expression levels of key a-cell identity genes were sustained in culture while some of the genes such as DLK1, GSN, SMIM24 were altered in a-pseudoislets in a time-dependent manner. In conclusion, we report a method to sort human primary a-cells with high purity that can be used for downstream analyses such as functional and transcriptional studies.
-
- Biochemistry and Chemical Biology
- Cell Biology
Eukaryotic cells control inorganic phosphate to balance its role as essential macronutrient with its negative bioenergetic impact on reactions liberating phosphate. Phosphate homeostasis depends on the conserved INPHORS signaling pathway that utilizes inositol pyrophosphates and SPX receptor domains. Since cells synthesize various inositol pyrophosphates and SPX domains bind them promiscuously, it is unclear whether a specific inositol pyrophosphate regulates SPX domains in vivo, or whether multiple inositol pyrophosphates act as a pool. In contrast to previous models, which postulated that phosphate starvation is signaled by increased production of the inositol pyrophosphate 1-IP7, we now show that the levels of all detectable inositol pyrophosphates of yeast, 1-IP7, 5-IP7, and 1,5-IP8, strongly decline upon phosphate starvation. Among these, specifically the decline of 1,5-IP8 triggers the transcriptional phosphate starvation response, the PHO pathway. 1,5-IP8 inactivates the cyclin-dependent kinase inhibitor Pho81 through its SPX domain. This stimulates the cyclin-dependent kinase Pho85-Pho80 to phosphorylate the transcription factor Pho4 and repress the PHO pathway. Combining our results with observations from other systems, we propose a unified model where 1,5-IP8 signals cytosolic phosphate abundance to SPX proteins in fungi, plants, and mammals. Its absence triggers starvation responses.
