Lipidome remodeling in response to nutrient replenishment requires the tRNA modifier Deg1/Pus3 in yeast

In the yeast Saccharomyces cerevisiae, the absence of the pseudouridine synthase Pus3/Deg1, which modifies tRNA positions 38 and 39, results in increased lipid droplet (LD) content and translational defects. In addition, starvation‐like transcriptome alterations and induced protein aggregation were observed. In this study, we show that the deg1 mutant increases specific misreading errors. This could lead to altered expression of the main regulators of neutral lipid synthesis which are the acetyl‐CoA carboxylase (Acc1), an enzyme that catalyzes a key step in fatty acid synthesis, and its regulator, the Snf1/AMPK kinase. We demonstrate that upregulation of the neutral lipid content of LD in the deg1 mutant is achieved by a mechanism operating in parallel to the known Snf1/AMPK kinase‐dependent phosphoregulation of Acc1. While in wild‐type cells removal of the regulatory phosphorylation site (Ser‐1157) in Acc1 results in strong upregulation of triacylglycerol (TG), but not steryl esters (SE), the deg1 mutation more specifically upregulates SE levels. In order to elucidate if other lipid species are affected, we compared the lipidomes of wild type and deg1 mutants, revealing multiple altered lipid species. In particular, in the exponential phase of growth, the deg1 mutant shows a reduction in the pool of phospholipids, indicating a compromised capacity to mobilize acyl‐CoA from storage lipids. We conclude that Deg1 plays a key role in the coordination of lipid storage and mobilization, which in turn influences lipid homeostasis. The lipidomic effects in the deg1 mutant may be indirect outcomes of the activation of various stress responses resulting from protein aggregation.

In a previous study, we have shown that absence of either Elongator or Deg1 function causes an imbalance in neutral lipid levels (Bozaquel-Morais et al., 2018).Both deletion strains present a higher content of lipid droplets (LD), which store the neutral lipids triacylglycerol (TG) and steryl ester (SE), and are more resistant to the acetyl-CoA carboxylase (Acc1) inhibitor soraphen A (sorA).Acc1 is the main regulator of fatty acid synthesis and catalyzes carboxylation of acetyl-CoA to malonyl-CoA, the building block of lipids.It is finely regulated by the AMPK/Snf1 kinase, which upon phosphorylation, inhibits Acc1 activity (Figure 1a).Deletion of AMPK/Snf1 in yeast leads to a high LD phenotype, inability to ferment sucrose and to inositol auxotrophy (Bozaquel-Morais et al., 2010, 2017;Shirra et al., 2001).The first trait is attributed directly to Acc1 hyperactivity.Another report that studied hyperactivity of Acc1 by substituting the phosphorylatable serine residue (S1157A) has shown that Acc1 phosphorylation regulates the proportion of C16 and C18 fatty acids in free fatty acids, glycerophospholipids and neutral lipids (Hofbauer et al., 2014).This regulation controls the proper functioning of membranes through sequestration of the transcription factor Opi1 by binding to C16 phosphatidic acids, leading to AMPK/Snf1 regulation of fatty acid chain length.As phosphatidic acid is an intermediate between TG and phospholipid synthesis, their synthesis is coordinated.
It was also reported that Acc1 overexpression regulates the sterol and squalene content of the cell (Shin et al., 2012) showing tight coupling between Acc1 activity, neutral lipid and phospholipid synthesis to ensure lipid homeostasis.In this study, we investigated the lipid-specific effects of deg1 mutation in more detail.Our results show that a deg1 mutant deregulates the coupling between neutral lipid synthesis and phospholipids, leading to an imbalance between TG, SE and phospholipids during growth, independently of Acc1 activity.

| Genetic interaction of deg1 and ACC1 S1157A in soraphen A resistance
It was previously shown that a deg1 mutation induces partial resistance against the Acc1 inhibitor sorA and increases cellular LD content (Bozaquel-Morais et al., 2018).In some cases, loss of tRNA modifications is known to cause tRNA destabilization via the activation of exonucleolytic decay, which is routinely suppressed by mutation of MET22 (Phizicky & Hopper, 2023;Figure S1a).Since MET22 is neither required for the high LD nor the sorA resistance phenotypes of deg1 (Figure S1b,c), tRNA destabilization is unlikely mechanistically involved.Consistent with this assumption, a previous study revealed in deg1 mutants unaltered levels of tRNA Gln UUG, a tRNA that critically depends on the Deg1-dependent pseudouridylation for function (Han et al., 2015).Since upregulation of Acc1 activity affects LD and sorA phenotypes in a similar manner (Hofbauer et al., 2014), we investigated whether the deg1 effect might occur through Acc1 activation, possibly due to interference with the inhibitory phosphorylation at S1157 (Figure 1a).We generated a yeast mutant carrying the hyperactivated ACC1 S1157A allele, that has been shown to be resistant to sorA (Hofbauer et al., 2014;Shen et al., 2004;Vahlensieck et al., 1994), as the only source of Acc1 and deleted the DEG1 gene in this background.When the sorA phenotypes of the respective single mutants and the double mutants were compared, additive resistance was observed (Figure 1b).
Hence, deg1 mutation can increase sorA resistance in a background where inhibitory phosphorylation of Acc1 at S1157 is prevented.To test whether other potential Snf1-dependent phosphorylation sites (such as S659; Shi et al., 2014) are involved in the sorA phenotype of deg1 cells, we combined deg1 and snf1 mutations and compared single and double mutant sorA phenotypes.Additivity in sorA resistance was again observed (Figure 1d), indicating that the deg1 effect on sorA resistance (and potentially Acc1 activity) hardly arises from changes in Snf1-dependent Acc1 phosphorylation targets.To verify that the S1157 phosphoregulation circuit in deg1 mutants is in general functional, we removed the negative Snf1 regulator Reg1 (Ruiz et al., 2011; Figure 1a) and tested for alteration of the deg1 sorA phenotype.Yeast mutants lacking REG1 were previously shown to exhibit a lowered LD level and increased sorA sensitivity (Bozaquel-Morais et al., 2010).As expected for a situation where Snf1 can still be activated and inhibit Acc1 via phosphorylation, increased sorA sensitivity was observed in absence of Reg1 regardless of the presence or absence of Deg1.Therefore, phosphoregulation of Acc1 via Snf1 is functional in the deg1 mutant (Figure 1c).

| Quantification of specific decoding errors in deg1 mutants
To obtain a more comprehensive picture of the effect of deg1 mutation on overall decoding, we mined a previously published transcriptional profiling data set for various tRNA modification mutants (Chou et al., 2017) and extracted ribosomal A-site occupancy changes for all codons in deg1 mutants normalized to the wild type.As shown in Figure S2a, several codons display higher A-site occupancy in deg1 mutants, including the Gln CAA codon as well as the Asn codons AAC and AAU.Higher A-site occupancy implies that the ribosome slows down at these codons during translation and indicates a reduced decoding efficiency in the deg1 mutant.
Similar to the CAA decoder tRNA Gln UUG, tRNA Asn GUU, decoding both AAC and AAU also contains a Deg1-mediated pseudouridine (Boccaletto et al., 2018;Jühling et al., 2009).We focused on these three codons for further analysis, since in vivo misreading reporters were available (Salas-Marco & Bedwell, 2005) and allowed the quantification of translational errors that might be associated with ribosomal slow-down in deg1 mutants.The assay utilizes a dual luciferase reporter encoding a renilla-firefly luciferase (R-luc/Fluc) fusion protein.In addition to a control encoding functional R-luc/F-luc, plasmids were used that encode mutated F-luc variants (K529N, K529Q and H245Q), which are inactive but can regain activity due to specific decoding errors (Figure S2b).In detail, misreading of Asn codons AAC or AAU by tRNA Lys in absence of Deg1-dependent modification of tRNA Asn may be increased and cause restoration of active F-luc K529N .An analogous misreading event triggered by inefficient decoding by a hypomodified tRNA was previously demonstrated (Khonsari & Klassen, 2020).
Similarly, misreading of CAA by tRNA His in F-luc H245Q or by tRN-A Lys in F-luc K529Q may be promoted by inefficient decoding of the CAA codon in deg1 mutants.Neither tRNA Lys UUU nor tRNA His- GUG, which could misread AAC/AAU or CAA, respectively and subsequently restore F-luc activity, contain a Deg1-dependent pseudouridine (Boccaletto et al., 2018;Jühling et al., 2009).We measured F-luc/R-luc ratios of the different reporters and the control encoding fully active luciferases from 10 biological replicates for each construct/mutant and calculated misreading frequencies by normalizing F-luc/R-luc ratios for the mutant plasmids (pDB825, pDB827, pDB872 and pDB865) to the control encoding fully active luciferases (pDB688; Figure S2).Indeed, deg1 mutants displayed increased error rates at AAC and AAU codons, whereas no significant change was recorded for the two CAA misreading reporters.It should be noted that other misreading events at the CAA codon (e.g., by a distinct tRNA, which cannot be assessed with this assay) may still be increased.Results obtained for AAC/ AAU misreading, however, demonstrate that specific misreading events are indeed elevated in deg1.

| Effect of deg1 mutation on LD levels in absence of inhibitory Acc1 phosphorylation
Since the genetic interaction of deg1 and ACC1 S1157A in sorA resistance argued for an S1157-phosporylation independent effect in the tRNA modification mutant, we compared cellular LD levels in WT, deg1, ACC1 S1157A and the double mutant in both, stationary and exponential growth phases.As previously described (Bozaquel-Morais et al., 2018;Hofbauer et al., 2014), both single mutations led to an increase in total LD levels as detected by BODIPY staining (Figure 1e).Importantly, we observe a further increase in LD levels in the double mutant (deg1 ACC1 S1157A ) when compared to either single mutant (Figure 1e,f).This result suggests that the tRNA modification defect can increase LD levels via a mechanism that is not entirely dependent on changes in the phosphorylation status of S1157 in Acc1.
Since Deg1 is important for decoding of several codons (Figure S2a) and improves functioning of tRNA in translation (Han et al., 2015;Klassen et al., 2016), we reasoned that its absence may affect the biosynthesis and cellular availability of Acc1.To test this, we utilized a strain expressing Acc1-GFP (enabling detection of Acc1 levels), introduced a deg1 deletion and compared Acc1-GFP protein levels by Western analysis.As shown in Figure 1g, the tRNA modification mutant indeed exhibits reduced Acc1-GFP levels, despite the increased tolerance toward the Acc1 inhibitor drug.Thus, a translational defect of Acc1 mRNA or decreased protein stability might account for reduced Acc1 levels and the observed sorA and LD phenotypes are not linked to increased Acc1 levels in absence of Deg1.

| Genetic interaction of deg1 and ACC1 S1157A with respect to growth in absence of inositol
Transcription of INO1 encoding inositol-3-phosphate synthase (Donahue & Henry, 1981) is known to be regulated by Acc1 activity.In the presence of activated Acc1, INO1 expression is downregulated resulting in reduced growth on inositol-free medium (Hofbauer et al., 2014).Hence, if deg1 mutation would upregulate Acc1 activity by interference of S1157 phosphorylation or through a different mechanism, reduced growth on inositol-free medium would be expected.Investigation of this phenotype, however, revealed that deg1, unlike ACC1 S1157A , does not reduce growth in absence of inositol.Instead, loss of DEG1 partially rescues inositol-free growth of the ACC1 S1157A mutant, suggesting an Acc1-independent effect on lipid homeostasis (Figure 2a).Similar to deg1 mutation, loss of the transcriptional repressor Opi1 was also shown to suppress the inositol auxotrophy of an ACC1 S1157A mutant by restoring INO1 expression (Hofbauer et al., 2014).Loss of Opi1 was also reported to result in inositol overproduction and secretion (Greenberg et al., 1982).Since deg1 equals opi1 in genetic interaction with ACC1 S1157A on inositol-free medium, we tested whether deg1 mutation also induces inositol secretion.
While inositol secretion of opi1 mutants was verified, no such effect is presented by the deg1 mutant (Figure 2b).

| Quantification of triacylglycerol and SE levels by thin layer chromatography
Since genetic analysis presented above suggested differential effects of ACC1 S1157A and deg1 on lipid-related phenotypes, we directly compared neutral lipid levels in WT, deg1 and ACC1 S1157A mutants in both exponential and stationary growth phases by thin layer chromatography (TLC) (Figure 3a).As shown by densitometric analysis of TLC images (Figure 3b), ACC1 S1157A mutation resulted in a drastic increase in TG levels in both growth phases, which is in agreement with previously published data (Hofbauer et al., 2014).
In contrast, loss of Deg1 hardly affected TG levels in the stationary phase and led to a moderate increase in the exponential phase of growth.Instead, SE levels were heavily induced in deg1, but not in ACC1 S1157A and this effect was detected in both, exponential and stationary phase.Thus, deg1 mutation significantly impacts neutral lipid composition with a strong upregulation of SE and more moderate induction of TG.We assume that the effect of increased LD abundance in the tRNA modification mutant involves an alteration of the relative abundances of the two neutral lipids present in this organelle.
F I G U R E 2 Inositol phenotypes of ACC1 S1157A and deg1 mutants.(a) Growth of the indicated strains was determined in YNB-based minimal medium containing (+) or lacking (−) inositol.(b) Inositol excretion assay for WT, deg1 and opi1 (control) cells.Strains were streaked at the bottom of a YNB plate lacking inositol.At a 90° angle, 5 spots of cell suspensions (5 μL, OD600 nm = 1) of the indicator/ control strains ino1 and WT were applied.The ino1 indicator strain can only grow if the tester strain excretes inositol into the medium.The opi1 strain is known to excrete inositol and was employed as a positive control.the top 50 most significantly changed lipid species together with volcano plots (Figure S3a,b) shows that the amount of TG is higher in deg1 mutant while a lower phospholipid levels are observed in both stationary and exponential phases (Figure 4a,b).Next, we analyzed the total content (sum of all species within a class) of neutral lipids.As shown in Figure 4c,d, neutral lipid synthesis in S. cerevisiae is highly regulated and responds to growth phase as a notable increase of total SE and TG occurred at stationary phase in WT cells.Such storage lipid accumulation allows the cells to maintain energy homeostasis in the absence of nutrients.As cells are shifted from starvation condition to a high glucose medium, neutral lipids are mobilized and the resulting fatty acids contribute to formation of phospholipids (Kurat et al., 2006;Rajakumari et al., 2010).These results contrast to those of deg1 mutant cells where the levels of SE and TG species are significantly higher compared to WT at stationary phase.Although they undergo lipolysis upon shifting to exponential phase, a significantly increased level is maintained in relation to WT (Figure 4c,d).In contrast to TG, diacylglycerol was moderately reduced in the deg1 mutant at exponential phase and unaffected in stationary phase (Figure 4e).
The main TG/SE and FFA species that were deregulated in deg1 are shown in Figure S4a-e.It should be mentioned that although both SE (18:1) and SE (16:1) were increased (Figure S3a,b), free ergosterol was not significantly altered in deg1 mutant.
Since SE and TG are upregulated in the deg1 mutant, we analyzed whether the lipidomic profile of the deg1 mutant resembled to that reported for the hyperactive ACC1 S1157A mutant, in which the acyl C18/C16 ratio in cellular lipids and free fatty acids is drastically increased (Hofbauer et al., 2014).Our lipidomic profiles for the deg1 mutant in contrast revealed that the C18/C16 ratio of free fatty acids and TG/SE acyl chains were not significantly affected in stationary phase.However, a moderate increase of the C18/C16 acyl chain ratio in diacylglycerol, TG and SE was detected exclusively in exponential phase (Figure 4f).
In general, neutral lipid and phospholipid synthesis compete for the limited pool of free fatty acids.Thus, when neutral lipid synthesis is increased, less fatty acids are available for the synthesis of phospholipids, which are readily synthetized when neutral lipids are degraded.Indeed, increased levels of neutral lipid was accompanied by reduced phospholipid content in the deg1 mutant (Figure 4a,b).

Such negative correlation between neutral lipids and phospholipids
was also seen in the WT when stationary and exponential growth phases were compared; the content of phospholipids was higher in exponential phase compared to stationary phase, when neutral lipids accumulate (Figure 5a-e phosphatidylethanolamine, and lysophosphatidylethanolamine), was significantly lower in the deg1 mutant (Figure 5).Moreover, the deg1 mutant did not display increased levels of C18 acyl chains in phospholipids as observed in the ACC1 S1157A mutant (Hofbauer et al., 2014).Our observation of major changes in neutral lipids and phospholipid composition, but not in the general acyl length of all cellular lipids in the tRNA modification mutant, suggests a mechanism distinct from Acc1 activation.Our results indicate that in deg1 mutant, storage lipids are not degraded properly upon switching to exponential phase, which may limit the availability of precursors for phospholipid synthesis and eventually, result in an imbalance between neutral and phospholipid levels.
In yeast, pleiotropic phenotypes of Elongator and deg1 mutants also overlap, including the induction of protein aggregates and high LD content (Bozaquel-Morais et al., 2018;Khonsari et al., 2021;Nedialkova & Leidel, 2015).Since different types of translational errors are induced in absence of functional Elongator (Joshi et al., 2018;Tavares et al., 2021;Tükenmez et al., 2015), we tested whether loss of Deg1 could also affect translational fidelity.While a tRNA lacking modifications could itself become error-prone in decoding, an indirect effect linked to an inefficiency of correct decoding is also possible (Khonsari & Klassen, 2020;Tavares et al., 2021).We show that in deg1 mutants two Asn codons read by a Deg1 target tRNA become more efficiently misread by another tRNA not modified by Deg1 (tRNA Lys ; Figure S2c).Considering that ribosomal profiling data indicated a larger number of inefficiently decoded codons, such indirect loss of translational fidelity might be more widespread in deg1 and possibly mechanistically linked to the observed induction of protein aggregates.A similar concept was proposed for protein aggregates induced in Elongator mutants and supported by proteomic analysis (Tavares et al., 2021).
In this study, we investigated the lipid-specific effects of the deg1 mutation in yeast in more detail.S1157A exchange in Acc1 prevents inhibitory S1157 phosphorylation by Snf1 and in turn upregulates its activity, LD levels and sorA resistance (Hofbauer et al., 2014).Since ACC1 S1157A and deg1 mutation induce similar lipid phenotypes, we investigated whether loss of the tRNA modifier could interfere with Acc1 phosphorylation at S1157.Genetic analysis, however, revealed that all analyzed deg1 lipid phenotypes are independent of S1157 phosphorylation (Figure 1).In addition, deg1 mutants do not phenocopy ACC1 S1157A with respect to inositol auxotrophy and the effects of both mutations on specific lipid classes are strikingly distinct.
ACC1 S1157A and snf1 mutants upregulate TG (but not SE) with a concomitant increase of the C18/C16 ratio of acyl chains in fatty acids, TG and phospholipids (Hofbauer et al., 2014).Absence of DEG1 in contrast strongly induces SE levels, whereas TG levels increase more subtly (Figures 3a-c and 4c,d).In addition, a general increase in the C18/C16 ratio of fatty acyl chains is not observed in deg1 cells (Figure 4f).These differences support the conclusion that the lipid phenotypes of deg1 arise independently of changes in Acc1 phosphorylation and activation.
A major effect in deg1 mutants is the diminished induction of phospholipid synthesis when stationary phase cells are refed with fresh medium and switch to exponential growth.Induction of phospholipid synthesis at the expense of storage lipid depletion is observed in the WT but occurs less efficiently in deg1 cells (Figure 4).
Hence, lipid remodeling to support phospholipid induction is impaired in deg1, which might involve reduced activity of lipases involved in storage lipid remobilization.However, published deg1 transcriptome data (Chou et al., 2017) do not indicate reduced expression of the lipase genes TGL1, TGL3, TGL4, TGL5 or YEH1 and YEH2.Instead, there are signs of inappropriate stationary phase gene expression programs executed in the absence of nutrient depletion.Similar to Elongator mutants, amino acid biosynthesis genes are globally upregulated in deg1 and stationary phase marker genes SNZ1 and SNO1 (Padilla et al., 1998) are the most strongly induced mRNAs (Chou et al., 2017).This may indicate an altered metabolic state of deg1 mutants, resembling stationary phase or nutrient depletion.Under such conditions, storage lipid levels are normally high, and lipolysis is inhibited, and these characteristics are observed in the deg1 mutant despite the refed growth condition and abundant nutrient supply.Hence, the lipid phenotype of deg1 fits into the assumption of an inappropriate stationary phase metabolic state.
Since combined deg1 elp3 mutants display signs of reduced TOR activity including autophagy activation and gene expression changes (Bruch et al., 2020), it is possible that nutrient availability is incorrectly sensed or signaled in the absence of critical modifications at or around the tRNA anticodon loop.A study in fission yeast further supports the implication of Elongator-dependent tRNA modifications in TOR signaling (Candiracci et al., 2019).
Interestingly, budding yeast TOR is also known to regulate neutral lipid dynamics in yeast (Madeira et al., 2015), and the reduced TOR activity in combined elp3 deg1 mutants (Bruch et al., 2020) might point to a connection between tRNA modification, TOR activity and lipid homeostasis.In detail, however, TOR inhibition by rapamycin increases TG rather than SE levels (Madeira et al., 2015).Since deg1 mutation affects SE levels more drastically than TG (Figures 3   and 4), other factors than TOR inhibition alone are likely involved in deregulation of lipid homeostasis in tRNA modification mutants.
It should be noted that U34 tRNA modification defects induce metabolic shifts partly resembling starvation situations (Gupta et al., 2019;Gupta & Laxman, 2020;Karlsborn et al., 2016;Laxman et al., 2013).Also, deg1 and Elongator mutations were found to significantly alter cellular amino acid levels (Mülleder et al., 2016), demonstrating that distinct tRNA modification defects may similarly affect metabolism.Such common effect might also be relevant for imbalances in lipid homeostasis observed in this study.In support, a deletion of TUM1, a gene required along with Elongator for mcm 5 s 2 U modification, was found to upregulate SE, but not TG levels, thus resembling part of the deg1 lipid phenotype (Uršič et al., 2017).In addition, mutations in yeast tRNA genes promoting mistranslation at alanine and glycine codons induce major changes in lipidome composition (Araújo et al., 2018).Hence, common lipid phenotypes may be induced by different tRNA defects and this might represent an additional aspect of the widespread metabolic changes caused by interference with the fine-tuning of tRNA function.It might be speculated that such common effects also contribute to related disease phenotypes in humans carrying mutations in anticodon loop modification genes.Since several of the metabolically important tRNA modifications are also required to maintain proteostasis, it seems possible that lipidomic effects (as observed in deg1) are indirect outcomes of protein aggregation which induces mis-activation of nutrient stress responses.

| Yeast strains
All strains used in this study are listed in Table 1.Standard methods were used for the cultivation of yeast (Sherman, 2002).Yeast strains were grown in yeast peptone dextrose medium (YPD) or in yeast nitrogen base (YNB) minimal medium lacking specific nutrients for selection of mutants and plasmids.Individual gene knockouts were introduced using PCR-based deletion cassettes and the pUG plasmid system as outlined (Gueldener et al., 2002).Yeast transformation was done according to (Gietz & Schiestl, 2007).

| Western blot
To isolate proteins from yeast, liquid cultures were grown in 50 mL YPD overnight at 30°C.The following day, fresh medium was added to adjust cultures to OD600 nm = 0.02 and grown to an OD600 nm = 1.0.Yeast (3 OD units) was harvested, washed and resuspended in 0.1 M NaOH followed by incubation for 10 min (Kushnirov et al., 2006).Cells were washed, resuspended in 50 μL 5× Laemmli buffer (312.5 mM Tris-HCl pH 6.8, 10% SDS w/v, 50% glycerol v/v, 25% β-mercaptoethanol v/v, 0.01% bromophenol blue w/v) and incubated for 10 min at 99°C.After separation of proteins by SDS-PAGE, transfer to PVDF membranes was carried out using the Mini PROTEAN® system (Bio-Rad) following recommendations by the supplier.

| Codon-specific misreading
Misreading assay was conducted using renilla/firefly luciferase plasmids pDB688 (normalization control), pDB825, pDB827, pDB872 and pDB865 (Salas-Marco & Bedwell, 2005).Plasmids were transformed into wild type and deg1 mutants and individual colonies used to inoculate uracil-free YNB medium.Renilla and firefly luminescence from independent cultures was measured using the dual luciferase reporter assay kit (Promega, Fitchburg, USA) and a GloMax luminometer (Promega).Ten biological replicates were conducted for each strain and condition and three technical replicates were measured per culture.Mistranslation rates were calculated as described before (Khonsari & Klassen, 2020).

| LD analysis by BODIPY staining
To visualize LDs, the fluorescent dye BODIPY® 493/503 was used.

| Neutral lipid analysis by thin layer chromatography
Lipids were extracted from yeast cells based on a Bligh and Dyer modified protocol (Bourque & Titorenko, 2009), dried under nitrogen and stored −20°C until used.Neutral lipids were separated in silica plates using a two-step separation system.The first separation to 2/3 of the height of the plate used light petroleum/diethyl ether/acetic acid (35:15:1 v/v) as a solvent system, followed by light petroleum/diethyl ether (49:1 v/v) solvent (Schmidt et al., 2013).The standards used were SE, TG, diacylglycerol, free fatty acids and monoacylglycerol (Sigma-Aldrich St. Louis, MO).Lipids were revealed with iodine vapor and quantified by densitometry using Image Master TotalLab 1.11 (Amersham Pharmacia Biotech, UK).

| Lipidomic assay
Yeast cells were inoculated in fresh YPD at a low density (OD 600 nm = 0.3) and collected after 7 or 48 h.Cells were lyophilized and sent frozen to the lipidomics facility at the Biochemistry Institute, São Paulo University, São Paulo, Brazil.Approximately 200 mg of lyophilized cells was resuspended in 350 μL of deionized water.

F
Additivity of lipid-related phenotypes of ACC1 S1157A and deg1 mutants.(a) Scheme depicting the roles of Snf1, Reg1 and soraphen A (sorA) in the regulation of Acc1 activity.See text for details and references.(b) SorA phenotype of wild type (WT), deg1, ACC1 S1157A and the respective double mutant.(c) SorA phenotype of WT, deg1, reg1 and the respective double mutant.(d) SorA phenotype of WT, deg1, snf1 and the respective double mutant.Note that all strains used in this assay contain the ACC1-GFP fusion as the only source of Acc1, which increases sorA sensitivity.(e) Representative fluorescence-microscopic visualization of BODIPY493/503 stained yeast cells with the indicated genotypes.Staining was done with stationary phase cells (stat.) and after refeeding of cells with fresh medium, followed by 6 h growth into exponential phase (exp.).Scale bar = 10 μm.(f) Relative fluorescence of strains shown in (g).Cells were fixed, stained with BODIPY493/503 and examined at magnification 400×.Fluorescence was measured using ImageJ and normalized to wild-type stationary phase levels.The significance was determined utilizing two-tailed t-test (*≤0.05;**≤0.01).(g) Western blot analysis of Acc1-GFP and Cdc19 protein levels of WT and deg1 cells.Signal intensities for the deg1 strain were normalized to WT and are indicated.

F
I G U R E 3 SE and TG levels in WT, deg1 and Acc1 S1157A strains.(a) TLC and (b) Densitometric TLC analysis of steryl ester (SE) and triacylglycerol (TG) from WT, deg1 and ACC1 S1157A strains grown to stationary (Stat) and exponential (Exp) phase.Data are shown as mean (n = 3) ± standard deviation of mean (SD).The significance was determined utilizing two-tailed t-test (*≤0.05;ns, non-significant).

2. 6 |
Global effect of deg1 mutation on the lipidomeSince the investigation of neutral lipid contents in deg1 mutant revealed a major shift in SE abundance, we considered the possibility that additional lipid classes are affected by the loss of the Deg1-dependent tRNA modification.Assessment of 207 lipid species by UPLC-ESI-TOF/MS in the WT and the deg1 mutant in stationary and exponential growth phases indeed revealed significant alterations in the cellular levels of 105 species (TableS1).These altered lipid species are represented by glycerophospholipids, SE, cardiolipin, and particularly multiple TG species.Heatmap analysis (Figure4a,b) using F I G U R E 4 Lipidomic profile of WT and deg1 mutant.(a) Heatmap of distinct clusters of the top 50 significantly changed lipid species in WT and deg1 mutant grown to stationary phase.The heat map was generated using MetaboAnalyst 5.0, and the significance was determined utilizing ANOVA with adjusted p-value (FDR) cutoff: 0.05 and Fisher's post hoc analysis, n = 3 independent experiments.(b) As in (a) but for WT and deg1 cells from exponential phase.(c) Levels of steryl ester (SE) from WT and deg1 cells in stationary (Stat) or exponential (Exp) phase.(d) Levels of triacylglycerol (TG) from WT and deg1 cells in stationary (Stat) or exponential (Exp) phase.(e) Levels of diacylglycerol (DG) from WT and deg1 cells in stationary (Stat) or exponential (Exp) phase.Log2 fold changes are indicated above bar diagrams (c-e).The significance was determined utilizing two-tailed t-test (*≤0.05;ns-non-significant (f) The ratio of C18/C16 acyl chains of neutral lipids in deg1 mutant was normalized to WT, both grown to exponential phase.The difference between deg1 mutant and WT was statistically significant for the neutral lipids shown in the figure (p ≤ 0.05).(c-f) Data are shown as mean (n = 3) ± standard deviation of mean (SD).
).In comparison to the WT at exponential phase, the total content of the phospholipid classes (i.e., phosphatidyl serine, phosphatidyl methylethanolamine, phosphatidylinositol, F I G U R E 5 The deg1 mutant shows lower abundance of phospholipid species in relation to WT. Phospholipid species that show a decrease in relative abundance in deg1 mutant grown to exponential phase: (a) phosphatidylserine (PS), (b) phosphatidyl methylethanolamine (PME), (c) phosphatidylinositol (PI), (d) phosphatidylethanolamine (PE), and (e) lysoPE (LPE).At stationary phase, phospholipid species were not significantly changed in deg1 mutant compared to WT.Data are shown as mean (n = 3) ± standard deviation of mean (SD).
the desired S1157A mutation ( 5′-AGC T GC GTT CT C C AC CTT TC C A AC TGT TA A A TC TA A A A T G GG TAT GA A C AG AGC TG T T GC TGT TT C AGA TTTGTCATATGTTGCAAACAG-3′). The fragment was obtained by hybridization of two complementary oligonucleotides.Colonies were checked for the ACC1 S1157A mutation by PCR amplification and sequencing and phenotypically (sorA resistance).Plate assays for yeast phenotyping were conducted as described previously (Bozaquel-Morais et al., 2018).

A 5
mM stock solution was prepared in 100% DMSO and stored at −20°C until used.To stain cells, 1-5 μL of BODIPY® 493/503 stock solution was added to formaldehyde-fixed cells and mixed carefully.The suspension was incubated in the dark for 10 min and washed.Cells were examined using an Olympus BX53 microscope (GFP channel with 200 ms exposure).Pictures were taken with the CellSens Standard software (Olympus), and fluorescent intensity was analyzed with Fiji(Schindelin et al., 2012).
were vortexed for 30 s and then incubated with MTBE (1 mL) in a Thermomixer (1000 rpm at 20°C) for 1 h.Phase separation was induced by adding water (300 μL), vortexing for 30 s and centrifugation at 1000 g for 10 min.The upper organic phase was collected, dried with nitrogen gas and dissolved in 100 μL isopropanol.The analysis of extracts containing internal standards of different classes of lipids was performed by high-performance liquid chromatography (UPLC Nexera, Shimadzu, Kyoto, Japan) coupled with ESI-TOF/MS mass spectrometry (Triple TOF 6600, Sciex, Concord, USA;Chaves-Filho et al., 2019).The sample injection volume was adjusted to 2 μL, and the compounds were separated on a CORTECS® column (C18, 1.6 μm, 2.1 × 100 mm) with a flow rate of 0.2 mL/min and oven temperature at 35°C.The mobile phase was composed of (A) water/acetonitrile (60:40) and (B) isopropanol/acetonitrile/water (88:10:2).A and B contain ammonium acetate or ammonium formate at a final concentration of 10 mM for experiments performed in negative or positive ionization mode, respectively.The following gradient was used in the chromatography: from 40% to 100% B over the first 10 min; 100% B from 10 to 12 min; and 100% to 40% B for 12-13 min, holding 40% B for 13-20 min.Mass spectrometry was operated in both ionization modes (positive and negative), with the "scan" performed for mass values of 200-2000 Da.MS/MS data acquisition was performed using the Analyst® 1.7.1 software; proceeding to their analysis with the PeakView® software.The lipid molecular species were identified and then quantified using the MultiQuant® software.The precursor ions areas were normalized by the internal standards.The MetaboAnalyst (http:// www.metab oanal yst.ca) was used for statistical analyses of the lipidomic data.Univariate statistical tests included analysis of variance (one-way ANOVA) or volcano plot (t-test and fold-change).Data from univariate analyses were also plotted as heatmaps taking into account the lowest p-values.The complete set of results is found in Table Writing -original draft; writing -review and editing; funding acquisition; data curation; validation; conceptualization.Gabriel Soares Matos: Investigation; data curation; writing -review and editing.Leonie Vogt: Investigation; data curation; writing -review and editing.Rosangela Silva Santos: Investigation; writing -review and editing; data curation.Aurélien Devillars: Investigation; data curation.Marcos Yukio Yoshinaga: Methodology; writing -review and editing; data curation; validation.Sayuri Miyamoto: Writing -review and editing; Detection of Acc1-GFP and Cdc19 involved specific