Translational control of one-carbon metabolism underpins ribosomal protein phenotypes in cell division and longevity

  1. Nairita Maitra
  2. Chong He
  3. Heidi M Blank
  4. Mitsuhiro Tsuchiya
  5. Birgit Schilling
  6. Matt Kaeberlein
  7. Rodolfo Aramayo
  8. Brian K Kennedy  Is a corresponding author
  9. Michael Polymenis  Is a corresponding author
  1. Department of Biochemistry and Biophysics, Texas A&M University, United States
  2. Buck Institute for Research on Aging, United States
  3. Department of Pathology, University of Washington, United States
  4. Department of Biology, Texas A&M University, United States
  5. Departments of Biochemistry and Physiology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore
  6. Centre for Healthy Ageing, National University of Singapore, National University Health System, Singapore
8 figures, 1 table and 1 additional file

Figures

Doubling time and normal Rpl levels are not associated with the replicative lifespan of single rpl deletion mutants.

(A) Scatterplot between the generation time (x-axis; from Tables S2 in Steffen et al., 2012) and replicative lifespan (y-axis; as percentage of the wild type lifespan, from Table 3 in Steffen et al., 2012 and Table S2 in McCormick et al., 2015). (B) Scatterplot between the abundance of each deleted Rpl typically found in wild type cells (x-axis; the median number of molecules per cell, from Table S4 from Ho et al., 2018) and replicative lifespan (y-axis; as percentage of the wild type lifespan, from Table 3 in Steffen et al., 2012 and Table S2 in McCormick et al., 2015). The Spearman correlation coefficients (ρ) shown in each case were calculated with the rcorr function of the Hmisc R language package, comparing the pair shown in each panel. All the values used as input for this figure and analyses are in Figure 1—source data 1.

Figure 1—source data 1

Lifespan, doubling time, and RP levels in ribosomal protein mutants.

https://cdn.elifesciences.org/articles/53127/elife-53127-fig1-data1-v2.xlsx
Figure 2 with 1 supplement
Loss of Rpl22 does not affect the relative abundance of other ribosomal proteins in ribosomes.

(A) Schematic overview of the approach to query ribosomal protein abundances in rpl22 deletion mutants (see Materials and methods). All strains were in the haploid BY4742 background. The doubling time values for each strain are shown on top (Td; in min, from Table S2 in Steffen et al., 2012). (B) Bar plot displaying on the x-axis the levels of Rpl22Ap and Rpl22Bp in wild type and rpl22 deletion mutants, as indicated on the y-axis. The abundances shown correspond to the peak area intensities obtained from SWATH-MS (see Materials and methods and Figure 2—source data 1). (C) The Log2-transformed ratios of the intensities corresponding to the ribosomal proteins detected are indicated for each pairwise comparison among the 4 strains we analyzed. Changes higher than 2-fold are outside the dashed red lines in each panel. Note that not all the changes were statistically significant, as indicated. The red arrows indicate the only cases where the differences were significant (Log2FC ≥ 1 and p-value<0.05; bootstrap-based ANOVA; see Materials and methods) The Spearman correlation coefficients (ρ) shown in each case were calculated with the rcorr function of the Hmisc R language package, comparing the pair shown in each panel.

Figure 2—source data 1

SWATH-mass spectrometry measurements of ribosomal protein abundances.

https://cdn.elifesciences.org/articles/53127/elife-53127-fig2-data1-v2.xlsx
Figure 2—figure supplement 1
Reduced protein synthesis in cells lacking Rpl22Ap.

(A) Schematic of the approach to measure incorporation of a methionine analog (HPG) in newly synthesized proteins (see Materials and methods). (B) The strains shown on the x-axis (all in the BY4742 background, the same as in Figure 1) were incubated with HPG for 30 m, processed for CLICK reaction with the Alexa dye, and then the fluorescence per cell was quantified with flow cytometry. At least 8,000–10,000 cells were quantified in each sample, from four independent experiments in each case. On the y-axis is the mean fluorescence per cell for each sample divided by the average of the four WT samples. Significant differences in pair-wise comparisons were indicated by the non-parametric Kruskal-Wallis test (p=0.004848), while the p-values shown were based on the posthoc Nemenyi test (performed with the PMCMR R language package). (C) Representative images of HPG-Alexa-labeled cells, which were also stained with DAPI to visualize the nuclei.

Querying synchronous, dividing cells lacking ribosomal protein paralogs.

(A) Schematic of our experimental approach to identify gene expression changes in rpl22 and rpl34 paralog mutants at the transcriptional or translational level, from cells at several different stages of the cell cycle. (B) Cell size (in fL, x-axis) and the percentage of budded cells (%Budded, y-axis) from each pool of the indicated strains. The values shown are the weighted averages, from the different elutriated samples in each pool. (C) On the y-axis, are the Log2-transformed TPM values from representative transcripts known to increase in abundance in S (histone H4; HHF1), G2 phase (cyclin CLB2), or are constitutively expressed (ACT1). Cell size is on the x-axis.

Transcripts with altered relative abundance in rpl22 and rpl34 paralog deletion mutants.

(A) Decision diagram for identifying transcripts that were differentially expressed in paralog deletion mutants. (B) The number of transcripts with significantly different levels (adjusted p-value/FDR < 0.05, Log2FC ≥ 2, identified both from the babel and DESeq2 R language packages) at any one cell size pool, between any pairwise comparison between WT, rpl22, and rpl34 mutants is shown. The data for wild type diploid cells were from Blank et al., 2017. The Gene Ontology terms (The Gene Ontology Consortium, 2019) that were most enriched in each case are shown, based on the PANTHER (Mi et al., 2019) platform classification, incorporating the Holm-Bonferroni correction. All the gene names of loci with significant changes in transcript levels are in Figure 4—source data 1 (sheet ‘cumulative’). The ‘UP’ and ‘DOWN’ groupings correspond to each pairwise comparison shown, with the ‘UP’, or ‘DOWN’, group higher, or lower, for the strain in the numerator of the ratio, respectively. (C) Reduced levels of transcripts encoding glycolytic/gluconeogenic enzymes in rpl22aΔ compared to rpl22bΔ cells. The Log2-transformed ratio of the corresponding TPM values is shown in each case, across the different cell size pools. The data were hierarchically clustered without supervision and displayed with the pheatmap R package. Each row corresponds to a separate 5 fL cell size interval, from 40 to 75 fL, from top to bottom. The arrows indicate transcripts for key enzymes of glycolysis and gluconeogenesis. mRNAs with missing values across the cell sizes we analyzed were not included in the heatmap.

Figure 4—source data 1

mRNAs with significantly different abundances between any two strains.

https://cdn.elifesciences.org/articles/53127/elife-53127-fig4-data1-v2.xlsx
Figure 5 with 4 supplements
Reduced translational efficiency of transcripts encoding enzymes of the methionine and 1C metabolic pathways in rpl22aΔ cells.

(A) Heatmap of the transcripts with significantly different translational efficiencies (see Figure 3) between the two Rpl22 paralog deletion strains. The ratio of the TPM values for the ribosome footprints ((ribo) against the corresponding values for the mRNA reads (rna) define the translational efficiency of each locus in each paralog mutant. The data were hierarchically clustered and displayed with the pheatmap R package. Each row corresponds to a separate cell size interval, from 40 to 75 fL, from top to bottom. The Gene Ontology terms highlighted were enriched, based on the PANTHER platform classification, incorporating the Holm-Bonferroni correction. (B) Diagram of the metabolic pathways involving enzymes whose mRNAs have altered translational efficiency in rpl22aΔ vs rpl22bΔ cells. Proteins whose mRNAs have lower translational efficiency in rpl22aΔ cells are shown in blue. (C) Immunoblots from WT (RPL22+), rpl22aΔ, or rpl22bΔ cells, carrying a MET3-TAP allele expressed from its endogenous chromosomal location. All the strains were otherwise isogenic. The signal corresponding to Met3p-TAP was detected with the PAP reagent, while the Pgk1p signal represents loading. The band intensities were quantified using the ImageJ software package, and the relative abundance of Met3p-TAP in rpl22aΔ: rpl22bΔ cells is shown at the bottom, from three independent such experiments.

Figure 5—source data 1

mRNAs with significantly different translational efficiency between any two strains.

https://cdn.elifesciences.org/articles/53127/elife-53127-fig5-data1-v2.xlsx
Figure 5—figure supplement 1
Transcripts with altered translational efficiency (TE) in rpl22 and rpl34 paralog deletion mutants.

(A) Decision diagram for identifying transcripts that had altered translational efficiency in paralog deletion mutants. (B) The number of transcripts with significantly different levels (adjusted p-value/FDR < 0.05, Log2FC ≥ 1, identified from all three statistical packages) at any one cell size pool, between any pairwise comparison between WT, rpl22, and rpl34 mutants is shown. The data for wild type diploid cells were from Blank et al., 2017. The Gene Ontology terms that were most enriched in each case are shown, based on the PANTHER platform classification, incorporating the Holm-Bonferroni correction. All the gene names of loci with significant changes in translational efficiency are also in Figure 5—source data 1 (sheet ‘cumulative’). The ‘UP’ and ‘DOWN’ groupings correspond to each pairwise comparison shown, with the ‘UP’, or ‘DOWN’, group higher, or lower, for the strain in the numerator of the ratio, respectively.

Figure 5—figure supplement 2
The translational efficiency of GCN4 is de-repressed in cells lacking Rpl22Ap.

The Log2-tranformed ratio of the TPM values for the ribosome footprints ((ribo) against the corresponding values for the mRNA reads (rna) define the translational efficiency of GCN4 (y-axis), for each of the cell sizes we queried (x-axis), in the strains shown. Asterisks indicate that the translational efficiency was significantly reduced (FDR < 0.05) compared to other mRNAs in that strain (based on the babel ‘combined’ output).

Figure 5—figure supplement 3
Dysregulation of the translational efficiency of transcripts encoding gene products involved in cytoplasmic translation in cells lacking Rpl22 or Rpl34 paralogs.

Heatmaps of the transcripts with significantly different translational efficiencies (see Figure 2) between wild type diploid cells (from Blank et al., 2017) and rpl22aΔ/rpl22aΔ (A), rpl22bΔ/rpl22bΔ (B), rpl34aΔ/rpl34aΔ (C), rpl34bΔ/rpl34bΔ (D) cells. The ratio of the TPM values for the ribosome footprints ((ribo) against the corresponding values for the mRNA reads (rna) define the translational efficiency of each locus in each strain. The data were hierarchically clustered and displayed with the pheatmap R package. Each column corresponds to a separate cell size interval, from 40 to 75 fL, from left to right. The Gene Ontology terms highlighted were enriched, based on the PANTHER platform classification, incorporating the Holm-Bonferroni correction. mRNAs with missing values across the cell sizes we analyzed were not included in the heatmaps. All the gene names of loci with significant changes in translational efficiency are also in Figure 5—source data 1 (sheet ‘cumulative’).

Figure 5—figure supplement 4
Features of mRNAs with altered translational efficiency in rpl mutants.

Top, All the mRNAs shown in Figure 5—figure supplement 3 that had altered translational efficiency between wild type and any of the rpl deletion strains were placed in the following groups: UP, if their median of their log2[(ribo/rna)WT/(ribo/rna) rplΔ] values from all cell cycle points was positive; DOWN, if their median of their log2[(ribo/rna)WT/(rio/rna)rplΔ] values from all cell cycle points was negative; OTHER if they were not identified as being under translational control between wild type and any of the rpl deletion strains. The values for ORF, 5’-leader (5’-UTR), and 3’-UTR length for each of the mRNAs were from Lin and Li, 2012. The protein abundance values were the median number of molecules per cell, from the studies analyzed in Ho et al., 2018. The violin plots were generated with the lattice R language package. Significant differences in pair-wise comparisons were indicated by the non-parametric Kruskal-Wallis test, while the p-values shown were for all significant differences based on the posthoc Nemenyi test (performed with the PMCMR R language package). Bottom, The same analysis as in the top panels for the mRNAs with different translational efficiency between the rpl22aΔ and rpl22bΔ mutants, based on their log2[(ribo/rna)rpl22aΔ/(ribo/rna)rpl22bΔ] values.

Figure 6 with 2 supplements
Metabolic profiling indicates reduced flux through central metabolic pathways and the folate cycle in rpl22aΔ cells.

(A) Eleven metabolites shown at the bottom had significantly reduced levels in rpl22aΔ cells (Log2FC ≥ 1, p<0.05; based on bootstrapped ANOVA; see Materials and methods) and they were significantly enriched for the metabolic pathways shown to the right (FDR < 0.05). Pathway enrichment analysis was done with the MetaboAnalyst R language package. The metabolites were identified with untargeted, MS-based profiling of primary metabolites and biogenic amines, and targeted amino acid analysis. Metabolites indicated with gray in the Table are part of the pathways shown to the right. (B) The Log2-transformed peak intensities from the MS-based profiling of the metabolites shown in A (except Glycine) are on the y-axis. The strains used in the analysis are on the x-axis. (C) The Log2-transformed levels (in nmoles) of amino acids, after PTH-derivatization, Edman degradation and HPLC detection, are shown on the y-axis. The red arrows indicate the only amino acids (Gly and Trp) whose levels were significantly lower in rpl22aΔ cells (Log2FC ≥ 1, p<0.05; based on bootstrapped ANOVA; see Materials and Methods). The strains used in the analysis are on the x-axis, and they were in the BY4742 background.

Figure 6—figure supplement 1
Exogenous addition of metabolites from pathways affected in rpl22aΔ cells does not suppress the slower growth of these cells.

Cultures from the indicated strains shown to the left (in the diploid BY4743 background in A and B; in the haploid BY4742 background in C), were spotted onto solid plates with standard undefined media (YPD; see Materials and Methods. In each case, aliquots from saturated cultures were spotted in 10-fold serial dilutions. The plates contained the compounds shown on top in each case (GSH is reduced glutathione and GSSG is oxidized glutathione), at the indicated concentrations. The plates were photographed after they were incubated for 3 days at 30°C.

Figure 6—figure supplement 2
Nobiletin is accumulated in rpl22aΔ cells.

(A) Structure of nobiletin. (B) Peak intensities assigned to nobiletin from mass spectrometry are on the y-axis. The strains used in the analysis are on the x-axis, and they were in the BY4742 background.

Figure 7 with 1 supplement
Deletion of enzymes of one-carbon metabolic pathways extends replicative lifespan in yeast.

Survival curves for MATα (BY4742) cells (shown in black), compared to experiment-matched cells (shown in red) lacking SHM1 (A), SHM2 (B), ADE17 (C), ADE2 (D), or ADE3 (E). Mean lifespans are shown in parentheses, along with the number of cells assayed in each case. In the case of shm1Δ, shm2Δ, ade17Δ, and ade3Δ cells, the lifespan extension was significant (p<0.0001; based on the log-rank test).

Figure 7—figure supplement 1
Deletion of enzymes of one-carbon metabolic enzymes does not further extend the replicative longevity of rpl22aΔ cells.

Survival curves for wild type MATα (BW885) cells (shown in black), compared to experiment-matched cells of the indicated genotype. Mean lifespans are shown in parentheses. The lifespans were determined from 60 cells in each case.

Figure 8 with 1 supplement
Loss of Shm2p and Ade3p impinges on multiple cell cycle phases but in distinct ways.

(A) Double shm2Δ,ade3Δ deletion mutants are slow-growing. Two representative tetrad dissections from shm2Δ x ade3Δ crosses are shown. The yellow diamond indicates the shm2Δ,ade3Δ segregants. All the strains used in B-E were segregants from the same shm2Δ x ade3Δ crosses and, except as indicated, isogenic otherwise. (B) Violin plots of the mean and birth size of the indicated strains, calculated from ≥12 asynchronous cultures in each case. The plots were generated with the lattice R language package. Significant differences in pair-wise comparisons were indicated by the non-parametric Kruskal-Wallis test, while the p-values shown were for all significant differences based on the posthoc Nemenyi test (performed with the PMCMR R language package). (C) Representative DNA content histograms from the indicated strains, from at least 10,000 cells and ≥5 independent asynchronous cultures in each case. On the x-axis is fluorescence per cell, while the cell number is on the y-axis. The average and sd of the percentage of cells with unreplicated DNA (%G1) is shown. There were no statistically significant differences among the strains, based on the non-parametric Kruskal-Wallis test. (D) The rate of cell size increase (k, in h−1, shown on the y-axis) was calculated as described in Soma et al., 2014, for the strains shown on the x-axis. Each data point is the average of two technical replicates, from synchronous elutriated cultures. (E) The critical size (y-axis) for the strains shown on the x-axis was calculated as described in Soma et al., 2014, from the same cultures as in D.

Figure 8—figure supplement 1
The translational efficiency of SHM2 is cell cycle-regulated.

Cell size (in fL, x-axis) and the translational efficiency (TE) of SHM2, shown as Log2(expressed ratios) across the cell size series (y-axis), in the indicated strains. The values shown are the average from the three independent pools, for each 5 fL interval across.

Tables

Key resources table
Reagent type
(species) or
resource
DesignationSource or
reference
IdentifiersAdditional
information
Strain, strain background (S. cerevisiae)BY4743Giaever et al., 2002RRID:SCR_003093MATa/α his3Δ1/his3Δ1 leu2Δ0/leu2Δ0 LYS2/lys2Δ0 met15Δ0/MET15 ura3Δ0/ura3Δ0
Strain, strain background (S. cerevisiae)32672Giaever et al., 2002RRID:SCR_003093rpl22aΔ::KanMX/rpl22aΔ::KanMX, BY4743 otherwise
Strain, strain background (S. cerevisiae)35844Giaever et al., 2002RRID:SCR_003093rpl22bΔ::KanMX/rpl22bΔ::KanMX, BY4743 otherwise
Strain, strain background (S. cerevisiae)30192Giaever et al., 2002RRID:SCR_003093rpl34aΔ::KanMX/rpl34aΔ::KanMX, BY4743 otherwise
Strain, strain background (S. cerevisiae)31445Giaever et al., 2002RRID:SCR_003093rpl34bΔ::KanMX/rpl34bΔ::KanMX, BY4743 otherwise
Strain, strain background (S. cerevisiae)BY4742Steffen et al., 2012MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0
Strain, strain background (S. cerevisiae)KS976Steffen et al., 2012rpl22aΔ::URA3, BY4742 otherwise
Strain, strain background (S. cerevisiae)KS979Steffen et al., 2012rpl22bΔ::URA3, BY4742 otherwise
Strain, strain background (S. cerevisiae)KS999Steffen et al., 2012rpl22a,bΔ::URA3, BY4742 otherwise
Strain, strain background (S. cerevisiae)BY4741Giaever et al., 2002RRID:SCR_003093MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0
Strain, strain background (S. cerevisiae)MET3-TAPDharmaconYSC1178-202231887MET3-TAP::HIS3M × 6, BY4741 otherwise
Strain, strain background (S. cerevisiae)HB147This studyrpl22aΔ::URA3, MET3-TAP::HIS3M × 6, BY4741 otherwise
Strain, strain background (S. cerevisiae)HB171This studyrpl22bΔ::URA3, MET3-TAP::HIS3M × 6, BY4741 otherwise
Strain, strain background (S. cerevisiae)13403Giaever et al., 2002RRID:SCR_003093shm1Δ::KanMX, BY4742 otherwise
Strain, strain background (S. cerevisiae)12669Giaever et al., 2002RRID:SCR_003093shm2Δ::KanMX, BY4742 otherwise
Strain, strain background (S. cerevisiae)16561Giaever et al., 2002RRID:SCR_003093ade17Δ::KanMX, BY4742 otherwise
Strain, strain background (S. cerevisiae)12384Giaever et al., 2002RRID:SCR_003093ade2Δ::KanMX, BY4742 otherwise
Strain, strain background (S. cerevisiae)6591Giaever et al., 2002RRID:SCR_003093ade3Δ::KanMX, BY4741 otherwise
Strain, strain background (S. cerevisiae)NM64This studyMATα, shm2Δ::KanMX, ade3Δ::KanMX, met -, lys-
Strain, strain background (S. cerevisiae)NM65This studyMATa, shm2Δ::KanMX, ade3Δ::KanMX, met -, lys-
Strain, strain background (S. cerevisiae)NM66This studyMATα, ade17Δ::KanMX, rpl22aΔ::URA, his-, lys-, leu-
Strain, strain background (S. cerevisiae)NM67This studyMATα, shm2Δ::KanMX, rpl22aΔ::URA, his-, leu-, met-
Strain, strain background (S. cerevisiae)NM68This studyMATα, shm1Δ::KanMX, rpl22aΔ::URA, his-, leu-, met-
Strain, strain background (S. cerevisiae)BW885This studyMATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0
OtherYeast extractSigma-AldrichY1625
OtherPeptoneSigma-AldrichP5905
Chemical compound, drugDextroseSigma-AldrichD9434
Chemical compound, drugCycloheximideCalbiochem239763 M
Chemical compound, drugSodium azideSigma-AldrichS2002
Chemical compound, drugTris(hydroxymethyl)aminomethaneSigma-Aldrich252859
Chemical compound, drugTris baseRocheTRIS-RO
Chemical compound, drugSodium chlorideSigma-AldrichS7653
Chemical compound, drugMagnesium chloride hexahydrateUSP1374248
Chemical compound, drugDTTSigma-AldrichD0632
Chemical compound, drugTriton X-100Sigma-AldrichT8787
Peptide, recombinant proteinTurbo DNase IThermoFisherAM2238
OtherGlass beadsScientific IndustriesSI-BG05
Other13 × 51 mm polycarbonate centrifuge tubesBeckman Coulter349622
Chemical compound, drugSucroseSigma-AldrichS0389
Chemical compound, drugPhosphate buffered saline (PBS)Sigma-AldrichP4417
Commercial assay or kitClick-iT HPG Alexa Fluor 488 Protein Synthesis Assay KitThermoFisherC10428
Chemical compound, drugDAPI (4',6-Diamidino-2-Phenylindole, Dihydrochloride)ThermoFisherD1306
Commercial assay or kitRibo-Zero Magnetic Gold Kit (Yeast)EpicentreMRZY1324
Commercial assay or kitSciptSeq v2 RNA-Seq Library Preparation KitEpicentreSSV21124
AntibodyPeroxidase Anti-Peroxidase (PAP) Soluble ComplexSigma-AldrichP1291(1:1000)
AntibodyAnti-Pgk1p antibody, rabbit polyclonalabcamab38007(1:1000)
OtherNovexWedgeWell4–12% Tris-Glycine gelsThermoFisherXP04125
Software, algorithmMetaboAnalysthttps://www.metaboanalyst.ca/RRID:SCR_015539Web server for statistical, functional and integrative analysis of metabolomics data
Software, algorithmAccuComp Z2Beckman Coulter383550Software to monitor number and size of cells with Z2 cell counter
Software, algorithmNIS-Elementshttps://www.nikoninstruments.com/Products/SoftwareRRID:SCR_014329Microscope imaging software suite used with Nikon products
Software, algorithmImageJhttps://imagej.net/RRID:SCR_003070Image processing software
Software, algorithmAdobe Photoshophttps://www.adobe.com/products/photoshop.htmlRRID:SCR_014199Image processing software
Software, algorithmRStudiohttp://www.rstudio.com/RRID:SCR_000432Software for the R statistical computing environment
Software, algorithmSGDhttp://www.yeastgenome.org/RRID:SCR_004694Saccharomyces Genome Database
Software, algorithmRhttps://www.r-project.orgv3.5.2 RRID:SCR_001905Statistical Computing Environment
Software, algorithmPANTHERhttp://www.geneontology.org/RRID:SCR_002811Gene ontology enrichment analysis

Additional files

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)

  1. Nairita Maitra
  2. Chong He
  3. Heidi M Blank
  4. Mitsuhiro Tsuchiya
  5. Birgit Schilling
  6. Matt Kaeberlein
  7. Rodolfo Aramayo
  8. Brian K Kennedy
  9. Michael Polymenis
(2020)
Translational control of one-carbon metabolism underpins ribosomal protein phenotypes in cell division and longevity
eLife 9:e53127.
https://doi.org/10.7554/eLife.53127