Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2019 Jun 14;17(6):e3000297.
doi: 10.1371/journal.pbio.3000297. eCollection 2019 Jun.

Cytosine-5 RNA Methylation Links Protein Synthesis to Cell Metabolism

Free PMC article

Cytosine-5 RNA Methylation Links Protein Synthesis to Cell Metabolism

Nikoletta A Gkatza et al. PLoS Biol. .
Free PMC article


Posttranscriptional modifications in transfer RNA (tRNA) are often critical for normal development because they adapt protein synthesis rates to a dynamically changing microenvironment. However, the precise cellular mechanisms linking the extrinsic stimulus to the intrinsic RNA modification pathways remain largely unclear. Here, we identified the cytosine-5 RNA methyltransferase NSUN2 as a sensor for external stress stimuli. Exposure to oxidative stress efficiently repressed NSUN2, causing a reduction of methylation at specific tRNA sites. Using metabolic profiling, we showed that loss of tRNA methylation captured cells in a distinct catabolic state. Mechanistically, loss of NSUN2 altered the biogenesis of tRNA-derived noncoding fragments (tRFs) in response to stress, leading to impaired regulation of protein synthesis. The intracellular accumulation of a specific subset of tRFs correlated with the dynamic repression of global protein synthesis. Finally, NSUN2-driven RNA methylation was functionally required to adapt cell cycle progression to the early stress response. In summary, we revealed that changes in tRNA methylation profiles were sufficient to specify cellular metabolic states and efficiently adapt protein synthesis rates to cell stress.

Conflict of interest statement

I have read the journal's policy and the authors of this manuscript have the following competing interests: MF is consulting for Storm Therapeutics.


Fig 1
Fig 1. Loss of NSUN2 triggers a shift of the metabolic state towards catabolism.
(A, B) Detection of Nsun2 RNA in Nsun2+/+ and Nsun2−/ mouse skin in early (A) and late (B) anagen. Scale bar: 50 μm. (C-E) Transcriptional changes in skin of wild-type (Nsun2+/+) and Nsun2 knockout (Nsun2−/) mice. Highlighted in red are significant FC expression differences (FDR < 0.05) in hair follicle stem cells (CD34+/ITGA6high) (C), progenitor cells (PCADhigh/ITGA6low) (D), and anagen skin (E). (n = 3–4 mice per genotype and condition). (F-H) Multivariate analyses of data obtained from MS (F) or NMR spectroscopy–based metabolic profiling (G) using mouse back skin (n = 3–5 mice) or human dermal fibroblasts (n = 5 samples per genotype) (H). Model parameters: R2X = 94.5%, R2Y = 99.9%, and Q2 = 95.8% using partial least square discriminant analysis (F), R2X = 70% and Q2 = 30% (G), and R2X = 85.9% and Q2 = 78.6% (H), using principal component analysis. (I-K) Metabolic differences between NSUN2+/− and NSUN2−/− normalised to NSUN2+/+ human dermal fibroblasts relating to the methionine cycle (I), free amino acids (J), and the TCA cycle (K). The underlying data for this figure can be found in S1–S3 Data and S1 File. BG, hair follicle bulge; FC, fold-change; FDR, false discovery rate; HB, hair bulb; IFE, interfollicular epidermis; ITGA6, integrin alpha-6; MS, mass spectrometry; NMR, nuclear magnetic resonance; PC1, Principal Component 1; PC2, Principal Component 2; PCAD, P-cadherin; TCA, tricarboxylic acid.
Fig 2
Fig 2. Methylation-dependent and -independent functions of NSUN2.
(A) Schematic representation of NSUN2-methylated tRNA sites in the anticodon loop (C34) and the VL (C46, C47). (B, C) Number of m5C per tRNA in all tRNAs (B) or tRNA leucine (C) quantified by mass spectrometry in NSUN2−/ cells reexpressing NSUN2, the enzymatic dead version of NSUN2 (K190M), or the empty (‘e.’) vector control. *padj < 0.05; ****padj < 0.0001 (ordinary one-way ANOVA, multiple comparisons). (D) Quantification of m5C levels in all rescued tRNAs (padj < 0.05) using RNA bisulfite sequencing. (E) Heatmaps of example tRNAs showing the rescued m5C sites in five replicates of NSUN2+/+ cells or NSUN2−/− cells reexpressing NSUN2, K190M, or the empty vector (‘e.v.’). (F) PCAs of tRFs differentially abundant in NSUN2-overexpressing NSUN2−/ cells. (G, H) Log2 coverage of tRFs smaller than 46 nucleotides (G) or larger than 46 nucleotides (H). (I) Polysome profile of NSUN2-expressing (NSUN2+/+) and -lacking (NSUN2−/) cells. Shown is one out three replicates. (J, K) Protein synthesis levels measured by flow cytometry using OP-puro in the indicated cells. CHX served as a control. Data represent mean, and error bars are ±SD. Student’s t test. *p < 0.05, **p < 0.01, ****p < 0.0001. (L-N) Metabolic differences between NSUN2−/ cells overexpressing the NSUN2 or K190 protein normalised (‘norm.’) to NSUN2−/ cell infected with the empty vector control (‘e.V.’) relating to the methionine cycle (L), free amino acids (M), and the TCA cycle (N). The underlying data for this figure can be found in S5–S7 Data and S1 File. ADMA, asymmetric dimethylarginine; CHX, cycloheximide; FC, fold-change; m5C, 5-methylcytosine; OP-puro, O-propargyl-puromycin; PCA, principle component analysis; SAH, S-adenosyl-homocysteine; SAM, S-adenosyl-methionine; SDMA, symmetric dimethylarginine; TCA, tricarboxylic acid; tRF, tRNA-derived fragment; tRNA, transfer RNA; VL, variable loop.
Fig 3
Fig 3. NSUN2 functions in the cell cycle to adapt dynamic protein synthesis in response to stress.
(A) Schematic representation how oxidative stress modulates global and gene-specific translation. (B) Treatment regime using arsenite to induce stress and OP-puro to measure protein synthesis. (C) Log2 FC of protein synthesis in NSUN2+/+, NSUN2+/− and NSUN2−/− cells in response to stress compared to the untreated controls (‘Ctr’). CHX served as a control. (n = 2–3 samples per time point). (D) Relative protein synthesis levels in response to stress in NSUN2+/+ and −/− cells measured as FC compared to CHX control. (E) Log2 FC of protein synthesis in NSUN2−/− cells rescued with NSUN2 or the enzymatic dead version K190M after exposure to stress at the indicated time points. (F-I) Polysome profile of NSUN2+/+ (F) and NSUN2−/− cells rescued with wt (I) or mutated NSUN2 (K190M) (H). The empty vector (‘e.V.’)-infected cells served as control (G). Shown is one out of two replicates. (J) Gating used for cell cycle analyses using DAPI incorporation. (K-M) Percentage of NSUN2+/+ (grey) and NSUN2−/− (red) cells in G0/G1- (K), S- (L), and G2/M- (M) phases of the cell cycle after treatment with sodium arsenite for the indicated time. (n = 3 samples per time point). Data presented as mean, error bars ± SD. p-Value: two-way ANOVA calculating row (grey; treatment) and column (black; genotype) factor variation. The underlying data for this figure can be found in S1 File. CHX, cycloheximide; eIF2, eukaryotic initiation factor 2; FC, fold-change; Norm. abs., normalised absorbance; OP-puro, O-propargyl-puromycin; wt, wild-type.
Fig 4
Fig 4. Levels of m5C changes site-specifically and dynamically in response to oxidative stress.
(A) Time course of sodium arsenite treatment. (B) Log2 FC of Nsun2 RNA expression in NSUN2+/+ and NSUN2+/− cells relative to GAPDH and normalised to the untreated control (‘Ctr’). Shown are 3 replicates. (C) Western blot analysis of the indicated proteins using whole cell lysates from NSUN2+/+ and NSUN2−/− cells. Hsp90 served as a loading control. (D,E) Detection of m5C in sodium arsenite–treated and untreated (‘ctr’) NSUN2+/+ and −/− cells using mass spectrometry. (n = 3 samples per time point). (F) Quantification of tRNA methylation percentage using RNA bisulfite sequencing of NSUN2+/+ and NSUN2−/− cells (n = 4 samples per time point). (G) Heatmap of methylation status of individual tRNA molecules shown in (F). (H,I) Quantification (H) and heatmap (I) of methylation changes in the tRNAs LeuCAA and AspGTC in NSUN2+/+ and NSUN2−/− cells. (J) Quantification of methylation in non-tRNA targets. Data represent median in F, H, and J. Error bars are ±SD. p-Values: Student’s t test, *p < 0.05 and **p < 0.01. ***p < 0.001. The underlying data for this figure can be found in S8 and S9 Data and S1 File. FC, fold-change; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; HSP90, heat shock protein 90; m5C, 5-methylcytosine; NPMI, nucleophosmin; tRNA, transfer RNA; VL, variable loop.
Fig 5
Fig 5. NSUN2-mediated tRNA methylation is dynamic and site-specific.
(A, B) Volcano plot depicting the significant methylation changes when NSUN2 was reexpressed in NSUN2−/− cells compared to empty (‘E.’) vector (A) or K190M controls (B). (C) Global methylation levels of all m5C sites identified in (B) after treatment with arsenite for 0, 2, or 4 hours. Shown are all sites >20% methylation in NSUN2-rescued cells. (D, E) Mass spectrometry analyses to quantify the number of methylated sites (m5C) in all tRNA (D) or only tRNA LeuCAA (E) in response to stress. (F) Heatmap showing all significantly different m5C sites (p < 0.05) changing upon stress in NSUN2-overexpressing cells. (G-L) Examples of m5C sites identified in (F). *padj < 0.05; **padj < 0.01; ****padj < 0.0001 (ordinary one-way ANOVA, multiple comparisons). (M) Mass spectrometry analyses to quantify m5C in tRNA leucine (upper panel) and all tRNAs (lower panel) in the presence of an angiogenin inhibitor (‘Angi’) and arsenite. Data presented as mean (n = 3), error bars ± SD. p-Value: padj ANOVA. (N-P) Methylation levels (pooled from 5 replicates) of cytosines along tRNA 74-Leu CAA (upper panels) and 145-Leu CAA (lower panels) detecting all m5C sites within the tRNA molecule with different dynamic changes in response to stress. The underlying data for this figure can be found in S10 Data and S1 File. m5C, 5-methylcytosine; tRNA, transfer RNA; VL, variable loop.
Fig 6
Fig 6. Site-specific tRNA methylation determines biogenesis of distinct tRFs.
(A) Density plot of tRNA-derived sequences beginning at the indicated positions. (B) Clustering of tRNA-derived fragments < 40 nucleotides in NSUN2+/+ and NSUN2−/ cells untreated (‘Ctr’) or treated with sodium arsenite for 2 or 4 hours. Shown are 3 out of 4 replicates per time point. (C, D) Heatmap (C) and log2 FC of tRFs shown in (C). (E, F) PCAs (E) and heatmap (F) of significantly different tRFs in NSUN2−/ cells rescued with a wild-type (NSUN2) or point mutated (K190M) NSUN2 construct after 4 hours of exposure to stress compared to the untreated control (‘0h’). (G, H) Violin plots showing the read distribution of the tRFs shown in (E, F). (I) Log2 FC of the up-regulated tRFs when NSUN2-overexpressing cells are exposed to stress for 4 hours. tRNA glutamic acid–derived tRFs are highlighted in red. Line indicates the mean. The underlying data for this figure can be found in S11 and S12 Data and S1 File. FC, fold-change; PCA, principle component analysis; tRF, tRNA-derived fragment; tRNA, transfer RNA.
Fig 7
Fig 7. Loss of NSUN2 alters mitochondrial function and catabolic pathways in response to stress.
(A) Venn diagram showing all significantly expressed genes in NSUN2/ cells compared to NSUN2+/+ when untreated (‘Ctr’) or treated for 2 and 4 hours with sodium arsenite. (B, C) Gene enrichment analysis for biological processes (GOrilla) using the 884 uniquely changed genes in NSUN2/ cells after 2 hours (B) or the 1,584 uniquely changed genes after 4 hours (C) of stress exposure. Colour code indicates p-value, and size reflects enrichment. The underlying data for this figure can be found in S13 and S14 Data.

Similar articles

See all similar articles

Cited by 1 article


    1. Van Heyningen V, Yeyati PL. Mechanisms of non-Mendelian inheritance in genetic disease. Hum Mol Genet. 2004;13 Spec No 2:R225–33. 10.1093/hmg/ddh254 . - DOI - PubMed
    1. Proudfoot NJ, Furger A, Dye MJ. Integrating mRNA processing with transcription. Cell. 2002;108(4):501–12. . - PubMed
    1. Frye M, Harada BT, Behm M, He C. RNA modifications modulate gene expression during development. Science. 2018;361(6409):1346–9. Epub 2018/09/29. 10.1126/science.aau1646 . - DOI - PMC - PubMed
    1. Boccaletto P, Machnicka MA, Purta E, Piatkowski P, Baginski B, Wirecki TK, et al. MODOMICS: a database of RNA modification pathways. 2017 update. Nucleic Acids Res. 2018;46(D1):D303–D7. 10.1093/nar/gkx1030 - DOI - PMC - PubMed
    1. Frye M, Blanco S. Post-transcriptional modifications in development and stem cells. Development. 2016;143(21):3871–81. 10.1242/dev.136556 . - DOI - PubMed

Publication types