A stress-induced tyrosine-tRNA depletion response mediates codon-based translational repression and growth suppression

Abstract

Eukaryotic transfer RNAs can become selectively fragmented upon various stresses, generating tRNA-derived small RNA fragments. Such fragmentation has been reported to impact a small fraction of the tRNA pool and thus presumed to not directly impact translation. We report that oxidative stress can rapidly generate tyrosine-tRNA_GUA fragments in human cells-causing significant depletion of the precursor tRNA. Tyrosine-tRNA_GUA depletion impaired translation of growth and metabolic genes enriched in cognate tyrosine codons. Depletion of tyrosine tRNA_GUA or its translationally regulated targets USP3 and SCD repressed proliferation-revealing a dedicated tRNA-regulated growth-suppressive pathway for oxidative stress response. Tyrosine fragments are generated in a DIS3L2 exoribonuclease-dependent manner and inhibit hnRNPA1-mediated transcript destabilization. Moreover, tyrosine fragmentation is conserved in C. elegans. Thus, tRNA fragmentation can coordinately generate trans-acting small RNAs and functionally deplete a tRNA. Our findings reveal the existence of an underlying adaptive codon-based regulatory response inherent to the genetic code.

Keywords: hnRNPA1; oxidative stress; tRNA; tRNA fragments; translation.

Figure 1. Identification of modulated mammalian tRNAs and tRFs in response to oxidative stress

1. Heatmap of tRNA profiling of MCF10A cells at 8 and 24 h post‐exposure to oxidative stress (200 μM H₂O₂). Biological triplicate data is depicted at each time point relative to control cells.

2. MCF10A cells were exposed to oxidative stress (200 μM H₂O₂) and processed for small RNA‐sequencing. The log₂‐fold induction levels for tRFs derived from distinct tRNAs are plotted.

3. Schematic depicts the overlap of tRNAs in (A) that decreased over time with tRFs from (B) that were induced.

Figure EV1. Fragmentation of tRNA^Tyr _GUA in response to oxidative stress is not due to cell death

1. MCF10A cell viability 1 h after H₂O₂ treatment (200 μM) was tested by a trypan blue exclusion assay (n = 6). Horizontal bars represent the mean. A two‐tailed Mann–Whitney test was used to test for statistical significance.

2. The log₂ fold induction of tRFs for all isoacceptors in the overlapping set of tRNAs found in Fig 1C.

Figure EV2. Fragmentation of tRNA^Tyr _GUA occurs with other sources of oxidative stress and additional cell lines

1. Northern blot for tRF^Tyr _GUA and tRF^Leu _HAG in MCF10A cells at 1 h post‐oxidative stress (200 μM H₂O₂) exposure.

2. Northern blot for tRF^Tyr _GUA in C. elegans cells at 15 min post‐oxidative stress (200 μM H₂O₂) exposure.

3. A Northern blot depicting a time course experiment ranging from 5 min to 24 h for HBEC30 cells in response to oxidative stress (200 μM H₂O₂). As before, a single probe complementary to pre‐tRNA^Tyr _GUA, mature tRNA^Tyr _GUA, and tRF^Tyr _GUA was ³²P‐labeled and used for detection in (C) and (D).

4. A Northern blot depicting two time points, 1 and 24 h, after exposure to oxidative stress (200 μM H₂O₂) in MCF10A cells.

5. Quantification of tRNA^Tyr _GUA, tRNA^His _GUG, and tRNA^Glu _YUC by Northern blot analysis from two independent experiments 24 h (normalized to U6 levels) after exposure to oxidative stress (200 μM H₂O₂) are shown (n = 4).

6. Northern blot after MCF10A cells were treated with a pharmacological agent used to induce oxidative stress (menadione).

7. Quantification of tRNA^Tyr _GUA by Northern blot analysis from two independent experiments 24 h (normalized to U6 levels) after exposure to menadione are shown (n = 4).

Data information: Data represent mean ± SEM. *P < 0.05. A one‐tailed Mann–Whitney test was used to test for statistical significance.

Figure 2. TRNA^Tyr _GUA abundance is reduced while the corresponding tRF is induced in response to oxidative stress

1. A Northern blot depicting a time course experiment ranging from 5 min to 24 h of MCF10A cells in response to oxidative stress. A single probe complementary to pre‐tRNA^Tyr _GUA, mature tRNA^Tyr _GUA, and tRF^Tyr _GUA expression was ³²P‐labeled and used for detection.

2. Quantification of pre‐tRNA^Tyr _GUA (left) and mature tRNA^Tyr _GUA levels (right) by Northern blot analysis from multiple independent experiments (normalized to U6 levels) are shown (n = 6).

3. MCF10A cells were exposed to oxidative stress (200 μM H₂O₂) once daily for five continuous days.

4. Quantification of mature tRNA^Tyr _GUA bands by Northern blot after cells were treated once daily for five continuous days (normalized to U6) from multiple independent experiments (n = 12).

5. Quantification of Northern blot analysis for pre‐tRNA^Tyr _GUA (left) and tRNA^Tyr _GUA (right) after 1 and 24 h, respectively, in HBEC30 cells upon exposure to oxidative stress (200 μM H₂O₂) as in (A) (n = 6).

Data information: Data represent mean ± s.e.m. A one‐tailed Mann–Whitney test (**P < 0.01 and ***P < 0.001) was used to test for statistical significance between the treated and control cell lines for each time point.

Figure 3. Cell growth repression upon oxidative stress and tRNA^Tyr _GUA depletion

1. Growth curves of MCF10A cells exposed to oxidative stress (200 μM H₂O₂) relative to control cells (n = 3). Two‐way ANOVA was used to test for significance.

2. Northern blot of MCF10A cells expressing a control short‐hairpin RNA or a hairpin targeting tRNA^Tyr _GUA.

3. Western blot of MCF10A expressing a control short‐hairpin RNA or a hairpin targeting the tyrosine‐tRNA synthetase, YARS (red arrow). HSC70 was used as a loading control.

4. Growth curves of MCF10A cells expressing RNAi targeting mature tRNA^Tyr _GUA or YARS relative to cells expressing a control hairpin (n = 3). Two‐way ANOVA was used to test for significance.

5. Cell growth of MCF10A cells transiently transfected with a tRNA^Tyr _GUA overexpression vector relative to an empty control vector (n = 3). A one‐tailed Mann–Whitney test was used to test for significance at day 3.

Data information: Data represent mean ± SEM. *P < 0.05 and ***P < 0.001.

Figure EV3. Validation of YARS knockdown and tRNA^Tyr _GUA overexpression

1. Total mRNA from MCF10A cells stably expressing a short‐hairpin targeting YARS was analyzed by qRT–PCR. The two hairpins with the best knockdown were used for subsequent experiments. A one‐tailed Mann–Whitney test (*P < 0.05) was used to test for statistical significance.

2. Northern blot of MCF10A cells transiently transfected with a tRNA^Tyr _GUA overexpression vector or an empty control vector. TRNA^Tyr _GUA overexpression is shown at 24, 48, and 72 h post‐transfection.

Data information: Data represent mean ± SEM.

Figure EV4. Proteomic analysis identifies a tRNA^Tyr _GUA‐dependent gene‐set

1. A plot showing the correlation between protein abundance changes with tRNA^Tyr _GUA depletion or YARS depletion (shYARS‐2) relative to control cells. A Pearson’s two‐sided test was used to assess the statistical significance of the correlation between tRNA^Tyr _GUA and YARS depletion effects.

2. Schematic showing locations of all tyrosine residues in the coding sequence of EPCAM, SCD, and USP3.

3. Levels of mRNA expression for target genes in cells depleted of either tRNA^Tyr _GUA or YARS as measured by qRT–PCR (n = 4). A one‐tailed Mann–Whitney test was used to test for statistical significance.

4. MCF10A cells transfected with two independent siRNA targeting EPCAM, SCD, or USP3 were analyzed by qRT–PCR at the end of each growth assay. A one‐tailed Mann–Whitney test was used to test for statistical significance (n = 3 except for siSCD‐2 has n = 2).

Data information: Data represent mean ± SEM. *P < 0.05.

Figure 4. A set of growth‐promoting genes are sensitive to tRNA^Tyr _GUA depletion

1. Cells depleted of tRNA^Tyr _GUA or YARS were processed for label free quantitation by mass spectrometry to identify proteins that were reduced by a log₂‐fold change of 0.5 or more. This set was overlapped with proteins containing a higher than median abundance of Tyr codon content to identify candidate mediators of the pleiotropic effects of tRNA^Tyr _GUA depletion.

2. GO functional analysis of the 109 candidate gene‐set from (A).

3. Quantitative Western blot validation depicting abundances of protein targets (EPCAM, SCD, and USP3) identified from (A). HSC70 was used as a loading control and is not modulated upon molecular perturbation of tRNA^Tyr _GUA.

4. Quantification of Western blot analysis in (C) (n = 4). A one‐tailed Mann–Whitney test was used to test for statistical significance.

5. Growth curves for MCF10A cells were transfected with either control siRNA or two independent siRNA targeting EPCAM, SCD, or USP3. Note that the control cell growth curve is the same in all graphs and were plotted separately for clarity and does not represent independent experiments. Two‐way ANOVA was used to test for significance (n = 3).

Data information: Data represent mean ± SEM. *P < 0.05, **P < 0.01. Source data are available online for this figure.

Figure 5. Global ribosome occupancy analysis from tRNA^Tyr _GUA‐depleted cells reveals reduced translation efficiency for Tyr‐enriched genes

1. Quantitative Western blot EPCAM, SCD, and USP3 in MCF10A cells 24 h after treatment with H₂O₂ (200 μM). HSC70 was used as a loading control.

2. Quantification of Western results in (A) (n = 9). A one‐tailed Mann–Whitney test was used to test for statistical significance.

3. A schematic of the codon‐based USP3 reporter. A Myc‐tagged WT or mutant reporter with 5 Tyr codons mutated to Ala codons were cloned upstream of a Luciferase used for transfection normalization.

4. Quantitative Western blot for the Myc‐tag and Luciferase (top) with normalized fluorescent intensities (bottom) are shown (n = 3).

5. Ribosome occupancy of 20‐22nt RPFs in tRNA^Tyr _GUA‐depleted cells compared to control cells reveal greater occupancy at both tyrosine codons in tRNA^Tyr _GUA‐depleted cells using a Wilcoxon test.

6. Genes were sorted based on their changes in GC‐corrected translation efficiency (TE) values, with reduced TE in tRNA^Tyr _GUA‐depleted cells shown in left and enhanced TE shown on right. The red bars over each column depict the range of values in that bin. Distribution of genes with high tyrosine codon content across these three bins were assessed using mutual information calculation and testing (see Materials and Methods for details). For visualization, we used the hypergeometric distribution to assign P‐values to the overlap between tyrosine‐rich genes and each bin. We then defined an enrichment score as –log of p‐value, if there was a significant enrichment. If the overlap is significantly fewer than expected by chance, log of p‐value is used instead (depletion). The resulting enrichment score is then shown as a heatmap with gold depicting positive enrichment.

Data information: Data represent mean ± SEM. *P < 0.05 and ***P < 0.001. Source data are available online for this figure.

Figure EV5. Proteomic and ribosomal profiling validation of tRNA^Tyr _GUA‐depleted cells

1. MRNA expression levels for target genes 24 h after treatment with H₂O₂ (200 μM) as measured by qRT–PCR (n = 9). A one‐tailed Mann–Whitney test was used to test for statistical significance.

2. Examples of the mapped position of the 5′‐end of reads near the start (top) or stop (bottom) codons are shown, revealing the characteristic 3‐nucleotide periodicity previously described by (Ingolia et al, 2009).

3. Histogram of the read length distribution of RPFs observed upon ribosome profiling sequence analysis.

4. Codon usage scatter plots of the ribosomal A‐site of shTyr vs shControl from replicate ribosomal profiling experiments. The five most affected codons are labeled with UAC (Tyr) being the most affected (red arrow). Gray outline illustrates the 95% confidence interval around the regression line.

Data information: Data represent mean ± SEM. ***P < 0.001.

Figure 6. Identification of proteins that interact with tRF^Tyr _GUA

1. A, B

   Quantification of tRF^Tyr _GUA induction in response to oxidative stress as a function of time in MCF10A (A) (n = 4) and in HBEC30 (B) (n = 6) with mean ± SEM shown for each cell line.

2. C

   Volcano plot of mass spectrometry results from a synthetic 5′‐biotinylated tRF^Tyr _GUA co‐precipitation experiment with cell lysate. Log₂ fold enrichment values of proteins identified from tRF^Tyr _GUA relative to scrambled tRF control samples.

3. D

   Western blot validation of mass spectrometry results for three of the top hits in (A), showing co‐precipitation of endogenous proteins with transfected tRF^Tyr _GUA relative to a scrambled control sequence (Scr).

Source data are available online for this figure.

Figure 7. Functional effects of interactions between hnRNPA1 and tRF^Tyr _GUA

1. Schematic depicting the expected visualization of a crosslinked immunoprecipitation by autoradiogram. Bands corresponding to smRNA‐RBP interactions (no RNase digestion) or a smear representing mRNA‐RBP interactions (low RNase digestion) by autoradiogram were processed for HITS‐CLIP.

2. IGV plots from an hnRNPA1 HITS‐CLIP (Huelga et al, 2012) reveals interactions with tRF^Tyr _GUA.

3. IGV plots representing SSB interacting with the tRF^Tyr _GUA in samples that were treated with low levels of RNase digestion. SSB bound tRF^Tyr _GUA reads mapped to multiple loci encoding tRNA^Tyr _GUA.

4. Similar to the IGV plots shown in (C), but depicting SSB interactions with tRF^Tyr _GUA loci in samples without RNase digestion.

5. A cumulative distribution in control and hnRNPA1 depleted cells of the stability levels for mRNA transcripts with 3′ UTR hnRNPA1 CLIP binding (Huelga et al, 2012). Transfection of tRF^Tyr _GUA led to a significant right‐shift in the expression levels of 3′ UTR bound hnRNPA1 transcripts. Statistical significance was measured using the Kolmogorov–Smirnov test.

6. Cumulative distribution as in (E). Transfection of locked nucleic acid against tRF^Tyr _GUA and treatment with 200 µM H₂O₂ led to a significant left‐shift in mRNA stability of 3′UTR bound hnRNPA1 transcripts. Statistical significance was assessed using the Kolmogorov–Smirnov test.

7. Model of tRNA^Tyr _GUA‐dependent gene regulatory response to oxidative stress.