Decoding of translation-regulating entities reveals heterogeneous translation deficiency patterns in cellular senescence

Abstract

Cellular senescence constitutes a generally irreversible proliferation barrier, accompanied by macromolecular damage and metabolic rewiring. Several senescence types have been identified based on the initiating stimulus, such as replicative (RS), stress-induced (SIS) and oncogene-induced senescence (OIS). These senescence subtypes are heterogeneous and often develop subset-specific phenotypes. Reduced protein synthesis is considered a senescence hallmark, but whether this trait pertains to various senescence subtypes and if distinct molecular mechanisms are involved remain largely unknown. Here, we analyze large published or experimentally produced RNA-seq and Ribo-seq datasets to determine whether major translation-regulating entities such as ribosome stalling, the presence of uORFs/dORFs and IRES elements may differentially contribute to translation deficiency in senescence subsets. We show that translation-regulating mechanisms may not be directly relevant to RS, however uORFs are significantly enriched in SIS. Interestingly, ribosome stalling, uORF/dORF patterns and IRES elements comprise predominant mechanisms upon OIS, strongly correlating with Notch pathway activation. Our study provides for the first time evidence that major translation dysregulation mechanisms/patterns occur during cellular senescence, but at different rates depending on the stimulus type. The degree at which those mechanisms accumulate directly correlates with translation deficiency levels. Our thorough analysis contributes to elucidating crucial and so far unknown differences in the translation machinery between senescence subsets.

Keywords: IRES elements; oncogene-induced senescence; replicative senescence; ribosome stalling; stress-induced senescence; translation deficiency; uORF/dORF.

FIGURE 1

Replicative senescence is accompanied by translation deficiency in vitro and in vivo. (a) Volcano plots demonstrating transcripts with significantly decreased (blue) and increased (red) translation rate in human W138 lung fibroblasts undergoing RS versus control. Bar graphs indicate the percentage of significant (p < 0.05) changes in translation efficiency. See also Table S2. (b) Same as (a) for aged (26 months old) mouse liver versus liver tissue from younger mice, whose age is displayed incrementally on the t axis. Translation deficiency changes are progressively diminished as 26‐month‐old mice are compared with mice approaching their age. See also Figure S2A–C.

FIGURE 2

Replicative senescence displays no distinct translation deficiency patterns. (a) Ribosome stalling in E‐, P‐ or A‐ sites in human W138 lung fibroblasts under RS versus control. Red coloring in the box plot (left) indicates codons where ribosomes are most stalled. (b) Top: The bar graphs indicate a non‐significant overall difference in the percentage of stalled codons per library between RS and control. Bottom: Cumulative distribution function (CDF) curve showing non‐significant differences in translation efficiency between stalled codons in RS versus control cells. See also Table S2. (c) Same as (a), for aged (32 months old) mouse kidney tissue versus respective young tissue (3 months old). (d) Top: No significant overall differences in the percentage of stalled codons per library were identified. See also Figure S3B. Bottom: CDF curve showing no significant differences in translation efficiency between stalled codons in 32‐month versus 3‐month mouse kidney cells. (e) Same as (a), for aged (32 months old) mouse liver tissue versus respective young tissue (3 months old). (f) Top: No significant overall differences in the percentage of stalled codons per library were identified. See also Figure S3C. Bottom: CDF curve showing no significant differences in translation efficiency between stalled codons in 32‐month versus 3‐month mouse liver cells. (g) Venn diagrams depicting the number of transcripts undergoing translation efficiency changes in cells under RS, OIS and oxidative stress (H₂O₂). n.s., non‐significant. Error bars indicate SEM.

FIGURE 3

uORF‐mediated translation deregulation upon oxidative stress‐induced senescence. (a) Volcano plots demonstrating genes with significantly decreased (blue) and increased (red) translation rate in H₂O₂‐treated HeLa cells versus control. Bar graphs indicate the percentage of significant (p < 0.05) changes in translation efficiency. See also Table S2. (b) Ribosome stalling differences derived by comparing the normalized EPA coverage per codon for H₂O₂‐treated HeLa cells versus control. Red coloring in the box plots indicates codons where ribosomes were most stalled. (c) Left: Bar graphs indicate overall differences in the percentage of stalled codons per library in E‐, P‐ and A‐ ribosome sites between H₂O₂‐treated HeLa cells and control. Right: CDF curve showing an overall non‐significant difference in translation efficiency between stalled codons in H₂O₂‐treated HeLa cells versus untreated counterparts. (d) Identification of uORF dominant motifs using the MEME motif finding platform. CUCUU sequences resembling candidate 5′ TOP motifs are found in the identified motifs. (e) Pie chart displaying the percentage (%) of total identified uORFs found in the indicated conditions. A statistically significant increase in mRNAs carrying uORFs was observed upon H₂O₂ treatment versus control. (f) Bar graph displaying the percentage (%) of total IRES elements found in the indicated conditions. (g) Pie chart displaying the distribution of uORF start codons derived by observing the nt sequences at the start of the periodicity at each 5′ UTR. No differences in uORF start codons were observed between H₂O₂‐treated HeLa and control cells. (h). Pathway enrichment analysis for genes regulated by uORFs using the WebGestalt platform. *p < 0.05, of Student's t‐test; n.s., non‐significant; FDR, False Discovery Rate. Error bars indicate SEM.

FIGURE 4

Oncogene‐induced senescence displays increased ribosome stalling, uORF/dORF patterns and IRES elements. (a) Volcano plots of genes with significantly decreased (blue) and increased (red) translation rate in human primary BJ fibroblast cells undergoing OIS versus control. Bar graphs indicate the percentage of significant (p < 0.05) changes in translation efficiency. See also Table S2. (b) Ribosome stalling differences derived by comparing the normalized EPA coverage per codon for OIS BJ fibroblast cells versus control. Red coloring in the box plots indicates codons where ribosomes are most stalled. (c) Top: Bar graphs indicating significant differences in the percentage of stalled codons per library in E‐, P‐ and A‐ ribosome sites between OIS and control BJ fibroblasts. Bottom: The CDF plot of the transcripts where stalling is observed shows a significant drop in translation efficiency in OIS. (d) Top: Ribosome dwell times heatmap referring to codons of the EPA ribosome sites where increased stalling may occur based on known lower translation elongation rates (Gobet et al., 2020). Combinations of codons with increased dwell times is likely to result in considerable ribosome stalling. Green color stands for increased, while red color for decreased dwell times. Bottom: Identified codon stalling resulting in respective changes in translated amino acids. (e) Identification of uORF dominant motifs using the MEME motif finding platform. CUCUU sequences resembling candidate 5′ TOP motifs are found in the identified motifs. (f) Left: The CDF curve is derived only from uORF‐carrying transcripts and exhibits a significant decrease in OIS translation efficiency versus control. Right: The pie chart shows the percentage (%) of total identified uORFs found in the indicated conditions. See also Figure S4A. (g) Bar graph with the percentage (%) of IRES elements found in uORFs of OIS versus control samples after folding the identified uORF domains with the Vienna algorithm (See also Section 4). (h) Pie chart showing the distribution of uORF start codons, with no significant differences between OIS and control BJ fibroblasts. (i) CDF plots with dORF‐carrying transcripts show significant increase of translation efficiency in OIS versus control BJ fibroblasts. (j) As in (e), dominant dORF motifs are evaluated using the MEME suite. (k) Pie chart presenting the dORF start codon distribution, with no significant changes between OIS and control BJ fibroblasts. Statistics for the CDF plots are extracted with a Wilcoxon rank sum test. *p < 0.05, of Student's t‐test; Error bars indicate SEM.

FIGURE 5

Translation deregulation patterns are experimentally recapitulated in senescence‐induced fibroblasts in vitro. (a) Schematic illustrating the experimental strategy. IMR‐90 human lung fibroblasts were forced to senesce via replication stress, H₂O₂ treatment or RAS ^G12V gene induction. Cells were subsequently lysed and subjected to RNA‐seq and ribosome profiling. See Table S2. (b) CDF curve showing non‐significant differences in translation efficiency between stalled codons in RS versus control (proliferating) IMR‐90 cells. RNA‐seq and Ribo‐seq data were retrieved by our published dataset (GEO accession number: GSE171780) (Sofiadis et al., 2021). See also Figures S5A,D,E and S6. (c) IMR‐90 cells were induced to senesce by H₂O₂ and subsequently compared to proliferating counterparts. The CDF curve displays an increased impact of uORFs on translation deregulation upon H₂O₂ treatment. (d) Pie chart displaying the distribution of uORF start codons derived by observing the nt sequences at the start of the periodicity at each 5′ UTR. (e) Identified dominant uORF motifs using the MEME platform. CUCUU motifs are again observed. (f) Pathway enrichment analysis for genes regulated by uORFs using the WebGestalt platform. See also Figures S5B,D,E and S7. (g) Western blotting of control and H₂O₂‐treated cell lysates with indicated antibodies, verifying enhancement of TGF‐β signaling (SMAD3, pSMAD3) and downregulation of VEGF signaling upon H₂O₂‐induced senescence. These observations are in line with our published dataset analyses. See also Figure S7F. (h) Bar graphs indicating significant differences in the percentage of stalled codons per library in E‐, P‐ and A‐ ribosome sites between OIS and control IMR‐90 fibroblasts. (i) Ribosome dwell times heatmap and common codon stalling in E‐, P‐ and A‐ sites. Green color stands for increased, while red color for decreased dwell times. Those in vitro results are in agreement with our published OIS versus control dataset analyses. See also Figures S5C–E and S8A–C. (j) Identification of uORF dominant motifs using the MEME platform in OIS versus control IMR‐90 cells. CUCUU sequences were detected (potential TOP‐like motifs). (k) The CDF curve is derived only from uORF‐carrying transcripts and exhibits a significant decrease in OIS translation efficiency versus control. See also Figures S7D and S8D. (l) Bar graph demonstrating an increased percentage (%) of IRES elements in uORFs of OIS versus control cells. (m) Dominant dORF motifs were similar to the ones identified in our published dataset analyses. See also Figure S8E. (n) CDF plots with dORF‐carrying transcripts show significant increase of translation efficiency in OIS versus control IMR‐90 fibroblasts. (o) Western blotting in control (proliferating) and OIS IMR‐90 cell lysates with indicated antibodies, verifying enhancement of Notch signaling via HES1 upregulation. See also Figure S8F,G. Statistics for the CDF plots are extracted with a Wilcoxon rank sum test. FDR, False Discovery Rate. Vectors were obtained from www.vecteezy.com.

FIGURE 6

Model. Although the frequency of translation deregulation mechanisms is similar between aged cells undergoing RS and young counterparts, SIS displays a clear manifestation of deregulation patterns. In the case of OSIS, uORFs are significantly enriched compared to control, which correlates with upregulation/downregulation of distinct signaling pathways. OIS, on the contrary, is characterized by significant enrichment of ribosome stalling, uORF/dORF patterns and IRES elements, accompanied by upregulation of Notch signaling activators. Our model demonstrates that the differential rates of translation deregulating mechanisms may be hallmarks of separate types of cellular senescence. Vectors were obtained from www.vecteezy.com.