Stress-Induced Translation Inhibition through Rapid Displacement of Scanning Initiation Factors

Abstract

Cellular responses to environmental stress are frequently mediated by RNA-binding proteins (RBPs). Here, we examined global RBP dynamics in Saccharomyces cerevisiae in response to glucose starvation and heat shock. Each stress induced rapid remodeling of the RNA-protein interactome without corresponding changes in RBP abundance. Consistent with general translation shutdown, ribosomal proteins contacting the mRNA showed decreased RNA association. Among translation components, RNA association was most reduced for initiation factors involved in 40S scanning (eukaryotic initiation factor 4A [eIF4A], eIF4B, and Ded1), indicating a common mechanism of translational repression. In unstressed cells, eIF4A, eIF4B, and Ded1 primarily targeted the 5' ends of mRNAs. Following glucose withdrawal, 5' binding was abolished within 30 s, explaining the rapid translation shutdown, but mRNAs remained stable. Heat shock induced progressive loss of 5' RNA binding by initiation factors over ∼16 min and provoked mRNA degradation, particularly for translation-related factors, mediated by Xrn1. Taken together, these results reveal mechanisms underlying translational control of gene expression during stress.

Keywords: RNA-binding sites; UV crosslinking; mass spectrometry; protein-RNA interaction; stress responses; yeast.

Graphical abstract

Figure 1

The Impact of Stress on the Yeast RNA-Binding Proteome (A) Summary of the TRAPP protocol. See main text for details. 4tU, 4-thiouracil; GTC, guanidinium thiocyanate. (B) Time course showing changes in RNA association during glucose starvation for individual RBPs in TRAPP analyses. (C) Same as (B) but for heat shock. (D) Density plot showing changes in RNA binding at 16 min following a mock shift, glucose starvation, or heat shock. (E) Bar chart showing all proteins with greater than 2-fold change in RNA association after 16 min of glucose starvation (left) or a mock shift (center). The right-hand panel shows changes in protein abundance following 16 min of glucose starvation (right). Error bars represent standard deviation. (F) Principal-component analysis (PCA) showing differences between conditions and time points. Axis titles show the extent of variation explained by a given principal component. (G) PCA comparing the changes in RNA binding for individual proteins following heat shock, glucose starvation, or mock shift of 16 min. See also Figure S1.

Figure 2

Changes in RNA Binding among Ribosomal Proteins (A) Time course showing changes in RNA binding during glucose withdrawal (upper; blue lines) and heat shock (lower; red lines) for various classes of RBPs in TRAPP analyses. Included in the figure are ribosome biogenesis factors (left), large ribosomal subunits (center), and small ribosomal subunits (right). (B) Scatterplot comparing the effects of glucose starvation and heat shock at 16 min on RNA binding for ribosomal proteins. (C) Time course showing RNA association for Rps2, Rps3, and Rps5. (D) Crystal structure (PDB: 3J77) of the yeast ribosome highlighting the changes in RNA association for each detected ribosomal protein. (E) A closeup view of the mRNA entry channel with amino acid-RNA crosslinking sites highlighted in green.

Figure 3

Changes in RNA Binding among Translation Initiation Factors (A) Overview of the translation initiation process. Each protein is colored according to its change in RNA association during glucose starvation in TRAPP analyses. Translation initiation factors shown in gray were not detected as RNA binding. (B) Time course showing changes in RNA binding for eIF4A, eIF4B, and Ded1 following glucose starvation (blue) and heat shock (pink) in TRAPP analyses. See also Figure S2B.

Figure 4

Genome-wide Analysis of the RNA Binding Profiles of eIF4A, eIF4B, and Ded1 (A) Breakdown of eIF4B-bound RNAs by biotype in CRAC analyses. (B) Boxplot showing the changes in eIF4B binding to individual mRNAs following either mock shift or stress in CRAC analyses. Each box represents the median with 25^th and 75^th percentiles. The whiskers show the 10^th and 90^th percentiles. ^∗∗∗p < 10⁻¹⁵ relative to mock, using the unpaired t test to compare means. For (B)–(E), all analyses are based on a set of 2,000 transcripts that show the strongest binding to eIF4B in unstressed conditions. (C) Same as (B) but for Ded1. (D) Scatterplots comparing the changes in RNA binding between control and either mock shift (16 min), glucose starvation (30 s), glucose starvation (16 min), or heat shock (16 min). (E) Metaplots showing the distribution of eIF4A, eIF4B, and Ded1 binding around the mRNA start codon in CRAC analyses. (F) Binding of eIF4A, eIF4B, and Ded1 across the ACT1, URA5, RPL6B, and TEF1 mRNAs. Each set of tracks is normalized to total library size using reads per million, with the exact value indicated in the upper right corner of each box. RNA-seq traces are shown at the bottom as a control. Each track is normalized to a spike-in control and thus represents the absolute abundance of each mRNA compared to the control. Each box represents a 3-kb window; a scale bar is shown at the bottom. The open reading frames (ORFs) are indicated as black boxes, with UTRs as flanking gray boxes of intermediate thickness. See also Figures S3–S5.

Figure 5

Global Analysis of mRNA Levels in Response to Stress (A) Bar graph showing the change in total mRNA abundance relative to an S. pombe spike-in control following glucose withdrawal (blue), heat shock (red), or heat shock plus cycloheximide (purple) for 16 min. (B) As in (A) but with boxplots showing changes in the abundance for 5,000 individual mRNAs. All samples were normalized to a single control sample (not shown). (C) Scatterplots comparing mRNA levels following glucose starvation (upper), heat shock (middle), or heat shock plus cycloheximide (lower) for 16 min relative to control. Points below the dotted red line indicate mRNAs with reduced abundance following stress. Each plot includes the 5,000 most-abundant mRNAs. RPKM (reads per kilobase per million) values were adjusted to account for the spike-in control. (D) GO (gene ontology) term enrichment among the 500 most-decreased mRNAs for each stress. (E) Violin plots showing the changes in mRNA levels for ribosomal protein (RP) mRNAs (upper) or ribosome biogenesis (RiBi) mRNAs (lower). (F) CRAC analysis showing binding of eIF4B across a selected genomic region (upper). RNA-seq tracks are normalized to a spike-in control and thus represent the absolute abundance of each mRNA compared to the control (lower). A scale bar is shown at the bottom. The ORFs are indicated as black boxes, with UTRs as flanking gray boxes of intermediate thickness. (G) Metaplots showing the distribution of eIF4B binding around the mRNA start codon. See also Figures S5 and S6B.

Figure 6

Cycloheximide Treatment or xrn1Δ Inhibits mRNA Decay during Heat Shock (A) Scatterplots comparing mRNA levels following heat shock for 16 min (x axis) relative to heat shock plus cycloheximide (upper); heat shock in a strain lacking Ski2 (middle); heat shock in a strain lacking Xrn1 (lower). Each plot includes the 5,000 most-abundant mRNAs. (B) Violin plots showing changes in mRNA levels among transcripts encoding ribosomal protein or ribosome biogenesis factors. (C) Model of the translational response to glucose starvation and heat shock. Upon exposure to either stress, the 40S scanning factors eIF4A, eIF4B, and Ded1 dissociate from the 5′ end of mRNAs, halting translation initiation. Already-initiated ribosomes continue translating before eventually terminating, potentially leaving “naked” mRNAs unprotected by the translational machinery. In the case of heat shock, Xrn1 is involved in degradation of a subset of these transcripts. With glucose starvation, by contrast, most mRNAs remain relatively stable.