Chronic starvation induces noncanonical pro-death stress granules

Abstract

Stress granules (SGs) assemble under stress-induced conditions that inhibit protein synthesis, including phosphorylation of eIF2α, inhibition of the RNA helicase eIF4a proteins or inactivation of mTORC1. Classically defined SGs are composed of translation initiation factors, 40S ribosomes, RNA-binding proteins and poly(A)⁺ mRNAs. As such, they represent an important compartment for storage of mRNAs and regulation of their translation. Emerging work on SGs indicates that these structures might promote cellular survival in diverse disease states. Yet, much work on SG formation and function employs acute stress conditions, which might not accurately reflect the chronic stresses that manifest in human disease. Here, we used prolonged nutrient starvation to model and investigate SG formation and function during chronic stress in a human cell line and mouse embryonic fibroblasts. Surprisingly, we found that SGs that form under chronic nutrient starvation lack 40S ribosomes, do not actively exchange their constituent components with cytoplasmic pools and promote cell death. We named these SGs starvation-induced SGs (stSGs). Our results on stSGs imply that SG assembly and function in the context of prolonged nutrient starvation stress differ significantly from what has been described for acute stress conditions.

Keywords: Cell death; G3BP1; Starvation; Stress granule; Translation; eIF2α.

Fig. 1.

Chronic nutrient starvation induces SGs with slow assembly kinetics. (A) U2OS cells were starved of nutrients for the indicated times prior to fixation and staining with antibodies against G3BP1 (green) and eIF3b (red). (B) Quantification of the percentage of cells with stSGs from A. (C) Western blot analysis of starvation stress kinetics. Puromycin labeling was used to analyze ongoing translation, and pathways known to mediate translation repression were examined (mTORC1 and eIF2α), as indicated. Molecular weight standards are indicated. (D) ATP levels during starvation at 8 h and 16 h time points. ATP was detected using a luciferase-based luminescence assay. ATP levels are expressed as a percentage of control (fed) values at each time point. (E) U2OS cells were starved for 16 h prior to the addition of glucose. Cells were then incubated for the indicated times before harvesting and processing as in A. (F) Quantification of the percentage of cells with stSGs from E. (G) U2OS cells were treated as in E before ATP analysis. ATP levels are expressed as a percentage of control (fed) values at each time point. (H) U2OS cells were treated as indicated and harvested for polysome profiling. M/P represents the monosome/polysome ratio for the indicated graph. The y-axis is voltage units proportional to OD254 and the x-axis is fraction number. (I) Western blot analysis of the unfractionated extracts shown in H. Results from all panels are representative of a minimum of three experimental replicates, and data are presented as mean±s.d.

Fig. 2.

stSGs have similar composition to other types of stSG, but lack 40S ribosomes. (A) U2OS cells were starved for 16 h followed by fixation and staining for the indicated SG constituents. G3BP1 is the most well-studied RNA-binding protein within stSGs; HuR and Tia1 are other well-studied stSG constituents; poly(A)⁺ mRNAs are included in canonical, acute SGs; eIF3b is a translation initiation factor and its localization with stSGs defines stSGs; rps6 is a subunit of the 40S ribosome. (B) RNA FISH was performed with probes directed against 18S and 28S rRNA after U2OS cells were exposed to either 250 μM arsenite stress for 1 h or chronic nutrient starvation (as indicated). (C) Quantification of colocalization between 18S rRNA and G3BP1. Colocalization was measured using a Manders’ correlation test, where 0 indicates no colocalization and 1 indicates perfect colocalization. Each data point represents the Manders’ coefficient for each cell, spanning eight fields per condition. Microscopy was performed with a 100× objective. A two-tailed Student's t-test was used to measure statistical significance (***P<0.001), and data are presented as mean±s.e.m. (D) U2OS cells were starved for 16 h prior to the addition of glucose for 1 h. Cells were fixed and stained as in B with 18S probes. Results from all panels are representative of a minimum of three experimental replicates.

Fig. 3.

Chronic nutrient starvation induces static SGs that do not exchange material with translating mRNPs. (A) U2OS cells were either treated with 250 μM arsenite for 1 h or starved for 16 h prior to fixation and staining with antibodies against G3BP1 (green) and TiaR (red). During the last hour of stress, 20 μM emetine or 50 μg/ml cycloheximide were added to trap SG constituents within translating mRNPs by stalling elongating ribosomes on mRNAs to assess exchange of material with stSGs. (B) Quantification of results shown in A. The y-axis depicts the percentage of cells with stSGs for each condition. Statistical analysis was performed with a Chi-test (NS, not significant; ***P<0.001). (C) U2OS cells were starved for 16 h before addition of glucose, cycloheximide or glucose and cycloheximide, as indicated, for 1 h prior to fixation. Cells were stained as in A. (D) Quantification of the results shown in C. Results from all panels are representative of a minimum of three experimental replicates, and data are presented as mean±s.d.

Fig. 4.

Role of eIF2α kinases in translation repression and stSG induction during chronic nutrient starvation. (A) Wild-type or S51A eIF2α mutant MEFs were stressed for 12 h prior to fixation and staining of stSGs with antibodies against G3BP1 (green) or eIF3b (red). (B) Quantification of the results shown in A. Statistical analysis was performed with a Chi-test (***P<0.001). (C) Wild-type or eIF2α kinase KO MEFs indicated were stressed as in A followed by fixation and staining. (D) Quantification of the results shown in C. Statistical analysis was performed with a Chi-test (***P<0.001). (E) MEFs were starved and lysates were prepared for western blotting for the puromycin-labeled polypeptides to assess ongoing translation, or the indicated proteins. Molecular weight standards are indicated. Quantification of western blots was performed using image studio for Li-Cor to avoid saturation effects, and ratios of total to phosphoproteins are indicated. (F) Wild-type and S51A eIF2α mutant MEFs were starved in the presence of ethidium homodimer-1, as described above. Ethidium fluorescence was normalized to 0 h and plotted against time. Results from all panels are representative of a minimum of three experimental replicates, and data are presented as mean±s.d.

Fig. 5.

Chronic nutrient starvation induces pro-death stSG. (A) Wild-type and ΔΔ U2OS cells were starved for 16 h, followed by western blotting for puromycin or the indicated proteins. Molecular weight standards are indicated. (B) Cells treated as in A were stained for stSGs using eIF3b (green) and Tia1 (red) after 16 h starvation. (C) Phase-contrast images taken with a 10× objective, showing the appearance of the respective monolayers for wild-type and ΔΔ U2OS cells. (D) Flow cytometry was performed on fed and starved U2OS cells (Fig. S4) stained with annexin V and propidium iodide to monitor the proportion of cells undergoing cell death. (E) Graphical illustration of flow cytometry results in U2OS cells. The y-axis is the percentage of cells in each category, with the total equaling 100%. (F) Wild-type and S51A eIF2α mutant MEFs were subjected to the same analysis as in D and are represented as in E. (G) U2OS cells were stressed with arsenite or starved for 16 h, as indicated, before fixation and staining for RACK1 (green) and G3BP1 (red). Results from all panels are representative of a minimum of three experimental replicates.