Cytoplasmic circular dsDNA is a key constituent of stress granules
- Laboratory of Nucleic Acids, National Heart, Lung, and Blood Institute, National Institutes of Health, United States
Figures
Figure 1 with 1 supplement
DNA is abundant in yeast and mammalian stress granule cores.
(A) Abundance of histones (brown) and other representative DNA-binding proteins (gray) in the proteomes of yeast stress granule cores (strain JD1370 grown in synthetic defined (SD) media), normalized to that of the stress granule marker eIF4A [peptide spectrum matches (PSMs); means from n = 2 biological replicates ± s.d.; n.s., no significance; one-tailed Student’s t-test]. (B) Abundance of histones and other representative DNA-binding proteins in the proteomes of HEK293T cell stress granule cores, normalized to that of the stress granule marker G3BP1 PSMs; means from n = 3 biological replicates ± s.d.; *p < 0.05; one-tailed Student’s t-test. Stress granule markers, histones and non-histone DNA-binding proteins in green, yellow, and gray, respectively. (C) Negative-stain immuno-electron micrograph of stress granule cores from yeast strain YAG1021 carrying a chromosomally encoded FLAG-tagged histone H3. Anti-FLAG antibodies were conjugated to NanoGold beads (arrows; Methods). Magnification (×23,000). (D) Native agarose gel electrophoresis (SYBR Gold stain) of yeast stress granule cores treated with DNase I. BP, bromophenol blue. (E) DNase I, psDNase, and RNase H treatment (the first at two different concentrations) of total nucleic acids extracted from yeast stress granule cores analyzed as in (D). Bottom graph, integrated signal for the dashed box. (F) Comparison of nuclease susceptibility of total nucleic acids extracted from yeast and mammalian stress granule cores, analyzed as in (D). (G) Alkaline hydrolysis of total nucleic acids extracted from yeast stress granule cores (15, 10, and 5 ng) and ribosomal RNA (rRNA; 15 ng), analyzed by non-denaturing agarose gel electrophoresis (SYTOX Green stain). Samples marked (+) were incubated with 50 mM KOH for 15 min at 95°C. Bottom graph, integrated signal for the dashed box.
-
Figure 1—source data 1
Original file for agarose gel analysis displayed in Figure 1D.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-data1-v1.zip
-
Figure 1—source data 2
TIFF file containing uncropped agarose gels for Figure 1D, indicating the relevant bands and/or treatments.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-data2-v1.zip
-
Figure 1—source data 3
Original file for agarose gel analysis displayed in Figure 1E.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-data3-v1.zip
-
Figure 1—source data 4
TIFF file containing uncropped agarose gels for Figure 1E, indicating the relevant bands and/or treatments.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-data4-v1.zip
-
Figure 1—source data 5
Original file for agarose gel analysis displayed in Figure 1F.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-data5-v1.zip
-
Figure 1—source data 6
TIFF file containing uncropped agarose gels for Figure 1F, indicating the relevant bands and/or treatments.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-data6-v1.zip
-
Figure 1—source data 7
Original file for agarose gel analysis displayed in Figure 1G.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-data7-v1.zip
-
Figure 1—source data 8
TIFF file containing uncropped agarose gels for Figure 1G, indicating the relevant bands and/or treatments.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-data8-v1.zip
Figure 1—figure supplement 1
Proteinaceous and DNA constituents of stress granule cores.
(A) Abundance of representative DNA-binding proteins in the proteomes of stress granule cores isolated from early-log HEK293T cells (Demeshkina and Ferré-D’Amaré, 2025). Green color denotes DNA-binders also found in immunoprecipitated stress granules from late-log U-2 OS cells (Jain et al., 2016); orange color marks DNA-binders additionally identified in the cores of HEK293T cells. Relative abundance of proteins between datasets was normalized using G3BP1 (data are mean ± s.d. from n = 3 biological replicates). (B) Native agarose gel electrophoresis (ethidium bromide stain) of the yeast large ribosomal subunit (y60S) and yeast stress granule cores treated with proteinase K, ribonucleases A and T1 (A/T1), and combination of these enzymes. Incubation was performed at 37°C for 24 hr. (C) Native agarose gel electrophoresis (ethidium bromide stain) of y60S and yeast stress granule cores treated with DNase I. Incubation was performed at 37°C for 1 hr. (D) Negative-stain electron micrograph of yeast stress granule cores treated with DNase I for 30 min at 30°C. Magnification (×13,000). (E) Alkaline hydrolysis of total nucleic acids extracted from yeast stress granule cores (15, 10, and 5 ng), ribosomal RNA (rRNA; 15 ng), and 11 kbp plasmid (15 ng), analyzed by non-denaturing agarose gel electrophoresis (SYBR Gold stain). Samples marked (+) were incubated with 50 mM KOH for 15 min at 95°C. Bottom graph, integrated signal for the dashed box. In (B) and (C), BP stands for bromophenol blue.
-
Figure 1—figure supplement 1—source data 1
Original file for agarose gel analysis displayed in Figure 1—figure supplement 1B.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-figsupp1-data1-v1.zip
-
Figure 1—figure supplement 1—source data 2
TIFF file containing uncropped agarose gels for Figure 1—figure supplement 1B, indicating the relevant bands and/or treatments.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-figsupp1-data2-v1.zip
-
Figure 1—figure supplement 1—source data 3
Original file for agarose gel analysis displayed in Figure 1—figure supplement 1C.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-figsupp1-data3-v1.zip
-
Figure 1—figure supplement 1—source data 4
TIFF file containing uncropped agarose gels for Figure 1—figure supplement 1C, indicating the relevant bands and/or treatments.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-figsupp1-data4-v1.zip
-
Figure 1—figure supplement 1—source data 5
Original file for agarose gel analysis displayed in Figure 1—figure supplement 1E.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-figsupp1-data5-v1.zip
-
Figure 1—figure supplement 1—source data 6
TIFF file containing uncropped agarose gels for Figure 1—figure supplement 1E, indicating the relevant bands and/or treatments.
- https://cdn.elifesciences.org/articles/111336/elife-111336-fig1-figsupp1-data6-v1.zip
Figure 2 with 1 supplement
Characteristics of eccDNAs from yeast and mammalian stress granule cores.
(A) Abundance and length distribution of high-confidence eccDNAs from yeast stress granule cores (unless noted, the data are from strain JD1370 cultured in SD medium). (B) Genomic coverage, per chromosome, of eccDNAs from yeast stress granule cores. The length of each chromosome is shown for comparison (overall correlation coefficient between relative eccDNA abundance and chromosome length is 0.95). (C) Abundance and length distribution of eccDNA from HEK293T cell stress granule cores. (D) Genomic coverage, per chromosome of eccDNA from HEK293T cell stress granule cores. The length of each chromosome is shown for comparison (overall correlation coefficient between relative eccDNA abundance and chromosome length is 0.95). (E) Relative abundance of gene elements within assemblies of eccDNAs from yeast and human stress granule cores. TSS, transcription start site; TTS, transcription termination site. (F) Repetitive elements within assemblies of eccDNAs of yeast stress granule cores. SINEs and LINEs, short and long interspersed nuclear elements, respectively. (G) Repetitive elements within eccDNAs from human stress granule cores.
Figure 2—figure supplement 1
Workflow for analysis of DNA from stress granule cores by eccDNA assemblers.
(A) Substrate specificity of psDNase. Species of DNA hydrolyzed by psDNase in the presence of ATP are in black. (B) Main steps of modified Circle-Seq treatment applied to enrich (step 1), amplify (step 2), and debranch (step 3) circular DNA from stress granule cores for further sequencing. (C) Bioinformatic workflow of processing DNA reads for assembly into eccDNA circles. Examples of genomic regions covered by short (Illumina) and long (Nanopore; ONT) reads, whose sequences are assembled in eccDNAs for yeast (D) and (E) and human (F) stress granule cores. Panels (A) and (B) created with BioRender.
© 2026, BioRender Inc. Parts of this image created with BioRender are made available under a Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Figure 3 with 1 supplement
Cytoplasmic eccDNAs colocalize with stress granules in human cells.
(A) Partitioning of AHNAK DNA signal between cytoplasm (c), perinuclear (pn), and nuclear (n) volumes of a HEK293T cell. Nuclear envelope is labeled with Lamins A and C. (B, C) Localization of MYC and AHNAK DNA signals in relation to the stress granule marker G3BP and its activator protein Caprin1. (D, E) Concurrent detection of DNA and RNA targets within AHNAK and MYC genes in relation to the stress granule marker G3BP. As in (A–C), denaturing conditions at 75°C were used to prioritize capturing DNA targets over mRNA targets (red arrows). (F, G) Detection of DNA by conventional anti-DNA IF combined with HCR-FISH against AHNAK and MAPT DNA targets. DNA signals (framed and numbered) proximal to translocase TOMM20 point to relationship with mitochondria. Cytoplasmic DNA without proximal TOMM20 label is marked by red arrows. (H) Colocalization of histone H4 with stress granules turnover protein VCP. Stress granules are demarcated by T30 signal (framed), which represents stress-induced condensation of polyadenylated nucleic acids. (I) Colocalization of histone H3 with the granule marker G3BP and control AHNAK mRNA. Colocalized signals for H3, G3BP, and mRNA are framed and numbered. Arrows point to some H3 and G3BP signals without AHNAK mRNA foci. In all panels, HEK293T cells underwent oxidative stress with 0.5 mM sodium arsenite for 1 hr and 20 min at 37°C, and then were fixed with formaldehyde. Unless otherwise stated, detection of proteins and nucleic acids was carried out by HCR IF and HCR-FISH, respectively. All images represent one confocal Z-stack of 0.2 μm with nuclei stained by Hoechst 33342; proteins and DNA are pseudocolored for clarity. Scale bars: 1 μm in (A) and (F–I), 2 μm in (B–E), and 0.5 μm in inserts of (B, C).
Figure 3—figure supplement 1
Detection of stress granule core constituents in HEK293T cells by HCR fluorescence.
(A) Exemplary design of FISH probes for targeting DNA (magenta) and RNA (red) counterparts of a gene of interest. (B) Schematic of heat-denaturation step of HCR-FISH protocol aimed to detect DNA coding region (magenta). Concurrent detection of transcribed RNA with corresponding region of non-coding DNA strand is indicated (red). Representative DNA Illumina and Nanopore reads covering regions within AHNAK (C) and MYC (D) genes to which HCR-FISH probes (magenta) were designed. (E, F) Abundance of histone H4 in the cytoplasm of HEK293T cells. Two gains (top and bottom) demonstrate sensitivity and specificity of HCR immunolabeling. (G, H) HCR amplifiers are highly specific to primary targets (FISH probes or antibodies). In (E–H), all images represent one middle Airyscan confocal Z-stack of 0.2 µm. Nuclei were stained by Hoechst 33342 (blue) in (G) and (H).
Figure 4 with 3 supplements
Cytoplasmic eccDNA is required for stress granule formation in yeast.
(A–C) Confocal microscopy of S. cerevisiae with endogenous PAB1-GFP (green) and transiently expressed (GAL promoter) cytoplasmic CRISPR machinery (Cas9NES) with gRNA (Ty1). CRISPR transformant variants are indicated together with treatment conditions. (D) Analysis of total DNA isolated from stress granule cores from wild-type (no transformation) and CRISPR-treated (variants indicated) cells after exposure to oxidative stress with 0.5% (wt/vol) sodium azide for 45 min (means from n = 2 biological replicates ± s.d.). (E) Depletion of 20-nt Ty1 (Ty12HDV) targets by enzymatically active cytoplasmic Cas9NES (3rd generation; Methods). Mann–Whitney U-test. (F) Effect of cytosolic CHD1NES on abundance of circular double-stranded DNA from stress granules isolated from cells co-expressing active CRISPR machinery with gRNA as in (B) (means from n = 2 biological replicates ± s.d.). (G, H) Confocal microscopy of untreated and oxidatively stressed S. cerevisiae with an endogenous PAB1-GFP fusion (green) and transient (GAL promoter) co-expression of cytoplasmic active CRISPR machinery with cytosolic yeast CHD1NES or GCN5NES. In (D) and (F), one-tailed t-tests are used; the ordinates are the mean depth of coverage over sixteen chromosomes (mitochondrial genome is omitted); n.s., no significance; *p < 0.05. All images represent one middle confocal Z-stack of 0.14 μm with nuclei stained by DAPI (blue). Scale bar: 2 μm.
Figure 4—figure supplement 1
Design of CRISPR experiments for targeting cytoplasmic eccDNA in yeast stress granule cores.
(A) Major regulatory and coding motifs of CRISPR elements in modified vector pML104. Bottom panel displays experimental variants of yeast transformants with CRISPR-pML104. NES, nuclear export signal; NLS, nuclear localization signal. (B) Transiently expressed Cas9NESmCherryNES localizes to cytosol together with PAB1-GFP (condition A, see (D), no stress). (C) Detection of gRNA in yeast cytoplasm using fluorophore TO1-Biotin and fluorogenic RNA aptamer Mango-III(A10U) (see Methods) encoded in HDV ribozyme at 5'-end of gRNA-2 (see (A)). The experiments were conducted with strain JD1370 free from endogenous or transient fluorescent fusions. (D) Growth steps of CRISPR-pML104 transformants with indication of major alterations in media, viz. changing D(+)-glucose to D(+)-galactose ratio to induce expression of Cas9 and addition of sodium azide to stress cells. Condition A: 0.2% (wt/vol) D-(+)-glucose and 2% (wt/vol) D-(+)-galactose for 4 hr. Condition B: 1.5% (wt/vol) D-(+)-glucose and 0.75% (wt/vol) D-(+)-galactose, overnight. In (B) and (C), Airyscan images of middle Z-stack plane (0.14 μm). Scale bar: 2 μm. Panels (A) and (D) created with BioRender.
© 2026, BioRender Inc. Parts of this image created with BioRender are made available under a Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Figure 4—figure supplement 2
Phenotypic changes of S. cerevisiae in response to transient expression of individual CRISPR elements.
(A–C) Confocal microscopy of S. cerevisiae with endogenous PAB1-GFP (green) and transiently expressed Cas9 (GAL promoter) or gRNAs (see Figure 4—figure supplement 1A). Treatment conditions and nuclear (NLS) or cytoplasmic (NES) tags are indicated. In (A), arrows (red) indicate accumulation of PAB1-GFP signal observed in the absence of stress. (D, E) Colocalization of PAB1-GFP with mitochondria in the absence of stress (framed in (B)). Upper images (488 and 405 nm channels) are split into individual channels to mark mitochondria (lower panels; arrows). All images represent one middle Airyscan confocal Z-stack of 0.14 µm with nuclei and mitochondria stained by DAPI (blue). Scale bars: 2 μm in (A–C) and 0.5 μm in (D) and (E).
Figure 4—figure supplement 3
Chromatin binders CHD1 and GCN5 compete out CRISPR and restore stress granule formation in cytoplasm.
(A) Schematics of pBY011 and pML104 main elements used to express chromatin remodeler CHD1 or histone acetyltransferase GCN5. (B) Transient co-expression of cytoplasmic CHD1 encoded into pBY011 backbone with cytoplasmic CRISPR preserves canonical stress response. (C) Transient co-expression of cytoplasmic GCN5 encoded into pBY011 backbone with cytoplasmic CRISPR preserves canonical stress response. (D) Transient co-expression of cytoplasmic CHD1 encoded into pML104 backbone with cytoplasmic CRISPR preserves canonical stress response. Right panels in (B–D), cells treated with 0.5% (wt/vol) sodium azide (Methods). All images represent one middle Airyscan confocal Z-stack of 0.14–0.17 μm with nuclei and mitochondria stained by DAPI (blue). Scale bar: 5 μm.
Figure 5
CRISPR-mediated suppression of stress granules compromises recovery from hypoxic stress.
(A) Recovery of CRISPR transformants after exposure to oxidative stress (45 min at 30°C) applied in the early log phase (Methods). Spotting assay was carried out under moderate Cas9NES induction on SD(-URA) solid medium with D(+)-glucose [1.5% (wt/vol)] and D(+)-galactose [0.75% (wt/vol)] using indicated 10-fold serial dilutions. Stress-induced phenotypes of the CRISPR variants with controls are presented in Figure 4, Figure 4—figure supplement 2. (B) Quantitation of post-stress recovery efficiency as in (A) for indicated CRISPR transformants. Spotting assays were carried out on solid medium at 30°C for 2.5 days as in (A) and quantified using the second (10–2) dilution. Data are mean for n = 3 ± s.d. and 44 technical replicates. Data were normalized to values of gRNAs alone; n.s., no significance; *p < 0.05; **p < 0.01; ****p < 0.0001 (one-way ANOVA followed by Tukey’s test with a 95% confidence interval). (C) Representative time-course for post-stress recovery of CRISPR transformants in the SD(-URA) liquid medium as in (A) (n = 2 ± s.d.; technical replicates). The growth was initiated with the second (10–2) dilution.
Author response image 1
Tables
Key resources table
| Reagent type (species) or resource | Designation | Source or reference | Identifiers | Additional information |
|---|---|---|---|---|
| Strain, strain background (S. cerevisiae) | BY4741 | https://www.yeastgenome.org | SGD:S000000967, GFP-clone | Systematic name: YER165W/Invitrogen, Yeast GFP Clone Collection |
| Strain, strain background (S. cerevisiae) | W303 | The Gunjan Laboratory | YAG1021 | |
| Genetic reagent (S. cerevisiae) | W303 | The Dinman Laboratory | JD1370 | |
| Gene (S. cerevisiae) | CHD1 | https://www.yeastgenome.org | SGD:S000000966 | Systematic name: YER164W |
| Gene (S. cerevisiae) | GCN5 | https://www.yeastgenome.org | SGD:S000003484 | Systematic name: YGR252W |
| Cell line (Homo sapiens) | HEK293T | Millipore Sigma | 96121229-1VL | |
| Gene (Homo sapiens) | AHNAK, MYC, MAPT | NCBI | Gene ID 79026, Gene ID: 4609, Gene ID: 4137 | |
| Biological sample (Homo sapiens) | Stress granule cores | This paper | Isolated from HEK293T cells, protocol: https://doi.org/10.1016/j.celrep.2025.115738 | |
| Biological sample (S. cerevisiae) | Stress granule cores | This paper | Isolated from yeast cells, protocol: https://doi.org/10.1016/j.celrep.2025.115738 | |
| Antibody | Anti-FLAG M2, mouse, monoclonal | Millipore Sigma | F1804 RRID:AB_262044 | |
| Antibody | 1.4 nm Nanogold-IgG, Goat anti-mouse | Nanoprobes | 2001 RRID:AB_2877644 | |
| Antibody | Anti-VCP, mouse, monoclonal | Santa Cruz Biotechnology | sc-57492 RRID:AB_793927 | (1:100), Supplementary file 1 |
| Antibody | Anti-histone H4, rabbit, monoclonal | Abcam | ab177840 RRID:AB_2650469 | (1:70–1:100), Supplementary file 1 |
| Antibody | Anti-histone H3.1, rabbit, monoclonal | Novus Biologicals | NBP3-26228 RRID:AB_3638417 | (1:90), Supplementary file 1 |
| Antibody | Anti-G3BP, mouse, monoclonal | Abcam | ab56574 RRID:AB_941699 | (1:200), Supplementary file 1 |
| Antibody | Anti Lamin A + Lamin C, rabbit, monoclonal | Abcam | ab108595 RRID:AB_10866185 | (1:400), Supplementary file 1 |
| Antibody | Anti-Caprin1, rabbit, polyclonal | Abcam | ab244360 | (1:400), Supplementary file 1 |
| Antibody | Anti-TOMM20, rabbit, monoclonal | Abcam | ab186735 RRID:AB_2889972 | (1:250), Supplementary file 1 |
| Antibody | Anti-DNA, mouse, monoclonal | Thermo Fisher Scientific | 690014S | (1:200), Supplementary file 1 |
| Antibody | Anti-mouse, donkey | Molecular Instruments, Inc | Used with B5-647 initiator, Supplementary file 1 | |
| Antibody | Anti-rabbit, donkey | Molecular Instruments, Inc | Used with B5-647 or B4-546 initiator, Supplementary file 1 | |
| Antibody | Anti-mouse, goat, Alexa Fluor 568 | Invitrogen | A11004 RRID:AB_2534072 | (1:800), Supplementary file 1 |
| Recombinant DNA reagent | pML104 (plasmid) | Addgene | 67638 RRID:Addgene_67638 | |
| Recombinant DNA reagent | pBY011 (plasmid) | DNASU | ScCD00095307 | Refer to Supplementary file 5 |
| Recombinant DNA reagent | pBY011 (plasmid) | DNASU | ScCD00011312 | Refer to Supplementary file 5 |
| Recombinant DNA reagent | pML104 (various modifications) | This paper | Refer to Supplementary files 3 and 5 | |
| Sequence-based reagent | Primers and fragments for modification of pML104 | This paper | Refer to Supplementary file 4 | |
| Commercial assay or kit | HCR RNA-FISH | Molecular Instruments, Inc | ||
| Commercial assay or kit | HCR IF | Molecular Instruments, Inc | ||
| Chemical compound, drug | Sodium azide | VWR Life Science | 0639-250G | |
| Chemical compound, drug | Sodium arsenite | LabChem | LC228709 | |
| Chemical compound, drug | Phenol solution, pH 8.0 | Sigma Life Science | P4557-400ML | |
| Software, algorithm | ZEISS ZEN Microscopy Software | https://www.zeiss.com | RRID:SCR_013672 | |
| Software, algorithm | Fiji (ImageJ) | https://www.fiji.sc | RRID:SCR_002285 | |
| Software, algorithm | GraphPad Prism | https://www.graphpad.com | RRID:SCR_002798 | |
| Software, algorithm | BioRender | BioRender, Toronto, Ontario, Canada | https://www.biorender.com/ | |
| Other | 5PRIME Phase Lock Gel Heavy | Quantabio | 2302830 | Extraction of total nucleic acids from stress granule cores |
| Other | 8 Well Chambered Cover Glass | Cellvis | C8-1.5H-N | Immobilization of HEK293T cells or yeast spheroplasts for light microscopy analysis |
| Other | Sephacryl S-500 HR, HiPrep 16/60 | Cytiva | 28935606 | https://doi.org/10.1016/j.celrep.2025.115738 |
| Other | Sephacryl S-1000 Superfine | Cytiva | 17-0476-01 | https://doi.org/10.1016/j.celrep.2025.115738 |
Additional files
-
Supplementary file 1
Antibodies used in IF and HCR IF light microscopy imaging of HEK293T cells.
- https://cdn.elifesciences.org/articles/111336/elife-111336-supp1-v1.docx
-
Supplementary file 2
Sequence Read Archive deposition summary for raw DNA sequences.
- https://cdn.elifesciences.org/articles/111336/elife-111336-supp2-v1.docx
-
Supplementary file 3
Cloning materials for CRISPR targeting cytoplasmic eccDNA with Ty1 in yeast.
- https://cdn.elifesciences.org/articles/111336/elife-111336-supp3-v1.docx
-
Supplementary file 4
Primers and fragments designed for modification of the pML104 backbone.
- https://cdn.elifesciences.org/articles/111336/elife-111336-supp4-v1.docx
-
Supplementary file 5
Addgene deposit IDs for plasmids used in the study.
- https://cdn.elifesciences.org/articles/111336/elife-111336-supp5-v1.docx
-
MDAR checklist
- https://cdn.elifesciences.org/articles/111336/elife-111336-mdarchecklist1-v1.docx
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Cytoplasmic circular dsDNA is a key constituent of stress granules
eLife 15:RP111336.
https://doi.org/10.7554/eLife.111336.2
