An aberrant phase transition of stress granules triggered by misfolded protein and prevented by chaperone function

Abstract Stress granules (SG) are membrane‐less compartments involved in regulating mRNAs during stress. Aberrant forms of SGs have been implicated in age‐related diseases, such as amyotrophic lateral sclerosis (ALS), but the molecular events triggering their formation are still unknown. Here, we find that misfolded proteins, such as ALS‐linked variants of SOD1, specifically accumulate and aggregate within SGs in human cells. This decreases the dynamics of SGs, changes SG composition, and triggers an aberrant liquid‐to‐solid transition of in vitro reconstituted compartments. We show that chaperone recruitment prevents the formation of aberrant SGs and promotes SG disassembly when the stress subsides. Moreover, we identify a backup system for SG clearance, which involves transport of aberrant SGs to the aggresome and their degradation by autophagy. Thus, cells employ a system of SG quality control to prevent accumulation of misfolded proteins and maintain the dynamic state of SGs, which may have relevance for ALS and related diseases.

Thank you for submitting your manuscript for consideration by The EMBO Journal and my apologies for the extended duration of the review period. Your manuscript has now been seen by three referees whose comments are shown below.
As you will see from the reports, referees #2 and #3 express high interest in the findings reported in your manuscript and recommend publication, following the inclusion of a number of additional control experiments and clarifications. At the same time referee #1 is concerned about the partial overlap with previously published work from both your lab and others, and consequently asks for a more extensive characterization of the difference between normal and aberrant stress granules.
Given the referees' overall positive recommendations, I would like to invite you to submit a revised version of the manuscript, addressing the comments of all three reviewers. In our view, the constructive suggestions and requests from refs #2 and #3 would also go a long way to address the major concern raised by ref #1. I should add that it is EMBO Journal policy to allow only a single round of revision, and acceptance or rejection of your manuscript will therefore depend on the completeness of your responses in this revised version.
For the revised manuscript I would particularly ask you to focus your efforts on the following points: -> Clarify all points related to experimental strategy, sample acquisition and quantification.
-> Expand data on the selective accumulation of mutant SOD1 in SGs as pointed out by refs #1 and #2 -> In our view the suggestions from ref #2 to use Super resolution microscopy for improved insights on the structure and dynamics of the SGs would clearly add to the manuscript; however, we also realize that this may be technically difficult to establish during revision. The manuscript by Mateju et al investigates the interaction between stress granule-associated proteins and misfolded proteins, mainly disease-linked mutants of SOD1. Stress granule resident proteins are hot spots for ALS-causing mutations, as is the SOD1 protein. The authors set out to investigate the mechanism of converting physiological SGs into pathological entities. The groups that collaborated in the study have authored a number of important papers on this topic in the past. This work is a continuation of a recent paper in Molecular Cell, where the authors propose that a chaperone network mediates SG disassembly.
The proposed novelty of this manuscript rests almost entirely on the interpretation of colocalization dynamics of SG proteins and SOD mutants. Otherwise, many of the observations are somewhat similar to previous reports (for example Farrawell et al., 2015 -showing that mutant and wt FUS and TDP-43 partition to different kinds of inclusions, some of which can recruit wt SG components. Also, the idea that SOD mutants can co-aggregate with less aggregation-prone proteins and substantially decrease their mobility by titrating Hsp70 and other proteostasis components is also well established. In the view of this reviewer the novelty of this paper hinges on proving that there is a mechanistic distinction between what the authors call normal and aberrant SGs. SGs are an under-defined entity, with potentially aggregation-prone and (albeit slightly) overexpressed proteins being used to mark them. Without a more extensive characteriazation of normal versus aberrant SGs it is very difficult to be convinced that there is a distinction between SOD mutants "infecting" normal SGs and the opposite scenario, namely SOD mutants aggregating and inducing the co-aggregation of metastable SG proteins. Major points: 1. Figure EV2D does not show the SGs and aggresome precursors simultaneously. The data only argue that ubc aggregates G3BP is already out of visible puncta. 2. The authors argue that in their system SG proteins are expressed at physiological levels. However, it's not endogenous proteins that are being tagged here -there is at least a 2-fold overexpression. For metastable aggregation prone proteins this may influence their aggregation kinetics. 3. Page 7 line 22 is not clear at all. How is the data showing that "recruitment to SGs is specific for misfolded variants of SOD1" when all SOD variants accumulate? 4. Page 8, line 9 -Fluorescence microscopy does not provide enough information to conclude that SGs are completely devoid of mutant SOD.

Referee #2:
Mateju and colleagues present an interesting manuscript that contains experimentally robust and thoughtful data using a number of technologically impressive techniques. They describe a novel mechanism by which misfolded proteins are enriched in cytoplasmic compartments known as a stress granules (SGs) prior to clearance by two independent mechanisms, involving chaperones and autophagy. This work is consistent with a number of trending subjects within the fields of cellular bodies and stress responses (concentration-dependent phase transitions and autophagy, respectively.) The experiments are well described and logical, which is emphasized by the well-written manuscript. I have some small comments and suggestions to improve the manuscript, but strongly believe that this is suitable for publication in the EMBO Journal. Figure 1: 1) Do Ubc9s or SOD1 mutants form SG-like structures in vitro in the absence of FUS or other in vitro nucleators of body formation (such as TDP-43, which the authors identified as decreased in SG localization following prolonged heat shock)? 2) Regarding the high throughput experiments, how did the authors decide upon the signal intensity threshold, and what was it? 3) In later figures, the authors state that there was a mostly binary accumulation of misfolded proteins in SGs. 4) Please provide cell-level % in addition to SG-level analysis. 5) Did the authors observe a change in SG number under different conditions? This is especially relevant for later figures, where a change in protein composition was reported over time. Where these changes in total SG composition reflective of protein relocalization/degradation or the formation of new SGs? 6) Throughout all high throughput imaging experiments, can the authors correlate protein expression with SOD1+ SG appearance (or other factors, such as SG size or cell cycle)? 7) There is a typo on page 7, line 21 - Figure 2G should be Figure 1G. 8) The authors may want to consider using another method to induce protein misfolding or SG assembly (such as ....) to show that this is a universal phenomenon. Do misfolded SOD1 proteins accumulate naturally in SGs in laminopathy cell models, as a model system to study ageing? Figure 2 9) The authors state that mutant SOD1 is aggregated and accumulated in SGs but provide no direct evidence for this. 10) 3D super resolution microscopy (rather than the 2D projected images shown) that are modeled using software such as Imaris (Bitplane) may indicate whether misfolded protein localization to the SG is a shell or more dispersed throughout the whole structure. 11) The manuscript will be vastly improved by a temporal dissection of SG ultrastructure using structural illumination microscopy (SIM). Is there a change in misfolded protein localization that correlates with the suggested decrease in RBP SG localization? Figure 3 12) Please provide mean square displacement (MSD, um2) data to support changes in SG movement. Figure 4 13) Did the authors observe any changes in SG size or number with composition? 14) 4C -please try to standardize the axes for reader comprehension. 15) EV4B -Were these changes due to protein degradation, relocalization away from the SG, or the creation of new SGs at these time points? Thus, was the number of DRiP+ SGs stable, but more SGs were assembled that don't contain DRiPS? Figure 5 16) Does HSP70 regulate accumulation of misfolded proteins in SGs in vitro? Figure 6 17) Do other PI3K inhibitors (LY294002 etc) also increase SG-aggresome association?
Referee #3: In this paper the authors investigate how unstable and aggregation prone proteins (mutants of Ubc and SOD1) interact with stress granules (bodies positive for FUS or G3BPs) and change their physical properties. They find that stress granules containing additional protein aggregates mature over time, and that the aggregated proteins somehow displace the initial SG components, leading to less dynamic and more solid-like cellular inclusions. The authors then go on to investigate the impact of heat-shock proteins 70 and 27 on SGs, and find that HSP presence correlates with that of aggregated proteins. Interestingly, an inhibitor of hsp70 manipulates the aggregates in a predictable way. Finally, the authors link hard-to-clear SGs with the aggresome, suggesting that cells possess multiple pathways for dealing with stress, and clearing its causalities.
Overall the paper presents a very interesting set of experiments, analysing the relationship between aggregating proteins, chaperones and stress granules. This reviewer recommends this paper for publication, with the following minor points for consideration.
Minor points 1. Figure 1A/B "Interestingly, misfolded Ubc9TS accumulated in FUS compartments while Ubc9WT remained largely in solution ( Fig 1A). This suggests that misfolded proteins may have a tendency to accumulate in phase-separated liquid compartments." From the images in the figure, when I manually adjust the contrast, it looks like there is zero Ubc9WT present in ∅ or WT panels, either in solution or in FUS compartments. It's not clear form these images that the protein has remained in solution. Perhaps it would be clearest if the change in partitioning were quantified as they do later in the paper? The change in partitioning looks like it's rather large.
2."9% of SGs were highly enriched for Ubc9TS". Presumably this is because of heterogeneity in the SGs rather than mass action? I think it would be interesting if this point were discussed further.
Is this (and other comparisons coming later in the manuscript) at the per-cell level or over a large population of cells? Is the high-content imaging data normalised for protein expression level? Also, what are the properties of this population of SGs -size, shape, number distribution compared to the remaining 91%? Presumably this data is accessible from the high-content automated imaging assay? In short, is it obvious why some of the SGs are enriched, but others are not?
3. What are the errors associated with the intensity ratios in Fig   Reply: We agree with the reviewer that SGs are an underdefined entity. For this reason, we are very careful to make sure that we study SGs and no other aggregate or assembly. This is also why we use multiple markers for SGs, such as FUS, G3BP, eIF3 (Figs 1E, 1F, Appendix Fig S2A). In some experiments, we had to rely on GFP-tagged SG proteins for live cell imaging, but in others we used immunofluorescence using cells that do not overexpress SG markers (Figs 4A, Appendix Fig S2A).
Our data are very clear: aberrant SGs that contain misfolded proteins also contain all the SG markers mentioned above (Figs 1E, 1F, Appendix Fig S2A). To address the reviewer's concern further, we performed fluorescence in situ hybridization, which revealed that both normal SGs and aberrant SGs contain poly(A) mRNA in similar amounts (Fig EV3A). In this specific experiment, no SG marker was overexpressed and SGs were detected by immunofluorescence. Thus, we conclude that aberrant SGs are indeed SGs: they contain several key SG marker proteins and have a high content of poly(A) mRNA.
The reviewer suggests that the opposite scenario could also be true, namely that SOD1 aggregates sequester and induce the aggregation of SG proteins. This was also one of the concerns that we had initially, but we could rule this out by performing a time-resolved analysis of SG formation and maturation. In time-lapse experiments with live cells, we observed SG formation before SOD1 formed visible aggregates (Figs 2A, 2B). In addition, we could directly observe the conversion of physiological SGs into aberrant SGs in some time-lapse movies (Movie EV5). Moreover, using FRAP, we find that SOD1 becomes immobilized only after prolonged heat stress, when SGs are already present (Figs 2C, 2D). Based on these findings, we conclude that SOD1 is recruited to SGs and not vice versa.
We agree with the reviewer that it is important to show a clear difference between normal and aberrant SGs. In order to provide such evidence, we investigated whether the structural integrity of aberrant SGs depends on RNA. The rationale was the following: physiological SGs are dynamic membrane-less compartments, whose formation depends on free mRNA (Kedersha et al, 2000;Bounedjah et al, 2014). However, aberrant SGs increasingly behave as protein aggregates with reduced dynamics, suggesting that they are no longer dependent on RNA. To test this hypothesis, we microinjected RNase A into heat-stressed cells that contained visible SGs. As expected, RNase injection triggered the rapid disintegration of SOD1-negative SGs. Remarkably, however, the same treatment had no visible effect on SOD1-positive SGs (Figs 3G, H). Based on these data, we conclude that aberrant SGs are protein-based assemblies, whereas physiological SGs are RNA-based assemblies. This suggests that the conversion into aberrant SGs involves a gradual replacement of physiological RNA-protein interactions with aberrant protein-protein interactions. It is easy to understand that this affects the dynamics of SGs. It is also clear that this creates a strong dependence on chaperones for disassembly, as observed in Figure 5C and D. We think that this experiment adds important weight to our study. We have therefore added these data to Figure 3.
Reviewer 1: Major points: 1. Figure EV2D does not show the SGs and aggresome precursors simultaneously. The data only argue that ubc aggregates G3BP is already out of visible puncta.
Reply: We agree. The simultaneous presence of SGs and aggresome precursors was visible only in a short time window in this experiment (between 4 and 6 h of MG132 treatment). For this reason, we added a new time point (Fig EV2D is now Fig EV1D). The new panel (5 h) shows a cell in which both SGs and aggresome precursors are visible but are not colocalizing. This can also be seen in Movie EV1.
Reviewer 1: 2. The authors argue that in their system SG proteins are expressed at physiological levels. However, it's not endogenous proteins that are being tagged here -there is at least a 2-fold over-expression. For metastable aggregation prone proteins this may influence their aggregation kinetics.
Reply: Because BAC transgenes have all the endogenous promoter and enhancer elements, they autoregulate their expression levels, and the amount of protein produced usually only deviates slightly from normal levels, if at all. To estimate the expression level of the BAC transgenes, we included an immunoblot. Based on these data, the expression level does not seem to be higher than 2-fold overexpression (Fig EV1A). To make sure that this slight increase in expression levels does not affect the aggregation propensity of the proteins, we generated different cell lines in which we tagged FUS or G3BP with different fluorophores (GFP or mCherry). The colocalization with misfolded proteins was not different between these cell lines (Figs 1B, C, E, F). We also performed experiments, where we detected endogenous SG markers using immunofluorescence. For these cells, we observed a similar number of aberrant SGs as in the presence of BAC transgenes. We therefore conclude that the presence of the transgenes does not affect protein aggregation. Reply: We consistently saw that mutant SOD1 variants were much more strongly enriched in SGs than wild type SOD1 (Fig 1G). All the mutant SOD1 variants that we tested (A4V, G37R, G85R, G93A) have been shown to be prone to misfolding and aggregation, in contrast to wild type SOD1 (Rakhit et al, 2007;Prudencio et al, 2009). We made changes to the text to make this more clear. Reply: We agree that "devoid" is an inaccurate term. We have changed the text to address this.

Reviewer 2: Mateju and colleagues present an interesting manuscript that contains experimentally robust and thoughtful data using a number of technologically impressive techniques. They describe a novel mechanism by which misfolded proteins are enriched in cytoplasmic compartments known as a stress granules (SGs) prior to clearance by two independent mechanisms, involving chaperones and autophagy. This work is consistent with a number of trending subjects within the fields of cellular bodies and stress responses (concentration-dependent phase transitions and autophagy, respectively.) The experiments are well described and logical, which is emphasized by the wellwritten manuscript. I have some small comments and suggestions to improve the manuscript, but strongly believe that this is suitable for publication in the EMBO Journal.
Reply: We thank the reviewer for the positive feedback.

) Do Ubc9s or SOD1 mutants form SG-like structures in vitro in the absence of FUS or other in vitro nucleators of body formation (such as TDP-43, which the authors identified as decreased in SG localization following prolonged heat shock)?
In the absence of FUS, purified Ubc9TS remains in solution and does not form droplets. We now include this data in Appendix Figure S1D.

Reviewer 2: 2) Regarding the high throughput experiments, how did the authors decide upon the signal intensity threshold, and what was it?
Reply: We chose an intensity ratio above which the accumulation of proteins in SGs is clearly visible by eye. As specified in the figure legends, we used a threshold of 1.4 for Ubc9 and SOD1.

Reviewer 2: 3) In later figures, the authors state that there was a mostly binary accumulation of misfolded proteins in SGs.
Reply: In reality, the accumulation of misfolded proteins in SGs was not binary, but there were gradual differences in the amount of accumulated misfolded protein in SGs. However, in order to see clear functional differences, we focussed on the extreme cases, rather than on SGs with an intermediate accumulation of SOD1. This was necessary because SGs with intermediate levels of SOD1 showed large variability in their properties and dynamics. We made changes to the text to make this more clear.

Reviewer 2: 4) Please provide cell-level % in addition to SG-level analysis.
Reply: We now provide these data in the Appendix Figure S2C, D. This new analysis shows the fraction of cells (in percent) containing aberrant SGs enriched for SOD1(A4V) or Ubc9TS. For cells that expressed misfolding-prone instead of wild type control proteins, there was a much higher probability that they contained one or more aberrant SGs per cell.

Reviewer 2: 5) Did the authors observe a change in SG number under different conditions? This is especially relevant for later figures, where a change in protein composition was reported over time.
Where these changes in total SG composition reflective of protein relocalization/degradation or the formation of new SGs?

Reply:
We do not think that the changes at later time points occur because of the formation of new SGs, but rather reflect the redistribution of proteins from/to SGs. In our time-lapse imaging experiments, SGs only formed in the first 60 minutes after stress onset. During prolonged stress (60-150 minutes), formation of new SGs was a very rare event. To test this further, we quantified the number of SGs per cell over time. This revealed that the SG number per cell was slightly decreasing after 60 minutes of heat stress, which is probably due to SG fusion (see Figure R1 below). Therefore, we can exclude that the formation of new SGs during prolonged stress affects SG composition.

Reviewer 2: 6) Throughout all high throughput imaging experiments, can the authors correlate protein expression with SOD1+ SG appearance (or other factors, such as SG size or cell cycle)?
Reply: We did not see any obvious correlations with protein expression or cell cycle, but this data cannot easily be extracted from our high-content imaging data. To gain more insight, we analyzed SG size and circularity (Appendix Figure S5). There is no obvious association between SG circularity and enrichment of SOD1(A4V) or Ubc9TS in SGs. Interestingly, however, we find that SOD1-positive SGs (and Ubc9TS-positive SGs) were on average slightly larger, but this difference was not statistically significant. We speculate that this could be due to the fact that aberrant SGs are less dynamic and thus become larger with time relative to physiological SGs.

Reviewer 2: 7)
There is a typo on page 7, line 21 - Figure 2G should be Figure 1G.
Reply: Thank you, we changed this. Reply: This is a very good idea and something that we will definitely test in the future. However, we feel that for this study, these experiments are beyond the scope. We would like to mention though that an increasing number of aberrant SGs have very recently been reported for aging C. elegans (Lechler et al, 2017), suggesting that there may be an interesting connection between SG homeostasis and aging.
Reviewer 2: Figure 2, 9) The authors state that mutant SOD1 is aggregated and accumulated in SGs but provide no direct evidence for this.
Reply: Figures 2A and 2B show the accumulation of mutant SOD1 in SGs. The evidence for aggregation comes from the FRAP experiments in Figures 2C-G and from super-resolution microscopy in Figures 2H, 2I. In the revised manuscript, we now provide additional data showing that aberrant SGs have properties of protein aggregates (Fig 3G). We find that aberrant SGs become independent of RNA, as they remain intact after degradation of RNA by RNase A microinjection. This is consistent with the idea that SGs convert from an RNA-based assembly into protein-based aggregates.
Reviewer 2: 10) 3D super resolution microscopy (rather than the 2D projected images shown) that are modeled using software such as Imaris (Bitplane) may indicate whether misfolded protein localization to the SG is a shell or more dispersed throughout the whole structure. Figure R3. Effect of LY294002 on SG-aggresome association. Cells expressing G3BP2-GFP and Ubc9TS-mCherry were treated with 50 µM LY294002 (Cell Signaling) or DMSO during heat stress (2 h) and subsequent recovery (6 h). After fixation, at least 100 cells expressing Ubc9TS were examined to determine the localization of SGs at the aggresome region. Mean values from 3 experiments are shown. Error bars = SEM. * p < 0.05 (t-test).

Reviewer 3: In this paper the authors investigate how unstable and aggregation prone proteins (mutants of Ubc and SOD1) interact with stress granules (bodies positive for FUS or G3BPs) and change their physical properties. They find that stress granules containing additional protein aggregates mature over time, and that the aggregated proteins somehow displace the initial SG components, leading to less dynamic and more solid-like cellular inclusions. The authors then go on to investigate the impact of heat-shock proteins 70 and 27 on SGs, and find that HSP presence correlates with that of aggregated proteins. Interestingly, an inhibitor of hsp70 manipulates the aggregates in a predictable way. Finally, the authors link hard-to-clear SGs with the aggresome, suggesting that cells possess multiple pathways for dealing with stress, and clearing its causalities.
Overall the paper presents a very interesting set of experiments, analysing the relationship between aggregating proteins, chaperones and stress granules. This reviewer recommends this paper for publication, with the following minor points for consideration.

Reply:
We thank the reviewer for the positive feedback.

Reviewer 3:
Minor points 1. Figure 1A/B "Interestingly, misfolded Ubc9TS accumulated in FUS compartments while Ubc9WT remained largely in solution (Fig 1A). This suggests that misfolded proteins may have a tendency to accumulate in phase-separated liquid compartments." From the images in the figure, when I manually adjust the contrast, it looks like there is zero Ubc9WT present in or WT panels, either in solution or in FUS compartments. It's not clear form these images that the protein has remained in solution. Perhaps it would be clearest if the change in partitioning were quantified as they do later in the paper? The change in partitioning looks like it's rather large.

Reply:
We added a quantification of Ubc9 accumulation in FUS droplets (Appendix Figure S1C). We also changed the contrast settings for the images, but the Ubc9WT signal is still much less bright than Ubc9TS. The reason for this is that Ubc9WT is diluted in solution and not enriched in foci as Ubc9TS. Reply: Indeed, enrichment of misfolded proteins in SGs is due to the heterogeneity of SGs, largely between cells, which may be due to the health or age of these cells. However, we not only observed variability between cells, but also identified cells that contained both physiological and aberrant SGs at the same time, which argues against a simple mass action mechanism. In addition, aberrant SGs also formed in the absence of overexpressed proteins (Fig 5B), further suggesting that aberrant SGs are not an artifact of protein overexpression. Generally, a correlation with expression levels was not apparent from our data.
Our high-content imaging assay analyzes a large number of individual SGs over a large population of cells. In our assay, the fluorescence intensity is normalized to the area around SG, which reflects cytosolic levels. In the revised manuscript, we have added additional cell-level analyses (Appendix Figs S2C, D). To gain more insight, we also analysed SG size, shape and number (Appendix Figure  S5). SG circularity did not correlate with the enrichment of SOD1(A4V) or Ubc9TS in SGs. Interestingly, SGs enriched for SOD1(A4V) or Ubc9TS were on average slightly larger and they were more prevalent in cells with a lower number of SGs. However, these were relatively small differences that cannot explain the observed variability in SG composition and other observations, such as decreased G3BP1 turnover (Fig 3E), reduced dynamics ( Fig 3D) or RNase insensitivity ( Fig  3H). Fig 4B? Reply: The enrichment ratios were calculated from the mean fluorescence intensities within the indicated ROIs, which were automatically segmented based on the signal of the SG marker. The whole distribution of the ratios can be seen in Figure 4C. We verified that our assay does not detect other structures than SGs. However, not all SGs in a sample could be detected, due to occasional autofocus errors or highly irregular (fuzzy) edges of some SGs. Some errors might be created due to SGs being closely adjoined. In general, the assay was robust and the high enrichment ratios reflected an accumulation that was visible by eye. Reply: In Figure 5B, we do not use heat stress and we compare four different conditions. The Ars+DMSO condition is the only conditions in which proteostasis is unimpaired and ubiquitinated proteins can be efficiently degraded. The other three conditions all involve some form of proteostasis inhibition (+ VER or + MG132 or both). We only observe a strong depletion of ubiquitin signal in the nucleus in the Ars+DMSO condition with no proteostasis impairment. Instead, when proteostasis is impaired (+VER or +MG132), ubiquitin is enriched in the nucleus. The strongest ubiquitin enrichment is seen, when both MG132 and VER are added to cells. This is in agreement with recent findings that ubiquitylated proteins are degraded by nuclear proteasomes (Park et al, 2013), and suggests that upon inhibition of the proteostasis machinery substrates of the ubiquitin/proteasome system accumulate in the nucleus. We have added a sentence to the figure legend to explain the difference in ubiquitin staining.

Reviewer 3: 3. What are the errors associated with the intensity ratios in
In Figure 5B, the accumulation of ubiquitin in SGs is only visible in the condition where VER is present. The other two conditions do not show a significant enrichment of ubiquitin. The different distribution of ubiquitin compared to Figure EV5B (previously EV6) is because of the different conditions used. In Figure EV5B we used heat stress, which in our experience causes more protein misfolding and thus stronger recruitment of ubiquitin to SGs. Additionally, the co-expression of Ubc9TS might have further enhanced this effect. Thus, we do not think that there is a problem with the specificity of the antibody, but that the differences in staining can be explained by the different conditions used. Reply: To address this concern, we performed fluorescence in situ hybridization on heat-stressed cells expressing SOD1(A4V)-GFP. We observed a strong poly(A) RNA signal in both SOD1negative and SOD1-positive SGs, without any noticeable difference in the RNA content. This suggests that aberrant SGs still contain mRNA. In the revised version of the manuscript, we also include new data showing that the structural integrity of aberrant SGs does not depend on RNA. To investigate the role of RNA in SG assembly, we microinjected RNase A into heat-stressed cells to degrade RNA. RNA degradation dissolved SOD1-negative SGs, but had minimal effect on SOD1positive SGs (Figs 3G, H). Therefore, RNA is still a component of aberrant SGs, but does not seem to be required for the structural integrity of these SGs. We thank the reviewer for asking us to perform these additional experiments, which we think have significantly improved the paper.
For animal studies, include a statement about randomization even if no randomization was used.
4.a. Were any steps taken to minimize the effects of subjective bias during group allocation or/and when assessing results (e.g. blinding of the investigator)? If yes please describe. Do the data meet the assumptions of the tests (e.g., normal distribution)? Describe any methods used to assess it.
Is there an estimate of variation within each group of data?
Is the variance similar between the groups that are being statistically compared?
Statistical tests were applied where appropriate.
We did not test for normal distribution. We use either paired t--test (where appropriate) or Welch's t--test without the assumption of equal variance.
Yes, we show SEM or SD (specified in figure legend) in all experiments with multiple replicates.
We do not use the Student's t--test with assumption of equal variance.

YOU MUST COMPLETE ALL CELLS WITH A PINK BACKGROUND ê
The sample size is 3 or higher (specified in figure legends), based on estimated variance in the sample. 1. Data the data were obtained and processed according to the field's best practice and are presented to reflect the results of the experiments in an accurate and unbiased manner. figure panels include only data points, measurements or observations that can be compared to each other in a scientifically meaningful way. graphs include clearly labeled error bars for independent experiments and sample sizes. Unless justified, error bars should not be shown for technical replicates. if n< 5, the individual data points from each experiment should be plotted and any statistical test employed should be justified Please fill out these boxes ê (Do not worry if you cannot see all your text once you press return) a specification of the experimental system investigated (eg cell line, species name).

B--Statistics and general methods
the assay(s) and method(s) used to carry out the reported observations and measurements an explicit mention of the biological and chemical entity(ies) that are being measured. an explicit mention of the biological and chemical entity(ies) that are altered/varied/perturbed in a controlled manner.
the exact sample size (n) for each experimental group/condition, given as a number, not a range; a description of the sample collection allowing the reader to understand whether the samples represent technical or biological replicates (including how many animals, litters, cultures, etc.).
Each figure caption should contain the following information, for each panel where they are relevant:

Captions
The data shown in figures should satisfy the following conditions: Source Data should be included to report the data underlying graphs. Please follow the guidelines set out in the author ship guidelines on Data Presentation. a statement of how many times the experiment shown was independently replicated in the laboratory.
Any descriptions too long for the figure legend should be included in the methods section and/or with the source data.
Please ensure that the answers to the following questions are reported in the manuscript itself. We encourage you to include a specific subsection in the methods section for statistics, reagents, animal models and human subjects. This checklist is used to ensure good reporting standards and to improve the reproducibility of published results. These guidelines are consistent with the Principles and Guidelines for Reporting Preclinical Research issued by the NIH in 2014. Please follow the journal's authorship guidelines in preparing your manuscript.