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.

Keywords: SOD1; protein aggregation; protein misfolding; proteostasis; stress granules.

Figure 1. SGs co‐assemble with misfolded proteins including ALS‐associated SOD1

1. Purified Ubc9TS accumulates in liquid compartments formed by 5 μM FUS(G156E)‐GFP in vitro. The Ubc9 concentration is at 4 μM.

2. Localization of Ubc9TS‐mCherry or Ubc9WT‐mCherry co‐expressed with FUS‐GFP in live HeLa cells exposed to heat stress (2 h at 43°C).

3. Ubc9TS‐positive SGs contain G3BP2‐GFP.

4. Fraction of SGs colocalizing with Ubc9TS‐mCherry or Ubc9WT‐mCherry in cells expressing G3BP2‐GFP and heat‐stressed for 2 h. A total of 11,794 SGs were analyzed. Colocalization is defined by fluorescent ratio > 1.4. Mean values from three experiments are plotted. Error bars = SEM. **P < 0.01 (t‐test, compared to WT).

5. Localization of SOD1(A4V)‐GFP or SOD1(WT)‐GFP co‐expressed with FUS‐mCherry in live cells exposed to heat stress (2 h at 43°C).

6. SOD1‐positive SGs contain G3BP1‐mCherry.

7. Fraction of SGs colocalizing with different variants of SOD1‐GFP in cells expressing FUS‐mCherry and heat‐stressed for 2 h. A total of 12,028 SGs were analyzed. Error bars = SEM. *P < 0.05, **P < 0.01 (t‐test, compared to WT).

Figure EV1. Characterization of HeLa cells expressing SG markers and mutant misfolding‐prone proteins

1. HeLa stable cell lines expressing BAC‐encoded FUS‐mCherry or FUS‐GFP were analyzed by Western blotting using anti‐FUS antibody. Positions of endogenous FUS and the transgenes are indicated.

2. HeLa cells expressing BAC‐encoded SG markers (FUS‐mCherry, G3BP1‐mCherry, FUS‐GFP, or G3BP2‐GFP) were transfected with plasmids expressing SOD1(A4V)‐GFP or Ubc9TS‐mCherry. Live cells were imaged under normal conditions (37°C, no treatment). FUS is localized in the nucleus, and G3BP is diffusely distributed in the cytoplasm. Misfolding‐prone proteins Ubc9TS and SOD1(A4V) are also diffusely distributed. Scale bar = 10 μm.

3. HeLa cells expressing SOD1(A4V)‐GFP, SOD1(WT)‐GFP, Ubc9TS‐mCherry, or Ubc9WT‐mCherry were treated with MG132 (10 μM) for 14 h. Cells were fixed and stained for p62. p62 localized to large perinuclear inclusions (arrows) together with misfolded proteins but not WT variants of SOD1 or Ubc9. Scale bar = 10 μm.

4. HeLa cells expressing G3BP2‐GFP and Ubc9TS‐mCherry were imaged during the treatment with MG132 (10 μM). G3BP2 assembled into SGs (arrowheads), Ubc9TS accumulated over time and localized to aggresome (arrows). Scale bar = 10 μm. Images are from Movie EV1.

Figure 2. Misfolded proteins form aggregates after accumulating in SGs

1. Recruitment of FUS‐mCherry and SOD1(A4V)‐GFP into SGs in living cells exposed to heat stress.

2. Quantification of SOD1(A4V) enrichment in the SGs from (A) (intensity in SG/intensity outside SG) at different time points of heat stress. Mean values are shown (13–27 SGs in each frame). Error bars = standard deviation.

3. FRAP analysis of SOD1(A4V)‐GFP in SGs after 40, 80, or 120 min of heat stress. Mean values are shown (8–13 SGs per condition).

4. Mobile fraction of SOD1(A4V) calculated from the FRAP analysis in (C). Error bars = SEM (8–13 SGs per condition).

5. SOD1‐positive SGs were induced with heat stress (2 h) and photobleached (arrows) in two channels. Fluorescence recovery of FUS‐mCherry, but not SOD1(A4V)‐GFP, was observed.

6. Two‐channel FRAP analysis of FUS‐mCherry and SOD1(A4V)‐GFP in SGs following a 2‐h heat stress. Mean values are shown (5 SGs).

7. Two‐channel FRAP analysis of G3BP1‐mCherry and SOD1(A4V)‐GFP in SGs following a 2‐h heat stress. Mean values are shown (12 SGs).

8. Super‐resolution image of a SG containing FUS‐mCherry and SOD1(A4V)‐GFP following a 2‐h heat stress.

9. Super‐resolution image of a SG containing G3BP1‐mCherry and SOD1(A4V)‐GFP following a 2‐h heat stress.

Figure EV2. Misfolded proteins accumulate in SGs and form immobile domains with prolonged heat stress

1. Quantification of relative FUS enrichment (intensity in SG/intensity outside SG) in the SGs from Fig 2A at different time points of heat stress. Mean values are shown (13–27 SGs in each frame). Error bars = standard deviation.

2. HeLa cells expressing G3BP1‐mCherry and SOD1(A4V)‐GFP were imaged at 42°C. SG formation was followed by accumulation of SOD1 in SGs. G3BP1 was partially depleted from SGs with prolonged stress.

3. Super‐resolution image of a SG after 30 min of heat stress. Cells are expressing FUS‐mCherry and SOD1(A4V)‐GFP.

4. Super‐resolution image of a SG after 2 h of heat stress. Cells are expressing FUS‐mCherry and SOD1(A4V)‐GFP.

Figure EV3. Properties of SOD1‐positive and SOD1‐negative SGs

1. HeLa cells expressing SOD1(A4V)‐GFP were heat‐stressed for 2 h and fixed. G3BP1 was then detected by immunofluorescence, and poly(A) RNA was detected by fluorescence in situ hybridization. In the same sample, some SGs were clearly enriched for SOD1(A4V) (lower cell), while other SGs were not (upper cell). Both types of SGs contained poly(A) mRNA signal.

2. HeLa cells expressing FUS‐mCherry and SOD1(A4V)‐GFP were heat‐stressed for 2 h and then imaged at 37°C (time indicates duration of recovery). SOD1‐negative SGs showed fusion (arrows) and fission (arrowheads).

3. Fusion of SGs from the cell shown in (B).

4. Fission of a SG from the cell shown in (B).

5. In other cells treated the same way as in (B), SOD1‐positive SGs (arrows) showed less dynamic behavior.

Figure 3. SGs that accumulate misfolded proteins show aberrant behavior

1. Prevalence of SG fusion in cells with SOD1‐negative SGs (SOD1⁻) and cells with SOD1‐positive SGs (SOD1⁺) during 2‐h recovery from heat stress (2 h). Cells express FUS‐mCherry and SOD1(A4V)‐GFP. Only cells with SGs persisting for 2 h were analyzed. Average from five experiments is plotted. Error bars = SEM. **P < 0.01 (t‐test).

2. Prevalence of SG fission in cells with SOD1‐negative SGs (SOD1⁻) and cells with SOD1‐positive SGs (SOD1⁺) during 2‐h recovery from heat stress. Average from five experiments is plotted. Error bars = SEM. **P < 0.01 (t‐test).

3. A cell containing two populations of SGs, SOD1‐negative (arrowheads) and SOD1‐positive (arrows). Time indicates duration of recovery from heat stress (2 h). See also Movie EV3.

4. SGs from the representative cell shown in (C). SOD1‐negative SGs undergo fusions and change shape. SOD1‐positive SGs are less dynamic. Scale bar = 2 μm.

5. FRAP analysis of G3BP1‐mCherry in cells co‐expressing SOD1(A4V)‐GFP. Following a 2‐h heat stress treatment, G3BP1 was photobleached in SOD1‐negative SGs or SOD1‐positive SGs in the same sample. Mean values from ≥ 10 SGs for each category are plotted.

6. Mobile fraction of G3BP1 in SGs calculated from the FRAP analysis in (E). Error bars = SEM. **P < 0.01 (t‐test).

7. Cells expressing G3BP1‐mCherry and SOD1(A4V)‐GFP were heat‐stressed for 2 h. RNase A was then microinjected into cells with SOD1‐negative SGs (upper panel) or cells with SOD1‐positive SGs (lower panel) in the same sample. Images were taken immediately before injection and after injection (+RNase). Arrows indicate SGs.

8. Quantification of SG area remaining after RNase microinjection (four cells from each category). Error bars = SEM. **P < 0.01 (t‐test).

9. Effect of misfolded proteins on in vitro reconstituted FUS compartments. FUS(G156E)‐GFP was incubated either alone (control) or with purified Ubc9WT or Ubc9TS for the indicated time. In control samples, FUS phase‐separated into droplets wetting the surface. In samples containing Ubc9TS, morphologically distinct particles with emanating fibers were prevalent (arrows).

Figure 4. Chaperones are specifically recruited into SOD1‐positive SGs

1. HeLa cells expressing SOD1(G93A)‐GFP were heat‐stressed for 2 h, fixed, and stained with antibodies against eIF3η, poly‐ubiquitin, HSP27, HSP70, or VCP.

2. Examples from the imaging assay used to quantify protein enrichment in SGs. Automatic segmentation of SGs is based on FUS‐mCherry signal. High enrichment of SOD1 and HSP27 in one SG is reflected by high fluorescence ratios (upper panel). Low intragranular levels of SOD1 and HSP27 are reflected by low fluorescence ratios (lower panel). In this cell, SOD1 and HSP27 localize to different foci (arrows) instead of accumulating in SGs. Note that SGs are always defined by the presence of FUS.

3. Variability in SG composition. HeLa cells expressing FUS‐mCherry and SOD1(A4V)‐GFP were heat‐stressed for 2 h, fixed, and stained with antibodies against polyubiquitin, HSP27 or TDP‐43. An automated imaging assay was used to quantify the enrichment of proteins in SGs (1,000 SGs plotted). The Pearson's correlation coefficient (r) is shown.

4. Temporal changes in SG composition. HeLa cells expressing FUS‐mCherry were heat‐stressed (HS) for 30, 60, 90, 120, or 150 min, fixed, and stained with antibodies against poly‐ubiquitin, HSP27 or TDP‐43. An automated imaging assay was used to quantify the percentage of SGs highly enriched for ubiquitin, HSP27 or TDP‐43, at given time points (using a fluorescent ratio 1.5 as threshold). Mean values from three independent experiments are shown, each sample (one replicate of one time point) with > 400 SGs (average 2,836). Error bars = SEM.

Figure 5. HSP70 prevents formation of aberrant SGs and promotes SG disassembly

1. Fraction of SGs enriched for SOD1(A4V), Ubc9TS, or poly‐ubiquitin. Cells were heat‐stressed for 2 h in the presence of DMSO or 40 μM VER followed by fixation. SOD1(A4V)‐GFP was co‐expressed with FUS‐mCherry, Ubc9TS‐mCherry was co‐expressed with G3BP2‐GFP, and poly‐ubiquitin was detected by immunofluorescence in G3BP2‐GFP‐expressing cells. Automated imaging assay was used with fluorescent ratio threshold 1.4. Mean values from three experiments are shown, each with > 300 SGs (average 2,028). Error bars = SEM. *P < 0.05, **P < 0.01 (t‐test, compared to DMSO).

2. HeLa cells expressing FUS‐mCherry were treated for 3 h with 10 μM MG132 or 2 h with 1 mM sodium arsenate and co‐treated with 40 μM VER or DMSO. Cells were fixed and stained for poly‐ubiquitin. Ubiquitin is visible in SGs in the conditions containing VER. Ubiquitin is also enriched in the nucleus in conditions where cells were treated with VER and/or MG132, both of which impair proteostasis, presumably causing accumulation of ubiquitinated proteins in the nucleus.

3. HeLa cells expressing G3BP2‐GFP were heat‐stressed for 2 h in the presence of DMSO or 40 μM VER and then imaged at 37°C for indicated time.

4. Quantification of SG persistence from live‐cell imaging as described in (C). Complete SG disassembly was scored in cells treated with DMSO (34 cells) or VER (29 cells) and plotted using survival analysis in R. Dashed lines = 95% confidence intervals.

Figure EV4. Disassembly of aberrant SGs

1. HeLa cells expressing FUS‐mCherry and SOD1(A4V)‐GFP were heat‐stressed for 2 h and then imaged during recovery at 37°C. Six independent experiments were performed. Complete SG disassembly was scored in cells containing SOD1‐negative SGs (109 cells) or cells containing SOD1‐positive SGs (76 cells) and plotted using survival analysis in R. Time indicates duration of recovery. Dashed lines = 95% confidence intervals. P < 0.01 (log‐rank test).

2. HeLa cells expressing FUS‐mCherry and SOD1(A4V)‐GFP were treated with heat stress for 2 h and then imaged during recovery at 37°C to observe the disassembly of SOD1‐containing SGs (arrowheads).

Figure 6. Persisting aberrant SGs are transported to the aggresome for degradation

1. HeLa cells expressing G3BP2‐GFP and Ubc9TS‐mCherry were heat‐stressed for 2 h and then imaged at 37°C. Time indicates duration of recovery. Ubc9TS‐containing SGs (arrows) are transported from cell periphery toward the aggresome, where they slowly disappear. Meanwhile, Ubc9TS accumulates in the aggresome and new SGs devoid of Ubc9TS are formed (arrowheads). Scale bar = 10 μm. See also Movie EV4.

2. Trajectories of SGs from the cell depicted in (A).

3. Fraction of cells with an aggresome. HeLa cells expressing G3BP2‐GFP and Ubc9TS‐mCherry were treated with 5 μM nocodazole or DMSO, followed by 2 h of heat stress, 6 h of recovery at 37°C and fixation. Cells with a single large Ubc9TS inclusion were counted from at least 100 cells expressing Ubc9TS. Mean values from three experiments are shown. Error bars = SEM. *P < 0.05 (t‐test).

4. HeLa cells expressing G3BP2‐GFP and Ubc9TS‐mCherry were treated with 2‐h heat stress followed by 6 h recovery at 37°C. The cells were then fixed and stained for p62 or vimentin. Scale bars = 10 μm. Arrow indicates the position of aggresome surrounded by vimentin cage.

5. Cells expressing G3BP2‐GFP and Ubc9TS‐mCherry were treated with 1 μM wortmannin (Wort) 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 presence of aggresome, presence of SGs, and the localization of SGs at the aggresome region. Mean values from three experiments are shown. Error bars = SEM. *P < 0.05 (t‐test).

Figure EV5. Transport of SG components to the aggresome

1. HeLa cells expressing FUS‐mCherry and SOD1(A4V)‐GFP were treated with heat stress for 2 h and then imaged during recovery at 37°C. SOD1‐containing SGs (arrows) are transported toward the aggresome, where they slowly disappear. Meanwhile, SOD1 accumulates in the aggresome and new SGs devoid of SOD1 are formed (arrowheads). Scale bar = 10 μm. Brightness was normalized across frames.

2. HeLa cells expressing G3BP2‐GFP and Ubc9TS‐mCherry were treated either with 2‐h heat stress (left) or 2‐h heat stress followed by 6 h of recovery at 37°C (right). The cells were then fixed and stained for poly‐ubiquitin, HSP27, HSP70, or VCP. After 2 h of heat stress, components of PQC machinery colocalized with Ubc9TS in SGs (insets). After additional 6 h of recovery, the proteins colocalized with Ubc9TS in the aggresome (arrows). Merge = immunofluorescence (green) + Ubc9TS (red). Scale bar = 10 μm.

Figure 7. Model illustrating the interplay between misfolded proteins and SGs

Misfolded proteins are generated due to mutations or stress conditions. These misfolded proteins can form aggregates that can assemble into large inclusions such as the aggresome. However, misfolded proteins can also form aggregates in SGs. The chaperone HSP70 prevents the accumulation of misfolded proteins in SGs. SGs that co‐assemble with misfolded proteins are more stable, less dynamic, do not exhibit the typical liquid‐like properties, and recruit chaperones. SG disassembly is promoted by HSP70, while persisting aberrant SGs are transported to the aggresome and targeted for degradation by autophagy.