G3BP1-dependent mechanism suppressing protein aggregation in Huntington’s models and its demise upon stress granule assembly

Abstract

Stress granules are membrane-less ribonucleoprotein organelles that assemble upon exposure to stress conditions, but rapidly disassemble upon removal of stress. However, chronic stress can lead to persistent stress granules, a feature of distinct age-related neurodegenerative disorders. Among them, Huntington’s disease (HD), which is caused by mutant expansion of the polyglutamine (polyQ) repeats of huntingtin protein (HTT), leading to its aggregation. To identify modulators of mutant HTT aggregation, we define its interactome in striatal neurons differentiated from patient-derived induced pluripotent stem cells (HD-iPSCs). We find that HTT interacts with G3BP1, a characteristic component of stress granules. Knockdown of G3BP1 increases mutant HTT protein levels and abolishes the ability of iPSCs as well as their differentiated neural counterparts to suppress mutant HTT aggregation. Moreover, loss of G3BP1 hastens polyQ-expanded aggregation and toxicity in the neurons of HD C. elegans models. Likewise, the assembly of G3BP1 into stress granules upon distinct stress conditions also reduces its interaction with HTT in human cells, promoting mutant HTT aggregation. Notably, enhancing the levels of G3BP1 is sufficient to induce proteasomal degradation of mutant HTT and prevent its aggregation, whereas the formation of stress granules blocks these ameliorative effects. In contrast, a mutant G3BP1 variant that cannot accumulate into granules retains its capacity to prevent mutant HTT aggregation even when the cells assemble stress granules. Thus, our findings indicate a direct role of G3BP1 and stress granule assembly in mutant HTT aggregation that may have implications for HD.

Introduction

Huntington’s disease (HD) is an autosomal dominant neurodegenerative disorder characterized by dementia, psychiatric manifestations and involuntary movement or chorea (1,2). HD is caused by an abnormal expansion of the CAG repeat in the huntingtin (HTT) gene (3). These mutations expand the polyglutamine (polyQ) tract of the N-terminal domain of HTT protein. In HD patients, HTT contains greater than 35 polyQ repeats, leading to its aggregation and proteotoxicity (4). Different components of the protein homeostasis (proteostasis) network can attenuate the accumulation of mutant HTT aggregates (4). For instance, overexpression of distinct molecular chaperones reduces aggregation in animal and cell models of HD (5–12). Moreover, activation of the ubiquitin-proteasome system induces the degradation of mutant HTT, reducing the accumulation of aggregates (13–19). Conversely, the age-associated decline of the proteostasis network is a major factor for mutant HTT aggregation (4).

Aggregation of polyQ-expanded HTT contributes to the neurodegeneration characteristic of HD (1,3,4,20). Although neuronal death occurs in many brain regions in HD, the striatum undergoes the greatest neurodegeneration (21). However, striatal neurons differentiated from patient-derived-induced pluripotent stem cells (HD-iPSCs) lack polyQ-expanded HTT aggregates and subsequent neurodegeneration (9,14,22,23). These findings support a rejuvenating process of proteostasis during cell reprogramming into immortal iPSCs, allowing these cells to differentiate into striatal neurons without pathological aggregates (24,25). Thus, iPSC-derived neurons can provide a system to define mechanisms suppressing mutant HTT aggregation. Based on this hypothesis, here we performed an analysis of the HTT interactome in iPSC-derived striatal neurons. Notably, we found that HTT interacts with both G3BP1 (Ras GTPase-activating protein-binding protein 1) and its structural and functional homologue G3BP2.

G3BPs are characteristic components of stress granules (26). These dynamic membrane-less ribonucleoprotein organelles appear when the cell is under stress, sequestering mRNA stalled in translation ensued from the global translational arrest in response to stress (27). The assembly of stress granules relies on the dynamic association of RNA-binding proteins such as G3BPs (27–29). Stress granules rapidly disassemble upon removal of stress, releasing sequestered mRNA that can be translated when protein synthesis is reactivated (27). However, the chronic stress and proteostasis failure ensued from aging and disease-related mutations can lead to persistent stress granules (30). For instance, the accumulation of persistent stress granules has been linked to different age-related neurodegenerative diseases that involve protein aggregation such as amyotrophic lateral sclerosis (ALS) and Alzheimer’s (30). In these lines, a recent study demonstrated that G3BP1-positive stress granules also accumulate in the brains of both HD patients and mouse models (31). Thus, stress granules are emerging as regulators of proteostasis with potential implications for aging and disease.

Here, we find an unexpected role of G3BP1 and stress granules in the regulation of polyQ-expanded mutant HTT. Under normal conditions, the interaction between G3BP1 and HTT promotes degradation of mutant HTT and prevents its aggregation. However, stress conditions induce the translocation and assembly of G3BP1 into stress granules, which then loses its interaction with HTT. As a consequence, G3BP1 cannot maintain proteostasis of HTT, triggering the accumulation of polyQ-expanded mutant HTT aggregates. Together, our results suggest an ameliorative role of G3BP1 in HD-related protein aggregation, which is suppressed by stress granule assembly.

Results

Wild-type and polyQ-expanded HTT interact with G3BP proteins

To define the interactome of HTT in human neurons, we assessed striatal neurons differentiated from control and patient-derived iPSC lines (HD-iPSCs) (14,32) (Fig. 1A). The control neurons express wild-type HTT alleles that do not exceed 21 polyQ repeats (14,22). The two independent HD-iPSC lines express one normal allele of HTT but also one polyQ-expanded mutant copy (i.e. HD #1: Q71, Q19; HD #2: Q180, Q18) (14,22,33). First, we performed co-immunoprecipitation (co-IP) experiments followed by label-free proteomics using an antibody against total HTT, which immunoprecipitates both normal and mutant HTT (34). To identify significant interactors of HTT in each neuronal line, we compared protein abundance in HTT-antibody pulldowns with control GFP-antibody pulldowns (Fig. 1B–D and Supplementary Material, Data S1).

Figure 1

Analysis of the interactome of HTT in control and HD iPSC-derived striatal neurons. (A) Schematic representation of the experimental procedure to define the HTT interactome of human neurons. Control and two independent HD iPSC lines expressing different mutant HTT alleles were differentiated into striatal neurons. Then, co-immunoprecipitation (co-IP) experiments with antibodies against either total HTT and polyQ-expanded HTT were followed by quantitative label-free proteomics. To identify significant interactors of HTT, we compared protein abundance in total HTT and polyQ-expanded HTT pulldowns with control GFP-antibody pulldowns. (B) Volcano plot of HTT interactome in control striatal neurons. Graph represents the –log10 (P-value) of a two-tailed t-test plotted against the log2 ratio of protein label-free quantification (LFQ) values from co-IP experiments using HTT antibody compared with control co-IP with GFP antibody. Orange dots indicate significance after correction for multiple testing (n = 3, False discovery rate (FDR) < 0.05 was considered significant). (C) Volcano plot of HTT interactome in HD #1 striatal neurons. Orange dots indicate significance after correction for multiple testing (n = 3, FDR < 0.05 was considered significant). (D) Volcano plot of HTT interactome in HD #2 striatal neurons. Orange dots indicate significance after correction for multiple testing (n = 3, FDR < 0.05 was considered significant). (E) Venn diagram represents the number of specific and common co-immunoprecipitated proteins with HTT antibody in control and HD iPSC-derived striatal neurons.

HTT was the most enriched protein in both control and HD striatal neurons, validating our IP assay (Fig. 1B–D). When we integrated data of the three striatal neurons lines, we identified 26 common interactors of HTT in the control and HD lines (Fig. 1E). Among them, we found previously reported interactors of HTT such as the chromatin modifiers ATF7IP and SETDB1 (34,35), the regulator of neuronal survival TRA2B (36), and the regulators of endosomal vesicle transport F8A1 and RAB11FIP2 (37). Besides previously known interactors, we found that HTT also interacts with the two human G3BP proteins, which is G3BP1 and its close structural and functional homologue G3BP2 (38) (Fig. 1E).

To assess whether G3BPs also interact with mutant HTT in striatal neurons, we performed co-IP experiments with an antibody to polyQ-expanded proteins (39) (Fig. 2A and B and Supplementary Material, Data S2). This antibody recognizes mutant HTT on western blots, but not wild-type HTT (14,34). As such, the anti-expanded polyQ antibody only immunoprecipitates mutant HTT (34) (Fig. 2C–E). Proteomics analysis of co-IP experiments revealed that polyQ71 and polyQ180-expanded mutant HTT also binds G3BPs (Fig. 2A and B and Supplementary Material, Data S2). In addition, co-IP experiments followed by western blot analysis further confirmed the interaction of both wild-type and mutant HTT proteins with G3BP1 in human striatal neurons (Fig. 2C–E).

Figure 2

G3BP1 interacts with both wild-type and polyQ-expanded mutant HTT in striatal neurons. (A) Volcano plot of the polyQ-expanded mutant HTT interactome in HD #1 striatal neurons. –log10 (P-value) of a two-tailed t-test is plotted against the log2 ratio of LFQ values from co-immunoprecipitation (co-IP) experiments using anti-expanded polyQ antibody compared with control co-IP with GFP antibody. Dark grey dots indicate significance after correction for multiple testing (n = 3, FDR < 0.05 was considered significant). Among significant enriched proteins, we detected HTT, G3BP1 and G3BP2 (orange dots). (B) Volcano plot of mutant HTT interactome in HD #2 striatal neurons (n = 3). FDR < 0.05 was considered significant interactors of mutant HTT (dark grey blots), including G3BP1 and G3BP2 (orange dots). (C–E) Co-IP with antibodies against total HTT, expanded-polyQ, G3BP1 and control GFP in control, HD #1, and HD #2 striatal neurons followed by western blot with antibodies to HTT, expanded-polyQ, and G3BP1. These experiments confirmed the interaction of either HTT and mutant HTT with G3BP1 in neurons. The images are representative of three independent experiments.

G3BP1 prevents polyQ-expanded HTT aggregation in HD-iPSCs

iPSCs express significant amounts of HTT protein, which are required to maintain their differentiation capacity (34,35). Despite expressing mutant HTT (Supplementary Material, Fig. S1), immortal HD-iPSCs do not accumulate aggregates indicating that these cells have an enhanced ability to prevent polyQ-expanded HTT aggregation (9,14). Notably, we found that iPSCs exhibit high protein levels of G3BP1, but these amounts decrease during differentiation into neural progenitors and striatal neurons (Fig. 3A). The downregulation of G3BP1 levels during differentiation occurred in both control and HD lines (Fig. 3B).

Figure 3

Knockdown of G3BP1 triggers mutant HTT aggregation in HD-iPSCs. (A) Western blot of control induced pluripotent stem cells (iPSC), neural progenitor cells (NPCs) and striatal neurons (Neurons) with anti-G3BP1 antibody. β-actin is the loading control. The images are representative of three independent experiments. (B) Western blot comparing control and HD-iPSCs with their striatal neuron counterparts using antibodies to G3BP1 and β-actin. The images are representative of three independent experiments. (C) Quantitative PCR (qPCR) analysis of G3BP1 mRNA levels in control and HD-iPSC lines. Graph (relative expression to non-targeting (NT) shRNA) represents the mean ± SEM of three independent experiments for each iPSC line. Statistical comparisons were made by Student’s t-test for unpaired samples. P-value: ^**P < 0.01, ^***P < 0.001. (D) Immunocytochemistry of control and HD-iPSCs upon G3BP1 knockdown. PolyQ-expanded and Hoechst 33342 staining were used as markers of mutant HTT aggregates and nuclei, respectively. Scale bar represents 10 μm. The images are representative of three independent experiments. (E–G) Filter trap analysis of control and HD-iPSC lines comparing non-targeting (NT) shRNA with G3BP1 shRNA-treated cells. G3BP1 knockdown triggers the formation of polyQ-expanded HTT aggregates in HD-iPSCs lines (detected by polyQ-expansion antibody). The images are representative of three independent experiments. (H–J) Western blot of control and HD-iPSC lines with antibodies to polyQ-expanded proteins, G3BP1 and β-actin. Images are representative of three independent experiments.

Thus, we asked whether endogenous increased levels of G3BP1 contribute to the enhanced ability of iPSCs to suppress mutant HTT aggregation. To this end, we knocked down G3BP1 using two independent shRNAs and assessed aggregation of mutant HTT by immunocytochemistry with anti-expanded polyQ antibody (Fig. 3C and D). Remarkably, a moderate reduction of G3BP1 levels was sufficient to trigger the accumulation of polyQ-expanded HTT foci in HD-iPSCs (Fig. 3D). To further confirm these results, we applied filter trap assay, which allows for specific visualization of SDS-insoluble polyQ aggregates. Indeed, knockdown of G3BP1 led to the accumulation of insoluble polyQ aggregates in HD-iPSCs, but not control iPSCs (Fig. 3E–G). Loss of G3BP1 did not alter the transcript levels of HTT, but it increased the total protein amounts of mutant HTT which could contribute to its aggregation (Fig. 3H–J and Supplementary Material, Fig. S2). Together, our results indicate that G3BP1 maintains proteostasis of polyQ-expanded mutant HTT in iPSCs.

Loss of G3BP1 hastens polyQ-expanded aggregation in nematode models

Prompted by our results in HD-iPSCs, we asked whether G3BP1 can also modulate proteostasis of polyQ-expanded HTT in differentiated cells. Indeed, we found that knockdown of G3BP1 induces an increase in both the levels and aggregation of mutant HTT protein in neural cells differentiated from HD-iPSCs (Fig. 4A–D).

Figure 4

Loss of G3BP1 induces mutant HTT aggregation in human neural cells differentiated from HD-iPSCs. (A) Immunocytochemistry of neural progenitor cells (NPCs) derived from HD-iPSC line #1 (Q71) on G3BP1 knockdown. PolyQ-expanded and Hoechst 33342 staining were used as markers of mutant HTT aggregates and nuclei, respectively. Scale bar represents 10 μm. The images are representative of three independent experiments. (B) Filter trap analysis of HD-NPCs #1. G3BP1 knockdown triggers the formation of polyQ-expanded mutant HTT aggregates in NPCs (detected by polyQ-expansion antibody). Right: SDS–polyacrylamide gel electrophoresis (PAGE) with antibodies to polyQ-expanded proteins, G3BP1 and β-actin. The images are representative of three independent experiments. (C) Immunocytochemistry of NPCs derived from HD-iPSC line #2 (Q180) on G3BP1 knockdown. PolyQ-expanded and Hoechst 33342 staining were used as markers of mutant HTT aggregates and nuclei, respectively. Scale bar represents 10 μm. The images are representative of three independent experiments. (D) Filter trap analysis of HD-NPCs #2. G3BP1 knockdown triggers the accumulation of mutant HTT aggregates in NPCs (detected by polyQ-expansion antibody). Right: SDS–PAGE with antibodies to polyQ-expanded proteins, G3BP1 and β-actin. The images are representative of three independent experiments.

We then assessed whether G3BP1 influences expanded polyQ aggregation in post-mitotic, terminally differentiated cells of organismal models. To this end, we knocked down gtbp-1, the C. elegans orthologue of G3BP1, in worm models that express polyQ-expanded peptides throughout the nervous system (40,41). In these animals, neuronal polyQ aggregation correlates with the length of the polyQ repeat, with a threshold of 40 polyQ repeats. Although knockdown of gtbp-1 after development did not affect total polyQ levels, it was sufficient to increase the amounts of SDS-insoluble polyQ67 aggregates in neurons (Fig. 5A and B). In contrast, knockdown of gtbp-1 did not induce aggregation of control polyQ19 peptides (Fig. 5B). We further confirmed upregulation of neuronal polyQ67 aggregation upon loss of GTBP-1 in two independent gtbp-1 knockout mutant worms generated by CRISPR-Cas9 methodology (i.e. gtbp-1(ax2073), gtbp-1(ax2029)) (Fig. 5C).

Figure 5

Loss of gtbp-1 increases aggregation of polyQ-expanded peptides in post-mitotic tissues of C. elegans. (A) Filter trap analysis of C. elegans expressing polyQ67::yellow fluorescent protein (YFP) (detected by anti-GFP antibody) under neuronal-specific F25B3.3 promoter. Right: SDS–PAGE with antibodies to GFP and α-tubulin loading control. Worms were analyzed at day 3 of adulthood. The images are representative of three independent experiments. (B) Filter trap analysis with anti-GFP antibody of worms expressing control polyQ19::YFP or polyQ67::YFP in neurons. Right: SDS–PAGE with antibodies to GFP and α-tubulin loading control. Worms were analyzed at day 3 of adulthood. The images are representative of two independent experiments. (C) Filter trap analysis with anti-GFP antibody of neuronal polyQ67 aggregation in gtbp-1(ax2073) and gtbp-1(ax2029) mutant knockout worms. Right: SDS–PAGE with antibodies to GFP and α-tubulin loading control. Worms were analyzed at day 3 of adulthood. The images are representative of three independent experiments. (D) Neuronal-specific knockdown of gtbp-1 induces intracellular polyQ67 aggregation in neurons. Right: SDS–PAGE with antibodies to GFP and α-tubulin loading control. RNAi rescued in the neurons alone of RNAi-deficient C. elegans (sid-1(pk3321); unc-119p::sid-1; F25B3.3p::Q67::YFP). Worms were analyzed at day 3 of adulthood. The images are representative of three independent experiments. (E) Knockdown of gtbp-1 hastens motility defects in neuronal polyQ67-expressing worms. In contrast, loss of gtbp-1 does not impair motility in control polyQ19-expressing worms. Bar graphs represent average (±SEM) thrashing movements over a 30-s period on day 3 of adulthood (n = 100 worms per condition from three independent experiments). (F) Knockout of gtbp-1 hastens motility defects in neuronal polyQ67-expressing worms. Bar graphs represent average (±SEM) thrashing movements over a 30-s period on day 3 of adulthood (n = 100 worms per condition from three independent experiments). (G) Knockdown of gtbp-1 increases expanded polyQ aggregation in worms that express polyQ44::YFP in the intestine alone (detected by anti-GFP antibody). Right: SDS–PAGE analysis with antibodies to GFP and α-tubulin loading control. Representative of three independent experiments. (H) Filter trap analysis in worms that express polyQ40::YFP in the muscle alone (detected by anti-GFP antibody). Right: SDS–PAGE analysis with antibodies to GFP and α-tubulin loading control. The images are representative of three independent experiments. (I) Knockdown of gtbp-1 promotes motility defects in worms expressing polyQ40::YFP in the muscle. Bar graphs represent average (±SEM) thrashing movements over a 30-s period on day 3 of adulthood (n = 35 worms per condition from two independent experiments). All the statistical comparisons were made by two-tailed Student’s t test for unpaired samples. P value: ^*P < 0.05, ^***P < 0.001, ^****P < 0.0001, NS = not significant (P > 0.05).

In C. elegans, gtbp-1 is ubiquitously expressed in all the tissues (42). To assess whether polyQ67 aggregation in neurons is regulated by intracellular GTBP-1 levels or by cell non-autonomous effects triggered by loss of GTBP-1 in other tissues, we performed neuronal-specific knockdown experiments. To this end, we used an RNAi-deficient mutant C. elegans strain (sid-1(pk3321)) in which RNAi efficiency has been rescued only in neurons by expressing wild-type sid-1gene under a neuronal-specific promoter (43). These worms were crossed with the strain expressing polyQ67 in neurons (41) and assessed by filter trap. We found that neuronal-specific RNAi against gtbp-1 is sufficient to trigger the accumulation of insoluble polyQ67 aggregates in neurons, indicating that intracellular levels of GTBP-1 modulate expanded polyQ aggregation in these cells (Fig. 5D). Aggregation of polyQ67 in neurons causes neurotoxicity, which leads to impaired coordinated movement, a disease-like phenotype (40,41). To examine the physiological consequences of neuronal polyQ67 aggregation induced by loss of gtbp-1 levels, we performed motility assays. We found that loss of gtbp-1 by either RNAi treatment or genetic knockout hastened the detrimental effects on motility induced by polyQ67 expression (Fig. 5E and F). However, loss of gtbp-1 did not affect the motility of control polyQ19 worms (Fig. 5E).

Since gtbp-1 is widely express through the organism, we asked whether gtbp-1 levels also modulate expanded polyQ aggregation in other tissues. To this end, we used C. elegans models that specifically express aggregation-prone polyQ peptides either in the intestine or muscle (41,44,45). Notably, we found that knockdown of gtbp-1 after development promotes intestinal and muscular aggregation of expanded polyQ repeats (Fig. 5G and H), indicating that these effects are not limited to the nervous system. The aggregation of expanded polyQ repeats within muscle cells also triggers intracellular detrimental effects, resulting in muscle dysfunction and decline of motility in C. elegans (45). Importantly, the accumulation of polyQ-expanded repeats in the muscle induced by gtbp-1 knockdown further hastened the motility defects (Fig. 5I). Collectively, our results suggest that lowering GTBP-1/G3BP1 levels in different tissues can trigger the pathological aggregation of expanded polyQ repeats.

Importantly, knockdown of gtbp-1 after development did not make worms more sensitive to acute proteotoxic conditions such as heat stress (34°C) or arsenite treatment (Supplementary Material, Fig. S3A and B). Thus, these data support that the increased polyQ aggregation and neurotoxicity ensued from loss of gtbp-1 are not due to a global decline in stress resilience. However, loss of gtbp-1 slightly shortened lifespan of wild-type worms at 20°C (Supplementary Material, Fig. S3C), as previously reported (46). To determine whether the effects triggered by knockdown of gtbp-1 on polyQ-expanded peptides are specific or caused by premature aging and subsequent loss of global proteostasis mechanisms, we examined other disease-related mutant proteins. To this end, we used worm models expressing ALS-related mutant variants of FUS and TDP-43 (47,48), which are linked with G3BP1 as they can accumulate into stress granules (49). In contrast to polyQ-expanded repeats, knockdown of gtbp-1 decreased the aggregation of ALS-related mutant FUS^P525L in neurons and ameliorated motility deficits (Supplementary Material, Fig. S4A and B). Although knockdown of gtbp-1 did not have a strong impact on TDP-43^M331V aggregation, it also attenuated the motility deficits ensued from this neurotoxic mutant TDP-43 variant (Supplementary Material, Fig. S4C and D). Together, our results indicate a beneficial role of GTBP-1/G3BP1 levels in preventing polyQ aggregation and subsequent neurotoxicity, but GTBP-1/G3BP1 could also influence organismal aging and proteostasis of other disease-related proteins through different mechanisms.

G3BP1 loses its interaction with HTT upon stress granule assembly

G3BPs contribute to the assembly of stress granules (26). As such, G3BP1 is a characteristic component of these ribonucleoprotein complexes that appear when the cell is under stress (29,50,51). In the absence of stress granules, G3BP1 displays a diffuse cytoplasmic distribution (Fig. 6A) (29,50,51). We found that the expression of mutant HTT does not induce the assembly of stress granules in human iPSCs under normal conditions (Fig. 6A). Concomitantly, G3BP1 retained its diffuse cytoplasmic distribution in untreated HD-iPSCs (Fig. 6A). The exposure to cellular stressors such as arsenite, which induces oxidative stress, results in the assembly of stress granules (50). Accordingly, the treatment with arsenite was sufficient to trigger the formation of G3BP1-positive granules in both control and HD-iPSCs (Fig. 6A). However, we could not detect HTT within G3BP1-positive stress granules, indicating that HTT does not translocate into these ribonucleoprotein complexes upon stress (Supplementary Material, Fig. S5).

Figure 6

HTT loses its interaction with G3BP1 upon stress granule assembly leading to mutant HTT aggregation. (A) Immunocytochemistry of control and HD-iPSCs without treatment or under 500 μm arsenite treatment for 1 h. G3BP1 and Hoechst 33342 staining were used as markers of stress granules and nuclei, respectively. Scale bar represents 10 μm. The images are representative of three independent experiments. (B) Immunocytochemistry of control and HD-iPSCs expressing non-targeting (NT) or G3BP1 shRNA #1 under normal and stress conditions (500 μm arsenite, 1 h). PolyQ-expanded, G3BP1, and Hoechst 33342 staining were used as markers of mHTT aggregates, stress granules, and nuclei, respectively. Scale bar, 5 μm. The images are representative of three independent experiments. (C) Immunocytochemistry of HD-iPSCs #1 (Q71) under different stress granule-provoking conditions (i.e. 500 μm arsenite for 1 h, 1 μm MG132 for 3 h or 2 μm thapsigargin for 4 h). Scale bar represents 10 μm. The images are representative of three independent experiments. (D) Co-immunoprecipitation (co-IP) with antibodies against control GFP, total HTT, expanded-polyQ, and G3BP1 in HD-iPSCs #1 (Q71) under normal conditions or treated with 1 μm MG-132 for 3 h. Co-IP experiments were followed by western blot with antibodies to total HTT, polyQ-expanded, and G3BP1. The images are representative of three independent experiments. (E) Co-IP with antibodies against control GFP, total HTT, and expanded-polyQ in HD-iPSCs #1 (Q71) treated with either 500 μm arsenite (1 h) or 2 μm thapsigargin (4 h). Co-IP experiments were followed by western blot with antibodies to total HTT, polyQ-expanded, and G3BP1. The images are representative of three independent experiments. (F, G) Stress granule assembly upon 500 μm arsenite (F) or 1 μm MG-132 treatment (G). Cells were collected after different times of exposure to the corresponding stress treatment (arsenite: 15, 30, and 60 min; MG-132: 2, 4, and 6 h). Then, cells were stained with antibodies to G3BP1 and polyQ-expanded as markers of stress granule and mutant HTT aggregates, respectively. (H, I) Stress granule disassembly after removal of stress. The medium containing arsenite (H) or MG-132 (I) was removed and replaced with fresh medium. In both cases, the recovery time was 3, 6, and 24 h. In F–I, scale bar, 20 μm. The images are representative of three independent experiments. Graphs represent the percentage of cells (at least 200 total cells per each time point in one representative experiment) divided into four groups: gray = cells with no stress granules or polyQ aggregates, red = cells with only stress granules, green = cells with only mutant HTT aggregates, black = cells with both stress granules and mutant HTT aggregates.

In addition to SG assembly, arsenite treatment alone was sufficient to trigger mutant HTT aggregation even in HD-iPSCs expressing normal levels of G3BP1 (Fig. 6B and C). When we induced mutant HTT aggregation through moderate knockdown of G3BP1 (Fig. 3C–J), HD-iPSCs were still able to form G3BP1-positive stress granules upon arsenite treatment although to a lesser extent when compared to control shRNA cells (Supplementary Material, Fig. S6). Importantly, polyQ-expanded mutant HTT aggregates did not co-localize with G3BP1-positive granules in cells with either downregulated or normal levels of G3BP1 (Fig. 6B and C). Together, these results indicate that a decrease in the levels of available cytoplasmic G3BP1 ensued from either its knockdown or assembly into stress granules promotes aggregation of mutant HTT.

To further assess this hypothesis, we first examined other stress granule-provoking conditions such as proteasome inhibition or endoplasmic reticulum (ER) stress. Similar to arsenite, the treatment of naïve HD-iPSCs with either the proteasome inhibitor MG-132 or the ER-stress inductor thapsigargin induced the formation of both stress granules and polyQ-expanded mutant HTT aggregates, which did not co-localize (Fig. 6C). Notably, the assembly of stress granules upon MG-132, arsenite or thapsigargin treatment suppressed the interaction of G3BP1 with both wild-type and mutant HTT in HD-iPSCs (Fig. 6D and E). In addition, stress granule assembly preceded the accumulation of mutant HTT aggregates (Fig. 6F and G), further supporting that the translocation of G3BP1 to stress granules contributes to polyQ-expanded HTT aggregation. Upon removal of stress, mutant HTT aggregates persisted after the rapid disassembly of SGs (Fig. 6H and I). Collectively, our results indicate that G3BP1 prevents the aggregation of mutant HTT under normal conditions. Upon stress conditions, G3BP1 accumulates within stress granules and loses its interaction with mutant HTT, leading to polyQ-expanded HTT aggregates that can persist after removal of stress.

Increase of G3BP1 levels suppresses mutant HTT aggregation

Given that lowering the levels of available cytoplasmic G3BP1 induces polyQ-expanded HTT aggregation, we asked whether increasing G3BP1 amounts can reduce the accumulation of mutant HTT aggregates. To this end, we generated HEK293 human cell models that express either control Q23-HTT or aggregation-prone Q100-HTT protein. Indeed, control Q23-HTT did not form aggregates whereas mutant Q100-HTT expression resulted in a pronounced accumulation of SDS-insoluble aggregates (Fig. 7A) (14). In these cells, knockdown of G3BP1 also increased the total levels of mutant Q100-HTT and hastened the accumulation of Q100-HTT aggregates (Fig. 7B). Conversely, overexpression of G3BP1 ameliorated Q100-HTT aggregation, a process that correlated with a decrease in the total levels of mutant HTT (Fig. 7A and B). In contrast, G3BP1 overexpression did not decrease Q23-HTT levels (Fig. 7A). Thus, we asked whether G3BP1 overexpression promotes proteasomal degradation of Q100-HTT. Indeed, proteasome inhibition upon MG-132 treatment blocked the reduction in Q100-HTT levels and aggregation induced by enhanced G3BP1 expression (Fig. 7C). Likewise, we observed similar results when we treated the cells with epoxomicin, a different proteasome inhibitor (Supplementary Material, Fig. S7). Together, these results suggest that the interaction of G3BP1 with HTT not only prevents aggregation of mutant HTT, but also induces its degradation through the proteasome. Autophagy is another major proteolytic system that regulates proteostasis of aggregation-prone proteins. We observed that neither knockdown nor overexpression of G3BP1 affects autophagy in cells expressing control or polyQ-expanded HTT (Supplementary Material, Fig. S8A–D), further supporting a role of the proteasome in G3BP1-regulated proteostasis of mutant HTT.

Figure 7

Increasing G3BP1 levels suppresses mutant HTT aggregation. (A) Filter trap with anti-GFP antibody of HEK293 cells expressing either Q23-HTT-GFP or Q100-HTT-GFP. Overexpression (OE) of GFP-G3BP1 ameliorates the accumulation of Q100-HTT aggregates. Right: SDS–PAGE with antibodies to HTT, G3BP1 and β-actin. The images are representative of two independent experiments. (B) Filter trap analysis with expanded-polyQ antibody of Q100 HTT-HEK293 cells upon knockdown or overexpression of G3BP1. Right: SDS–PAGE with antibodies to HTT, G3BP1 and β-actin. The images are representative of three independent experiments. (C) Filter trap analysis with anti-GFP antibody. Right: SDS–PAGE with antibodies to HTT, G3BP1 and β-actin. Proteasome inhibition with 0.5 μm MG-132 (16 h) blocks the degradation of mutant HTT and subsequent reduction of polyQ-expanded aggregation induced by G3BP1 overexpression. The images are representative of four independent experiments. (D) Co-immunoprecipitation (co-IP) with control IgG and antibodies against HTT, expanded-polyQ, and G3BP1 in Q100 HTT-HEK293 cells overexpressing wild-type G3BP1 under normal conditions or treated with 500 μm arsenite (1 h). Co-IP experiments were followed by western blot with antibodies to polyQ-expanded and G3BP1. The images are representative of two independent experiments. (E) Filter trap analysis with anti-GFP antibody of Q100 HTT-HEK293 cells overexpressing wild-type GFP-G3BP1 under normal conditions or treated with 500 μm arsenite for 1 h. Right: SDS–PAGE with antibodies to HTT, G3BP1 and β-actin. The images are representative of three independent experiments. (F) Immunocytochemistry of HEK293 cells expressing either GFP-G3BP1 or mutant GFP-G3BP1-ΔRBP under 500 μm arsenite treatment for 1 h. PABP1 and Hoechst 33342 staining were used as markers of stress granules and nuclei, respectively. Scale bar represents 20 μm. The images are representative of two independent experiments. (G) Filter trap of HEK293 cells expressing Q100-HTT-GFP with anti-HTT antibody. Overexpression of either GFP-G3BP1 or mutant GFP-G3BP1-ΔRBP ameliorates the accumulation of Q100-HTT aggregates. Right: SDS–PAGE with antibodies to HTT, G3BP1 and β-actin. The images are representative of two independent experiments. (H) co-IP with control IgG and antibodies against HTT, expanded-polyQ, and G3BP1 in Q100 HTT-HEK293 cells overexpressing mutant G3BP1-ΔRBP under normal conditions or treated with 500 μm arsenite (1 h). Co-IP experiments were followed by western blot with antibodies to polyQ-expanded and G3BP1. The images are representative of two independent experiments. (I) Filter trap analysis with anti-GFP antibody of Q100 HTT-HEK293 cells overexpressing mutant GFP-G3BP1-ΔRBP under normal conditions or treated with 500 μm arsenite for 1 h. Right: SDS–PAGE with antibodies to HTT, G3BP1 and β-actin. The images are representative of three independent experiments.

Upon arsenite treatment, the accumulation of G3BP1 into stress granules blocked its interaction with mutant HTT (Fig. 7D). Subsequently, the induction of stress granules suppressed the reduction of both mutant HTT levels and aggregation triggered by G3BP1 overexpression (Fig. 7E and F). To further determine whether the assembly of G3BP1 into stress granules modulates polyQ-expanded HTT aggregation, we assessed a mutant G3BP1 variant that cannot accumulate into stress granules due to deletion of its RNA-binding domains (G3BP1-ΔRBP) (Fig. 7F) (52). Under normal conditions, overexpression of G3BP1-ΔRBP induced a similar reduction in Q100-HTT protein levels and aggregation when compared with wild-type G3BP1 (Fig. 7G). In contrast to wild-type G3BP1, mutant G3BP1-ΔRBP did not assemble into stress granules on arsenite treatment as assessed by PABP1 staining, which is a different marker of stress granules (Fig. 7F). Upon stress granule assembly, G3BP1 retained its interaction with Q100-HTT in G3BP1-ΔRBP expressing cells (Fig. 7H). Concomitantly, overexpression of G3BP1-ΔRBP induced the degradation of Q100-HTT and prevented its aggregation even under arsenite treatment (Fig. 7I). Thus, our results indicate that G3BP1 maintains proteostasis of mutant HTT, but the assembly of stress granules diminishes its capacity to suppress mutant HTT aggregation.

Discussion

Persistent and aberrant G3BP1-positive stress granules are a common feature of distinct neurodegenerative disorders that involve protein aggregation such as ALS, frontotemporal dementia, Alzheimer’s, Parkinson’s and HD (30,31,53,54). Cumulative evidence indicates that aberrant stress granules contribute to the pathophysiology of neurodegenerative diseases (30,31,53,54). The link is more compelling in ALS, where several disease-related mutant proteins with low-complexity domains can accumulate within stress granules and alter their dynamics, including TDP-43 and FUS (55–60). Frontotemporal dementia (FTD) largely overlaps with ALS at the genetic level (61). As such, FTD and ALS are considered to represent two ends of a spectrum disorder with common alterations in key biological processes, such as stress granule dynamics (61).

However, the links between stress granules and HD are less defined. Here we found that stress granules can indeed influence the proteostasis of mutant HTT. By interactome analysis, we first observed that HTT interacts with G3BP proteins under normal conditions. Decreasing this interaction through knockdown of G3BP1 is sufficient to trigger mutant HTT aggregation, even in human cells that have a striking ability to prevent pathological aggregation such as iPSCs. In addition, loss of G3BP1 increased the total levels of mutant HTT protein, suggesting that G3BP1 promotes the degradation of HTT under normal conditions. Indeed, overexpression of G3BP1 in HEK293 cells dramatically reduced the levels of HTT, whereas the treatment with proteasome inhibitor blocked this effect. Given that loss of G3BP1 appears to have stronger effects on mutant HTT aggregation than protein levels (Fig. 7A), the interaction of G3BP1 may also prevent aggregation through other mechanisms such as a potential chaperone activity. In these lines, loss of the worm orthologue of G3BP1 did not affect total polyQ-expanded levels in C. elegans whereas it has pronounced effects on aggregation. Under challenging conditions such as oxidative and ER stress, the assembly of G3BP1 into stress granules reduces the amounts of available G3BP1 for other interactions in the cytoplasm. Indeed, the assembly of stress granules in cells expressing normal levels of G3BP1 was sufficient to increase mutant HTT amounts and aggregation, mimicking the phenotypes triggered by G3BP1 knockdown under normal conditions.

As a protective mechanism to counteract the proteotoxic effects triggered by exposure to stress, cells induce a global arrest in translation to decrease the loading of newly synthesized proteins that could undergo misfolding and aggregation (62). One of the biological functions ascribed to stress granules is to sequester mRNA stalled in translation and rapidly release them upon removal stress (27). This process would allow for the rapid recovery of normal translation once protein synthesis is restarted. In this paradigm, we speculate that the assembly of stress granules may represent a double-edged sword for cellular proteostasis in HD, which has been recently associated with the accumulation of persistent stress granules (31). Stress granules may relieve the burden on proteostasis caused by the chronic stress triggered by mutant HTT expression with age. On the other hand, this process might result in the sequestration of G3BP1 into stress granules, which then loses its capacity to prevent aggregation of mutant HTT. In summary, our findings reveal a regulation of polyQ-expanded aggregation by G3BP1 and stress granules, which could have implications to understand HD and define therapeutic approaches.

Materials and Methods

iPSC lines and culture

The control (Q21, ND42242) and HD iPSC #2 lines (Q180, ND36999) were obtained from NINDS Human Cell and Data Repository through Coriell Institute. These iPSC lines were established and fully characterized for pluripotency in Ref. (22). The HD iPSC line #1 was a kind gift from G.Q. Daley (33). iPSCs were maintained on Geltrex (Thermo Fisher Scientific) using mTeSR1 media (Stem Cell Technologies). Undifferentiated iPSC colonies were passaged using a solution of dispase (2 mg ml⁻¹) and scraping the colonies with a glass pipette. All the cell lines used in this study were tested for mycoplasma contamination at least once every two weeks. No mycoplasma contamination was detected.

Differentiation into striatal neurons

iPSCs were differentiated into striatal neurons through the induction of hedgehog signaling pathway (14,32). First, iPSCs were detached by treatment with dispase (1 mg ml⁻¹) for 20 min. The detached colonies were then cultured in suspension as free-floating embryoid bodies (EBs) in differentiation medium (Dulbecco’s Modified Eagle Medium (DMEM)/F12, 20% knockout serum replacement, 100 μm β-mercaptoethanol (Sigma), 2 mm l-glutamine, and 1× minimum essential medium (MEM) non-essential amino acids). On day 4, the medium was replaced with neural induction medium (DMEM/F12, N2 supplement (ThermoFisher Scientific), 2 mm glutamine, 1× MEM non-essential amino acids, and 2 μg ml⁻¹ heparin). On day 7, the EBs were attached to the laminin (ThermoFisher Scientific)-coated substrate in a 35 mm culture Petri dish and cultured in neural induction medium. In the next week, EBs flattened while columnar neuroepithelia organized into rosettes appeared in the center of individual colonies. On day 12, 0.65 μm purmorphamine (Stem Cell Technologies) was added until day 25 (14 days in total). From day 26, neuroepithelial spheres were dissociated with Accutase (1 unit ml⁻¹, Invitrogen) at 37°C for 5 min and placed onto poly-ornithine/laminin-coated plates in Neurobasal medium containing the following trophic factors: insulin-like growth factor-1 (10 ng ml⁻¹), brain-derived neurotrophic factor (20 ng ml⁻¹), glial-derived neurotrophic factor (10 ng ml⁻¹), and cAMP (1 μm) (all from R&D Systems). DARPP32-expressing neurons appeared by day 32 as assessed by immunohistochemistry with Rabbit anti-DARPP32 (Abcam, ab40801, 1:50). The experiments were performed between days 32 and 35 of the differentiation protocol.

Protein immunoprecipitation for interactome assays

Cells were lysed in modified Radioimmunoprecipitation assay (RIPA) buffer (50 mm Tris–HCl (pH 7.4), 150 mm NaCl, 0.25% sodium deoxycholate, 1% IgPal, 1 mm PMSF and 1 mm EDTA) supplemented with protease inhibitor (Roche). Lysates were centrifuged at 10 000g for 10 min at 4°C. Then, the supernatant was collected and incubated with total anti-HTT antibody (Cell Signaling, #5656, 1:50) or anti-polyQ-expanded HTT antibody (DSHB, MW1, 1:50) for 30 min and subsequently with 100 μl Protein A beads (Miltenyi, Germany) for 1 h on the overhead shaker at 4°C. As a control, the same amount of protein was incubated with anti-GFP (ImmunoKontakt, 210-PS-1GFP, 1:100) in parallel. After this incubation, supernatants were subjected to magnetic column purification. We then performed three washes using wash buffer 1 (50 mm Tris–HCl (pH 7.4), 150 mm NaCl, 0.05% IgPal, and 5% glycerol). Next, columns were washed five times with wash buffer 2 (50 mm Tris–HCl (pH 7.4), 150 mm NaCl). Then, we subjected the columns to in-column tryptic digestion containing 7.5 mm ammonium bicarbonate, 2 M urea, 1 mm DTT and 5 ng ml⁻¹ trypsin. Digested peptides were eluted using two times 50 μl of elution buffer 1 containing 2 M urea, 7.5 mm Ambic and 5 m IAA. Digests were incubated overnight at room temperature with mild shaking in the dark. The next day, samples were stage tipped for label-free quantification (LFQ) proteomics and analyzed with MaxQuant software. The downstream analyses were carried out on LFQ values with Perseus (v. 1.5.2.4). For co-IP assays followed by western blot, pellets were incubated after the washing steps with 2× Laemmli Buffer, boiled for 5 min and centrifuged 5 min at maximum speed. Then, the supernatant was collected and loaded onto a sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) gel for western blot analysis.

Lentiviral infection for knockdown experiments

Lentivirus (LV)-non-targeting short hairpin (shRNA) control, LV-G3BP1 shRNA #1 (TRCN0000008722), and LV-G3BP1 shRNA #2 (TRCN0000008723) in pLKO.1-puro vector were obtained from Mission shRNA (Sigma). Transient infection of iPSC lines was performed as follows. iPSCs colonies growing on Geltrex were incubated with mTesR1 medium containing 10 μm ROCK inhibitor (Abcam) for 2 h and individualized using Accutase (1 unit ml⁻¹, Invitrogen). Hundred thousand cells were plated on Geltrex-coated plates and incubated with mTesR1 medium containing 10 μm ROCK inhibitor for 1 day. Then, cells were infected with 5 μl of concentrated lentivirus. Plates were centrifuged at 800g for 1 h at 30°C. Cells were fed with fresh media the day after to remove the virus. After 1 day, iPSCs were selected for lentiviral integration using 2 μg ml⁻¹ puromycin (ThermoFisher Scientific). iPSCs were split and collected after 5–7 days of infection for further experiments such as filter trap, western blot and immunocytochemistry. For experiments on NPCs, we induced neural differentiation of shRNA-expressing iPSCs with STEMdiff Neural Induction Medium (Stem Cell Technologies) following the monolayer culture method (63). Briefly, undifferentiated shRNA-iPSCs were rinsed once with phosphate-buffered saline (PBS) and then we added 1 ml of Gentle Dissociation Reagent (Stem Cell Technologies) for 10 min. After the incubation period, we gently dislodged pluripotent cells and added 2 ml of DMEM/F12 (ThermoFisher Scientific) + 10 μm ROCK inhibitor. Then, cells were centrifuged at 300g for 10 min. Cells were resuspended on STEMdiff Neural Induction Medium + 10 μm ROCK inhibitor and plated on laminin (10 μg ml⁻¹)/poly-ornithine (15 μg ml⁻¹)-coated plates at a density of 200 000 cells cm⁻². After 8 days in vitro, NPCs were analyzed for further experiments. To knockdown G3BP1 in HEK293 lines, HEK293 cells (ATCC, #CRL-11268) were transduced with 5 μl of concentrated lentivirus and selected by adding puromycin at a concentration of 2 μg ml⁻¹.

Transfection of HEK293T cells

HEK293T cells (ATCC) were plated on 0.1% gelatin-coated plates and grown in DMEM supplemented with 1% MEM non-essential amino acids (ThermoFisher Scientific) and 10% fetal bovine serum at 37°C. HEK293T cells were transfected once they reached 80–90% confluency. 1 μg of pEGFP-C1-G3BP1-WT, pEGFP-C1-G3BP1-ΔRBP (52), pARIS-mCherry-httQ23-GFP, pARIS-mCherry-httQ100-GFP plasmids were used for transfection, using Fugene HD (Promega) following the manufacturer’s instructions. After 36 h of incubation, we harvested the cells for further experiments. The pARIS-mCherry-httQ23-GFP and pARIS-mCherry-httQ100-GFP plasmids were a gift from Saudou (64). pEGFP-C1-G3BP1-WT (Addgene plasmid #135997) and pEGFP-C1-G3BP1-DeltaRBP were a gift from A. Leung (Addgene plasmid #135999).

Western blot

Cells were scraped from tissue culture plates and lysed in protein cell lysis buffer (50 mm Hepes pH 7.4, 150 mm NaCl, 1 mm EDTA, 2 mm sodium orthovanadate, 1 mm PMSF, 1% Triton X-100, and protease inhibitor mix (Roche)). Lysates were homogenized through syringe needle (27 G) on ice followed by centrifugation at 8000g for 5 min at 4°C. We then collected the supernatants (with the exception of the western blot showed in Fig. 7C, where we loaded whole cell homogenates without centrifugation because proteasome inhibition induces high concentration of Q100-HTT in the pellet fraction) and determined protein concentrations with a standard BCA protein assay (ThermoFisher Scientific). Approximately 30 μg of total protein was separated by SDS–PAGE, transferred to polyvinylidene difluoride membranes (Millipore) and subjected to immunoblotting. Western blot analysis was performed with anti-G3BP1 (MBL, #RN048PW, 1:500), anti-β-actin (Abcam, #8226, 1:5000), anti-GFP (AMSBIO, #TP401, 1:5000), anti-LC3 (Sigma, #L7543, 1:1000) and anti–α-tubulin (Sigma-Aldrich, #T6199, 1:5000). To detect total HTT protein and polyQ-expanded mutant HTT, we used anti-HTT (Cell Signaling, #5656, 1:1000) and anti-polyQ-expansion diseases marker (Millipore, MAB1574, 1:1000), respectively (14,39).

Immunocytochemistry

Cells were fixed with paraformaldehyde (4% in PBS) for 20 min, followed by permeabilization (0.2% Triton X-100 in PBS, 10 min) and blocking (3% bovine serum albumin in 0.2% Triton X-100 in PBS, 10 min). Cells were incubated in primary antibody for 2 h at room temperature (Mouse anti-polyQ (Millipore, #MAB1574, 1:50), Rabbit anti-HTT (Cell Signaling, #5656, 1:200), Rabbit anti-G3BP1 (MBL, #RN048PW, 1:500), Mouse anti-G3BP1 (Abcam, #56574, 1:500), Rabbit anti-PABP (Abcam, #21060, 1:400)). Then, the cells were washed with 0.2% Triton-X/PBS and incubated with secondary antibody (Alexa Fluor 488 Goat anti-Mouse (ThermoFisher Scientific, #A11029, 1:500), Alexa Fluor 568F(ab′)2 Fragment of Goat Anti-Rabbit IgG (H + L) (ThermoFisher Scientific, #A-21069, 1:500), or Alexa Fluor 594 Goat Anti-Rabbit IgG (H + L) (ThermoFisher Scientific, #A-11037, 1:500)) and Hoechst 33342 (Life Technologies, #H3570, 1:1000) for 1 h at room temperature. PBS and distilled water washes were followed and then the cover slips were mounted on Mowiol (Sigma, #324590).

Filter trap

Human cell cultures were collected in non-denaturing lysis buffer (50 mm Hepes pH 7.4, 150 mm NaCl, 1 mm EDTA, 1% Triton X-100) supplemented with EDTA-free protease inhibitor cocktail (Roche). Then, human cell samples were lysed by passing 10 times through a 27 G needle attached to a 1 ml syringe. Regarding C. elegans samples, the worms were collected with M9 buffer and worm pellets were frozen with liquid N₂. Frozen worm pellets were thawed on ice and worm extracts were generated by glass bead disruption. Then, either worm or cellular debris was removed with 8000g spin for 5 min and protein concentration was determined with BCA protein assay. Approximately 100 μg of protein extract was supplemented with SDS at a final concentration of 0.5% and loaded onto a cellulose acetate membrane assembled in a slot-blot apparatus (Bio-Rad). The membrane was then washed with 0.2% SDS. In human cells and C. elegans extracts, the retained polyQ-expanded proteins were assessed by immunoblotting for anti-polyQ-expansion diseases marker (Millipore, MAB1574, 1:5000) and anti-GFP antibodies (AMSBIO, #TP401, 1:5000), respectively. In addition, extracts were also analyzed by SDS–PAGE to determine the total levels of the corresponding proteins using the aforementioned antibodies.

RNA isolation and quantitative reverse transcription-PCR

We extracted total RNA using RNAbee (Tel-Test Inc.) and generated complementary DNA (cDNA) using qScript Flex cDNA synthesis kit (Quantabio). SybrGreen real-time quantitative PCR (qPCR) assays were performed with a 1:20 dilution of cDNA using a CFC384 Real-Time System (Bio-Rad) following the manufacturer’s instructions. Data were analyzed with the comparative 2ΔΔCt method using the geometric mean of ACTB and GAPDH as housekeeping genes.

C. elegans strains and maintenance

C. elegans were grown and maintained at 20°C on standard nematode growth media seeded with Escherichia coli (OP50) bacteria (65). Wild-type (N2), JH3215 (gtbp-1(ax2073)IV) (66), JH3176 (gtbp-1(ax2029)IV) (66), TU3401 (sid-1(pk3321)V;uIs69[pCFJ90(myo-2p::mCherry) + unc-119p::sid-1]), and AM141 (rmIs133[unc-54p:: Q40::YFP]) strains were obtained from the Caenorhabditis Genetics Center (CGC) (University of Minnesota), which is supported by the National Institutes of Health Office of Research Infrastructure Programs (P40 OD010440). AM716 (rmIs284[pF25B3.3::Q67::YFP]) and AM23 (rmIs298[F25B3.3p::Q19::CFP]) strains were a kind gift from Morimoto (40). MAH602 (sqIs61[vha-6p::Q44::YFP + rol-6(su1006)]) was a gift from Hansen (44). ZM5844 (hpIs233[rgef-1p::FUS^P525L::GFP]) (48) and CK423 (Psnb-1::TDP-43^M337V, myo-2p::dsRED) (47) were a gift from P. St George-Hyslop and b.c. Kramer, respectively.

To obtain gtbp-1 knockout mutant worms expressing neuronal polyQ67, AM716 was crossed to JH3215 and JH3176, generating DVG278 (rmIs284[pF25B3.3::Q67::YFP]; gtbp-1(ax2029)IV)) and DVG279 (rmIs284[pF25B3.3::Q67::YFP]; gtbp-1(ax2073)IV) strains. To perform neuronal-specific RNAi, we used the TU3401 strain (sid-1(pk3321)V;uIs69[pCFJ90(myo-2p::mCherry) + unc-119p::sid-1]), in which RNAi treatment is only functional in neurons (43). These worms were crossed to AM716, generating DVG196 (rmIs284[F25B3.3p::Q67::YFP]; sid-1(pk3321)V;uIs69[pCFJ90(myo-2p::mCherry) + unc-119p::sid-1]) (41).

RNAi experiments in C. elegans

RNAi-treated strains were fed E. coli (HT115) containing either an empty control vector (L4440) or expressing double stranded gtbp-1 RNAi. gtbp-1 RNAi construct was obtained from the Ahringer RNAi library and sequence verified.

Motility assays

We transferred synchronized day 3-adult worms to a drop of M9 buffer. After 30 s of adaptation, we counted the number of body bends (i.e. change in direction of the bend at the mid-body) for 30 s (40).

Lifespan and survival assays in C. elegans

For lifespan assays, we synchronized larvae by egg laying protocol. Then, larvae were raised and fed OP50 E. coli at 20°C until day 1 of adulthood. Once hermaphrodite worms reached adulthood, they were shifted to plates with HT115 E. coli carrying empty vector or gtbp-1 RNAi constructs to monitor adult lifespan. Adult worms (n = 96 per condition) were scored every day or every other day (67). We censored the worms that were lost and those with ‘protruding vulva’ or that undergo bagging.

For acute heat stress assays, worms were also synchronized by egg laying protocol at 20°C. Worms were fed empty vector or gtbp-1 RNAi after development and maintained at 20°C. At day 3 of adulthood, worms were shifted to 34°C. After 8 h of heat stress, worms were kept at 34°C and scored every hour for viability. For arsenite stress survival, day 1-adult worms were treated with a final concentration of 0.03% arsenite and empty vector or gtbp-1 RNAi (68). Worms were cultured at 20°C and transferred every day to fresh plates containing the arsenite and RNAi treatment. Worms were scored every day for survival to arsenite stress.

To determine median lifespan and stress survival, we used GraphPad Prism software (version 6.0). We used OASIS software (version 1) for statistical analysis to calculate mean lifespan and survival rates (69). P values were obtained using the log-rank (Mantel–Cox) method. The P values refer to the comparison between control and RNAi-treated animals in a single lifespan or survival assay. In the corresponding figures, each graph presents a representative experiment. See Supplementary Material, Data S3 for statistical analysis and replicate data of lifespan and survival assays.

Acknowledgements

We thank the CECAD Proteomics for their advice and contribution to proteomics experiments and analysis. Figure 1A was generated using BioRender.com.

Conflict of Interest statement. The authors declare no competing interests.

Funding

The Deutsche Forschungsgemeinschaft (DFG) (Germany’s Excellence Strategy-CECAD, EXC 2030-390661388); the Köln Fortune Program/Faculty of Medicine/University of Cologne (Project #338/2019); and the Else Kröner-Fresenius-Stiftung (2021-EKSE.95) supported this research.

Data availability

All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. All the proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository (Project Accession: PXD037618). Additional data related to this paper can be requested from the corresponding author. There is no restriction on data availability.

Authors’ Contributions

R.G.G. designed and performed most of the experiments as well as data collection, analysis, and interpretation. R.G.G. also assembled the figures and contributed to writing the manuscript. S.K. and F.H. contributed to the experiments using HEK293 cells and performed several C. elegans experiments including stress assays. S.B. carried out C. elegans lifespan assays. H.J.L. contributed to analysis of aggregation. A.F. assessed the levels of G3BP1 in iPSCs and their differentiated counterparts. D.V. supervised the project and wrote the manuscript. All authors discussed the results and commented on the manuscript.