FUS and TDP-43 aggregation are uncoupled from toxicity in ageing yeast models

Abstract

Background

Protein aggregation is indicative of the loss of proteostasis associated with neurodegenerative diseases, including Amyotrophic Lateral Sclerosis (ALS) and Frontotemporal Dementia (FTD). Proteins like Fused in sarcoma (FUS) and Tar DNA-binding protein 43 (TDP-43) accumulate and aggregate in the cytosol of neurons in ALS/FTD. Yet, it remains unclear how ageing affects FUS and TDP-43 aggregation, and how these aggregates in turn influence neurodegeneration in ALS/FTD. In addition, mistranslation can reduce longevity, challenge proteostasis, and modulate protein aggregation. To investigate how ageing and mistranslation modulate FUS and TDP-43 aggregation and toxicity, we enlist tractable and reliable yeast models.

Results

Using optimized low-expression FUS and TDP-43 yeast models, we demonstrate that chronological ageing antagonizes proteostasis, the steady state levels and solubility of molecular chaperones, and aggregation of FUS and TDP-43. In addition, mistranslation caused by tRNA variants further antagonize FUS and TDP-43 aggregation and synergize to exacerbate FUS and TDP-43 cytotoxicity.

Conclusions

Our work provides new insights into factors that uncouple FUS and TDP-43 aggregation from toxicity and support a rather protective role for FUS and TDP-43 aggregates in promoting longevity.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12915-026-02537-3.

Background

Accumulation of misfolded proteins is a hallmark of neurodegenerative diseases, including Amyotrophic Lateral Sclerosis (ALS) [1–3]. ALS manifests most commonly in sporadic cases (~ 90%, sporadic ALS, sALS), while the remaining familial cases are inherited in a mendelian fashion (familial ALS, fALS). Mutations in more than 30 different genes cause fALS, most prominently those encoding C9orf72, Superoxide dismutase 1 (SOD1), Fused in sarcoma (FUS) and Tar DNA-binding protein 43 (TDP-43) [4, 5]. In most ALS cases, TDP-43 [1–3, 6] and, less frequently, FUS [7] mislocalize from the nucleus to the cytosol where they accumulate and aggregate. The prion-like and highly intrinsically disordered domains of FUS and TDP-43 predispose these proteins to mislocalization and misfolding [8–10]. Also, both proteins form condensates by liquid–liquid phase separation [11–13], which may promote conversion of the accumulated cytosolic misfolded proteins into more stable, insoluble aggregates in the neurons of ALS patients. Of note, the role of these aggregates in ALS pathogenesis remains unclear.

Key characteristics of TDP-43 and FUS, such as their misfolding, mislocalization, accumulation, and the ensuing cytotoxicity, have been recapitulated in yeast models [8–10, 14–17]. These yeast models identified genetic modulators of TDP-43 [18–26] and FUS [9, 10, 27, 28] toxicity, leading to the discovery of modulators of ALS [9, 10, 20, 22, 27]. Yeast models have also been instrumental in deciphering the molecular and cellular mechanisms underlying TDP-43 and FUS aggregation [8–10, 14, 15, 17, 23, 26, 28, 29].

Although the occurrence of protein aggregates in brain and spinal cord tissues is well-established neuropathological hallmarks of ALS/FTD, the causative link between these aggregates and disease progression remains tenuous. Numerous studies indicate only weak correlations between aggregation and eventual neuronal cell death [30–32]. Conversely, many findings even indicate potential protective functions of protein aggregates in diverse neurodegenerative diseases [33, 34]. Accordingly, genetic modifiers of the toxicity of misfolded proteins are not necessarily modifiers of their aggregation [8–10, 32, 35]. Although TDP-43 and FUS aggregation is a clear indicator of impaired proteostasis, it remains unclear exactly how TDP-43 and FUS aggregates contribute to cellular toxicity, if at all.

Moreover, the role of ageing in the formation and persistence of protein aggregates and toxicity of misfolded proteins has been notoriously difficult to ascertain. Indeed, chronological ageing hampers the ability of cells to maintain proteostasis [36–39]. However, the inability to form and maintain protein aggregates in aged cells is associated with reduced lifespan [40], indicating that age-dependent protein aggregates can be protective. While cultured mammalian cells are often used to model cellular senescence [41], they do not entirely recapitulate most physiological aspects of ageing. Studies of TDP-43 and FUS misfolding in the context of ageing have relied heavily on rodent models [42, 43], which are too complex to track cellular and molecular details linking ageing to TDP-43 and FUS aggregation and toxicity.

In addition to impaired proteostasis, ageing cells must contend with errors in protein biosynthesis, or mistranslation. Many have speculated [44] and shown experimentally [45–47] that mistranslation adversely affects longevity. While the specific mechanisms by which translation fidelity influences lifespan remain unclear, mistranslation clearly impairs proteostasis and induces protein misfolding and aggregation [48–52]. This plausibly contributes to the decline of proteostasis seen in ageing cells. Further, mistranslation modulates the misfolding, aggregation, and toxicity of neurodegenerative disease-associated proteins, such as polyglutamine expanded huntingtin and FUS [53, 54]. However, the degree to which ageing and mistranslation synergize to affect the misfolding of the ALS/FTD-associated proteins FUS and TDP-43 remains enigmatic.

Yeast models have a rich history in the study of ageing, including the identification of key cellular factors that influence longevity, such as TOR kinase complex 1 (TOR) signaling [55], the unfolded protein response [37], reactive oxygen species [56], and metabolism [57]. Additionally, yeast models have been employed to study how translation errors influence cellular fitness and proteostasis [50, 51, 58–61]. Yet, the combination of ageing and mistranslation has not yet been studied in yeast models of misfolded proteins, including TDP-43 and FUS.

To determine how ageing modulates the aggregation and toxicity of FUS and TDP-43, we establish optimized yeast models with low expression of FUS and TDP-43. These models allowed us to determine that mitochondrial respiration and β-oxidation enhance FUS and TDP-43 toxicity, but not necessarily their aggregation. Furthermore, we confirm proteostasis is impaired during the early prodromal phase of ageing, as evidenced by increased aggregation of a metastable reporter protein. By contrast, early ageing antagonizes FUS and TDP-43 aggregation while exacerbating the growth defect caused by FUS and TDP-43. Finally, we find that further impairing proteostasis in ageing cells by inducing translation errors synergistically prevents FUS and TDP-43 aggregation and exacerbates toxic phenotypes associated with FUS and TDP-43. Accordingly, our findings suggest that the loss of proteostatic capacity in early ageing cells reduces aggregation of FUS and TDP-43 and thus increases their toxicity in late-stage ageing. Our findings contribute to the growing literature indicating that protein aggregates are not necessarily the major toxic species in neurodegenerative diseases, highlighting the need to investigate misfolded monomer and oligomer species as the likely culprits. Finally, our findings genetically identify translation errors, such as those caused by tRNA variants, as potential contributors or risk factors for ALS/FTD and other age-associated protein misfolding diseases.

Results

Optimized low expression FUS and TDP-43 proteinopathy yeast models

Yeast models of FUS and TDP-43 aggregation and toxicity have been useful discovery tools for screening genetic interactors that antagonize toxicity [9, 10, 19, 20, 22, 27]. However, the high degree of toxicity caused by high expression levels of FUS and TDP-43 in these models has made identifying potential enhancers of toxicity challenging. To this end, we optimized models that express low levels of FUS and TDP-43 using the MET17 promoter and induce a milder growth defect than the commonly used GAL1 promoter (Additional File 1: Fig. S1).

Using our optimized expression system, we also explored how metabolic activity influences the toxicity and aggregation of FUS and TDP-43. We confirmed that oxidative phosphorylation induced by the metabolism of non-fermentable carbon sources (glycerol, potassium acetate, myristic acid, and oleic acid) exacerbates the growth defect of TDP-43 and FUS (Additional File 1: Fig. S2A-B) [24, 25]. Notably, the exacerbated growth defect of FUS and TDP-43 induced by oxidative phosphorylation did not correlate with an increase in inclusion formation (Additional File 1: Fig. S2C-F) or steady state levels (Additional File 1: Fig. S2G-J) of FUS-YFP and TDP43-YFP.

Our optimized yeast model demonstrates that altering metabolic activity changes FUS aggregation and toxicity. Similarly, we confirm previous findings indicating that respiratory metabolism enhances TDP-43 toxicity.

Increased FUS and TDP43 toxicity in chronologically aged cells

As post-mitotic cells, neurons must maintain proteostasis despite an accumulation of damage throughout their lifespan. Although protein aggregation is strongly associated with advanced age [36, 40], our options for monitoring protein aggregation in ageing mammalian cells are limited. Model organisms provide an effective alternative to study proteostasis and disease-associated protein aggregation in ageing cells. Unlike neurons, yeast cells age through two mechanisms: chronological and replicative ageing [62, 63]. Chronological ageing, which describes the lifespan of a yeast cell stationary phase [37, 55–57], more accurately recapitulates ageing experienced by neurons than replicative ageing, which describes the number of divisions a mother cell can sustain before senescence [64]. Using our optimized FUS and TDP-43 yeast models, we investigated whether FUS and TDP-43 exacerbate the toxicity associated with chronological ageing.

We first tested if ageing cells are sensitive to FUS and TDP-43 misfolding. We performed regrowth assays using cells expressing low levels of FUS or FUS-YFP that were aged for eight days. Of note, we performed our chronological ageing assays on cells starting at two days to control for the change in metabolism towards mitochondrial respiration associated with the diauxic shift [65]. Both FUS and FUS-YFP induced a growth defect in young cells (two days of ageing) (Fig. 1A). FUS and FUS-YFP also reduced the ability of cells to regrow after four and eight days of ageing (Fig. 1A). We then aged cells expressing FUS or FUS-YFP for eight days in caloric restriction media (0.2% glucose), which extends lifespan in yeast [57] and monitored the ability of these cells to regrow on standard media (2% glucose). While both FUS and FUS-YFP still induced a growth defect in young cells (two days of ageing) grown in caloric restriction media, eight days of ageing did not exacerbate this growth defect (Fig. 1B). We note an increase in the growth of cells expressing FUS after eight days of ageing in caloric restriction media likely due to a well-documented ‘gasping’ phenomenon whereby a subset of cells re-enter the cell cycle after prolonged ageing [66]. We next performed regrowth assays with cells expressing low levels of TDP-43 and aged for a time course of eight days. TDP-43 induced a growth defect in young cells and reduced the ability of cells to regrow after four and eight days of ageing (Fig. 1C). The growth defect associated with TDP-43 was exacerbated after four days of ageing in caloric restriction media but was not exacerbated after eight days (Fig. 1D).

Fig. 1.

Chronological ageing exacerbates growth defects associated with FUS and TDP-43. A, B Growth assays of cells expressing either vector control, FUS-YFP or FUS after two, four or eight days of ageing in media containing A 2% glucose or B 0.2% glucose. C, D Growth assays of cells expressing either vector control or TDP-43 in media containing C 2% glucose or D 0.2% glucose. E Flow cytometry measuring the percentage of cells stained for propidium iodide in the population of cells expressing either vector control or FUS aged for two, four or eight days in media containing either 2% glucose or 0.2% glucose. F Flow cytometry measuring the percentage of cells stained for propidium iodide in the population of cells expressing either vector control or TDP-43 aged for two, four or eight days in media containing either 2% glucose or 0.2% glucose. G Western blot of lysates from control cells or cells expressing FUS or TDP-43 after two and four days of chronological ageing. H–K Quantification of H FUS and TDP-43, I Hsp104, J Hsp42 and K Hsp26 steady state levels. The means (± SD) are represented graphically for three independent experiments. For flow cytometry, 200,000 cells were analyzed for each condition. The number sign “#” symbols indicate synergistic interactions when compared to the same sample at Day 2 (# p < 0.05, ## p < 0.01, ### p < 0.001, #### p < 0.0001)

We next tested whether FUS and TDP-43 exacerbated cell death associated with ageing. We used flow cytometry to measure propidium iodide staining as a proxy for cell death. We first used boiled cells as a positive control and generated a gate to measure the percentage of cells stained with propidium iodide (Additional File 1: Fig. S3). We then measured the percentage of dead cells in control cells and cells expressing FUS or TDP-43 after two, four, and eight days of ageing. Expression of FUS did not change the percentage of dead cells when compared to control cells at any point of ageing (Fig. 1E). In contrast, TDP-43 mildly reduced the percentage of dead cells when compared to control cells after eight days of ageing (Fig. 1F). Ageing cells in caloric restriction media reduced the percentage of dead cells after eight days of ageing, regardless of the expression of FUS or TDP-43 (Fig. 1E, F).

We performed western blots to test whether the steady state levels of FUS and TDP-43 change during the early prodromal phase of ageing preceding cell death (four days of ageing). Steady state levels of FUS and TDP-43 did not change when cells were aged (Fig. 1G, H). We then turned our attention to how FUS and TDP-43 influence expression of molecular chaperones in ageing cells. We found that expression of FUS and TDP-43 reduced steady state levels of Hsp104, Hsp42, and Hsp26 when compared to control cells after two days of ageing (Fig. 1G–K). Four days of ageing led to an increase in Hsp26 steady state levels, but not Hsp104 and Hsp42 levels, in FUS expressing cells (Fig. 1G–K). In contrast, four days of ageing led to an increase in Hsp104 steady state levels, but not Hsp42 and Hsp26 steady state levels, in TDP-43 expressing cells (Fig. 1G–K).

In sum, both FUS and TDP-43 impair the fitness of aged cells, although neither FUS nor TDP-43 exacerbate cell death caused by ageing. Furthermore, both FUS and TDP-43 reduce steady-state levels of Hsp104, Hsp42, and Hsp26, possibly further impairing proteostasis in ageing cells.

Ageing reduces FUS and TDP-43 inclusions

Neurodegenerative phenotypes have long been associated with the aggregation of misfolded proteins, like FUS and TDP-43, in ageing neurons. We sought to unravel the relationship between chronological ageing and the aggregation of FUS and TDP-43. To this end, we monitored the formation of FUS-YFP and TDP43-YFP foci preceding late-stage ageing that induces cell death (four days of ageing). Of note, the FUS and TDP-43 foci that form in yeast are not indicative of amyloid-like aggregates but instead most likely represent liquid-like detergent-sensitive condensates [8, 9, 15–17].

We aged cells expressing FUS-YFP in standard media (2% glucose) and caloric restriction media (0.2% glucose) for a time course of four days. Both the percentage of cells with FUS-YFP foci and the number of FUS-YFP foci per cell were reduced after four days of ageing (Fig. 2A–C). Caloric restriction also reduced the percentage of cells with FUS-YFP foci but masked the reduction in FUS-YFP foci caused by ageing (Fig. 2A–C). Although low levels of FUS or TDP-43 do not induce cell death, we hypothesized that the ability of cells to maintain FUS or TDP-43 aggregates during early ageing may influence the population of cells that die. To test this, we next stained cells expressing FUS-YFP with propidium iodide to assess the occurrence of FUS-YFP inclusions in dying cells. Across all conditions, cells stained with propidium iodide less frequently contained FUS-YFP foci when compared to their unstained counterparts (Fig. 2A, B).

Fig. 2.

Anti-correlation of ageing and cell death with aggregation of FUS and TDP-43. A Fluorescence microscopy of cells expressing FUS-YFP aged in media containing 2% or 0.2% glucose for two or four days and subsequently stained with propidium iodide. B Quantification of the percentage of cells containing FUS-YFP foci. C Quantification of the number of FUS-YFP foci per cell. D Fluorescence microscopy of cells expressing TDP43-YFP aged in media containing 2% or 0.2% glucose for two or four days and subsequently stained with propidium iodide. E Quantification of the percentage of cells containing FUS-YFP foci. F Quantification of the number of FUS-YFP foci per cell. The means (± SD) are represented graphically for three independent experiments. At least 500 cells were analyzed for each condition

We next aged cells expressing TDP43-YFP for four days. Similarly to FUS-YFP, both the percentage of cells with TDP43-YFP foci and the number of TDP43-YFP foci per cell were reduced after four days of ageing (Fig. 2D–F). In contrast to FUS-YFP, the percentage of cells with TDP43-YFP foci was not reduced by caloric restriction (Fig. 2D–F). In addition, when monitoring cells stained with propidium iodide, we find that cells stained with propidium iodide less frequently contained TDP43-YFP foci when compared to their unstained counterparts (Fig. 2D, E).

To confirm that ageing impairs proteostasis, we aged cells expressing metastable Firefly Luciferase-GFP (FFL-GFP) as a proteostasis reporter that is not directly linked to neurodegenerative diseases. The percentage of cells with FFL-GFP foci and the number of FFL-GFP foci per cell increased after four days of ageing (Additional File 1: Fig. S4A-C). To assess whether caloric restriction influences proteostasis decline in ageing cells, we aged cells expressing FFL-GFP in caloric restriction media. Caloric restriction did not change the percentage of cells with FFL-GFP foci, nor the number of FFL-GFP foci per cell (Additional File 1: Fig. S4A-C). As dead cells lose the ability to maintain FUS and TDP-43 inclusions, we sought to assess whether they also lose the ability to maintain FFL-GFP inclusions. Similar to FUS-YFP and TDP43-YFP, cells stained with propidium iodide showed a drastic reduction in the percentage of cells with FFL-GFP foci (Additional File 1: Fig. S4A-B).

In sum, our findings indicate a reduction in both FUS and TDP-43 aggregation in ageing cells. In contrast, FFL aggregation increases in ageing cells, both showing specificity and confirming impaired proteostasis in ageing cells. Finally, our findings reveal a negative correlation between FUS and TDP-43 aggregation and cell death.

FUS and TDP-43 reduce molecular chaperone solubility

Molecular chaperones are major modulators of protein folding and aggregation [67]. In particular, the AAA-ATPase disaggregase Hsp104 and the holdases Hsp42 and Hsp26 cooperate to regulate protein aggregation in yeast cells. Importantly, the loss of Hsp104, Hsp42, and Hsp26 impairs the fitness of ageing cells [62]. We explored whether FUS and TDP-43 dysregulate the function of Hsp104, Hsp42, and Hsp26 during early ageing, leading to the loss of aggregation.

First, we monitored the solubility of FUS and TDP-43 to confirm the loss of FUS and TDP-43 aggregation in ageing cells. We performed sedimentation assays using ageing cells expressing either FUS or TDP-43 to separate proteins into soluble (s) and insoluble (p) fractions. Both FUS and TDP-43 were less present in the insoluble fraction, and thus more soluble, after four days of ageing (Fig. 3A–C).

Fig. 3.

FUS and TDP-43 impair protein quality control in ageing cells. A. Sedimentation assay of FUS and TDP-43 in ageing cells. B, C Quantification of the ratio of B FUS and C TDP-43 in supernatant compared to the pellet. D–F Sedimentation assay of lysates extracted from D control cells or cells expressing E FUS or F TDP-43. G–I Quantification of the ratio of G Hsp104, H Hsp42 and I Hsp26 in the supernatant compared to the pellet. J Fluorescence microscopy of cells expressing FFL-mCherry and either vector control, FUS or TDP-43 after two and four days of ageing. K Quantification of the percentage of cells containing FFL-mCherry foci. L Quantification of the number of FFL-mCherry foci per cell. The means (± SD) are represented graphically for three independent experiments. For microscopy, at least 500 cells were analyzed for each condition. The asterisk “*” symbols indicate significant differences at Day 4 when compared to Day 2 (*p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001)

We next turned our attention to the solubility of Hsp104, Hsp42, and Hsp26 in ageing cells expressing FUS and TDP-43. Expression of FUS, but not TDP-43, reduced the solubility of Hsp104 when compared to control cells (Fig. 3D–G). In both control cells and, to a lesser extent, cells expressing FUS, Hsp104 became more soluble after four days of ageing (Fig. 3D, E, G). However, in cells expressing TDP-43, Hsp104 became less soluble after four days of ageing (Fig. 3F, G). Both FUS and TDP-43 reduced the solubility of Hsp42 when compared to control cells (Fig. 3D–F, H). While Hsp42 became less soluble after four days of ageing in control cells, Hsp42 became more soluble after four days of ageing in cells expressing FUS, but not TDP-43 (Fig. 3D–F, H). In similar fashion to Hsp42, both FUS and TDP-43 reduced the solubility of Hsp26 when compared to control cells (Fig. 3D–F, I). In both control cells and, to a lesser extent, cells expressing FUS, Hsp26 became less soluble after four days of ageing (Fig. 3D, E, I). In contrast, Hsp26 solubility did not change in cells expressing TDP-43 after four days of ageing (Fig. 3F, I).

We next monitored FUS-mCherry and TDP43-mCherry aggregation in cells deleted for Hsp104 (Δhsp104) or Hsp42 and Hsp26 (Δhsp42Δhsp26) and grown in media containing glucose. Loss of Hsp104 reduced the percentage of cells with FUS-mCherry and TDP43-mCherry foci and the number of FUS-mCherry and TDP43-mCherry foci per cell (Additional File 1: Fig. S5A-C). However, loss of Hsp42 and Hsp26 only reduced the percentage of cells with TDP43-mCherry foci and the number of TDP43-mCherry foci per cell (Additional File 1: Fig. S5A-C). We also performed growth assays to determine whether loss of FUS and TDP-43 aggregation affects growth on media supplemented with glucose. Similar to previous findings with high expression TDP-43 yeast models [8], loss of Hsp104 did not change the growth of cells expressing FUS or TDP-43 (Additional File 1: Fig. S5D-E). However, loss of Hsp42 and Hsp26 exacerbated the growth defect caused by FUS but rescued the growth defect caused by TDP-43 (Additional File 1: Fig. S5D-E).

As both FUS and TDP-43 modulate the solubility of molecular chaperones, we tested whether FUS and TDP-43 impair proteostasis during early ageing. We expressed a metastable FFL-mCherry reporter in ageing cells co-expressing FUS and TDP-43. FUS and TDP-43 do not increase the percentage of cells with FFL-mCherry foci, nor the number of FFL-mCherry foci per cell in young cells (Fig. 3J–L). However, after four days of ageing, FUS increased the number of FFL-mCherry foci per cell when compared to control cells (Fig. 3J, L). In stark contrast, TDP-43 decreased both the percentage of cells with FFL-mCherry foci and the number of FFL-mCherry foci per cell when compared to control cells (Fig. 3J–L).

In sum, both FUS and TDP-43 reduce the solubility of the molecular chaperones Hsp104, Hsp42, and Hsp26 in ageing cells. Furthermore, the loss of Hsp104, Hsp42, and Hsp26 function impairs the aggregation of FUS and TDP-43, indicating a potential mechanism underlying the reduced aggregation of FUS and TDP-43 in ageing cells. Finally, although both FUS and TDP-43 reduce the solubility of molecular chaperones, they have different effects on protein folding, as measured by FFL-mCherry aggregation.

Mistranslation exacerbates toxicity of FUS and TDP-43 in ageing cells

The ability of organisms to maintain translation fidelity is strongly linked to increased longevity [45–47]. Inducing translation errors (mistranslation) is detrimental to proteostasis [48, 51, 59–61], and can even modulate aggregation of endogenous proteins [49], metastable proteins [50], and proteins associated with neurodegenerative diseases like polyglutamine expanded Huntingtin protein [53] and FUS [54]. However, how mistranslation impacts proteostasis in ageing cells is poorly understood.

We sought to investigate whether mistranslating tRNA variants affect chronological lifespan of cells. To induce mistranslation, we expressed serine tRNA variants with anticodon mutations that induce proline to serine misincorporation at rates of approximately 0.5% (P > S weak) or 5% (P > S strong), or arginine to serine misincorporation at a rate of approximately 3.5% (R > S) [60]. This model of mistranslation is well described by ourselves and others to impair several proteostasis mechanisms in the cell [52, 68]. We performed growth assays using cells expressing mistranslating tRNA variants and aged over eight days in either standard media (2% glucose) or caloric restriction media (0.2% glucose). We confirm that both P > S strong and R > S induce growth defects in young cells (two days of ageing) in both standard media and caloric restriction media (Fig. 4A, B). The growth defect associated with P > S strong and R > S is exacerbated after eight days of aging in standard media, but not caloric restriction media (Fig. 4A, B). We stained mistranslating cells with propidium iodide to measure the percentage of dead cells. While mistranslating tRNA variants did not exacerbate cell death in cells aged in standard media, P > S strong and R > S partially antagonized the protective effects of caloric restriction (Fig. 4C, D).

Fig. 4.

Mistranslation exacerbates toxicity of FUS and TDP-43 in ageing cells. A, B Growth assays of cells expressing either vector control, wt tSer, or the mistranslating tSer variants P > S weak, P > S strong, or R > S after two, four, or eight days of ageing in media containing A. 2% glucose or B. 0.2% glucose. C, D Flow cytometry measuring the percentage of cells stained with propidium iodide in the population of cells expressing either vector control, wt tSer, or mistranslating tSer variants after two, four, or eight days of ageing in media containing C 2% glucose or D 0.2% glucose. E, F Growth assays of cells expressing either vector control, wt tSer, or mistranslating tSer variants and either E FUS or F TDP-43 after two, four, or eight days of ageing. G, H Flow cytometry measuring the percentage of cells stained with propidium iodide in the population of cells expressing either vector control, wt tSer, or mistranslating tSer variants and either E. FUS or F. TDP-43 after two, four, or eight days of ageing. The means (± SD) are represented graphically for three independent experiments. For flow cytometry, at least 150,000 cells were analyzed for each condition

We explored whether mistranslating tRNA variants augment the toxicity associated with FUS and TDP-43 in ageing cells. We aged cells co-expressing mistranslating tRNA variants and either FUS or TDP-43 for eight days. P > S strong and R > S synergistically impaired the ability of cells expressing FUS and TDP-43 to regrow on media supplemented with 2% glucose after eight days of ageing (Fig. 4E, F). P > S strong and R > S also induce synthetic cell death in cells expressing FUS and TDP-43 after four and eight days of ageing (Fig. 4G, H). Of note, P > S strong and R > S do not increase the steady state levels of FUS and TDP-43 (Fig. 4I–L).

In sum, mistranslating tRNA variants impair the fitness of ageing cells, but they do not exacerbate cell death when expressed in isolation in ageing cells. However, when expressed in addition to FUS and TDP-43, mistranslation synergistically impairs the fitness of ageing cells by inducing synthetic cell death.

Mistranslation reduces FUS and TDP-43 inclusions in ageing cells

Thus far, we have demonstrated an inverse correlation between ageing and inclusion formation of FUS and TDP-43, indicating a potential protective function of FUS and TDP-43 inclusions. We speculated that the synergistic toxicity of mistranslating tRNA variants and FUS and TDP-43 in ageing cells is due to an inability of cells to form protective inclusions of FUS and TDP-43 during the early phase of ageing. To test this hypothesis, we monitored FUS-YFP and TDP43-YFP aggregation in ageing cells expressing mistranslating tRNA variants.

P > S weak, P > S strong, and R > S all reduced the percentage of cells containing FUS-YFP foci, while P > S strong and R > S reduced the number of FUS-YFP foci per cell (Fig. 5A–C). After ageing cells for four days, P > S weak, P > S strong, and R > S led to a greater decrease in the percentage of cells with FUS-YFP foci and the number of FUS-YFP foci per cell when compared to cells expressing wt tSer (Fig. 5A–C). While mistranslating tRNA variants had no effect on the percentage of cells with TDP43-YFP foci, P > S strong and R > S both reduced the number of TDP43-YFP foci per cell (Fig. 5D–F). Although four days of ageing reduced both the percentage of cells with TDP43-YFP foci and the number of TDP43-YFP foci per cell regardless of the tRNA expressed, mistranslating tRNA variants did not change the percentage of cells with TDP43-YFP foci and the number of TDP43-YFP foci per cell after four days of ageing (Fig. 5D–F).

Fig. 5.

Translation errors reduce inclusion formation of FUS and TDP-43 in ageing cells. A Fluorescence microscopy of cells expressing FUS-YFP and either wt tSer or the mistranslating tSer variants P > S weak, P > S strong or R > S aged for two or four days. B Quantification of the percentage of cells with FUS-YFP foci. C Quantification of the number of FUS-YFP foci per cell. D Fluorescence microscopy of cells expressing TDP43-YFP and either wt tSer or the mistranslating tSer variants P > S weak, P > S strong or R > S aged for two or four days. E Quantification of the percentage of cells with FUS-YFP foci. F Quantification of the number of FUS-YFP foci per cell. The means (± SD) are represented graphically for three independent experiments. At least 500 cells were analyzed for each condition. (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001)

We used the metastable FFL-GFP reporter to test whether mistranslating tRNA variants reduces aggregation of FUS and TDP-43 by impairing proteostasis. P > S strong and R > S both increased the percentage of cells containing FFL-GFP foci, but not the number of FFL-GFP foci per cell in young cells (Additional File 1: Fig. S6A-C). After ageing cells for four days, P > S strong and R > S both led to a greater increase in the number of FFL-GFP foci per cell, but not the percentage of cells containing FFL-GFP foci, when compared to cells expressing wt tSer.

In sum, our findings suggest that mistranslating tRNA variants reduce aggregation of FUS and TDP-43 in both young and aged cells, supporting our hypothesis that FUS and TDP-43 inclusions are protective. In contrast, mistranslating tRNA variants increase aggregation of FFL-GFP in young and aged cells, confirming that mistranslation impairs proteostasis and indicating that impaired proteostasis inhibits aggregation of FUS and TDP-43.

Discussion

Like many other neurodegenerative diseases, Amyotrophic Lateral Sclerosis (ALS) is associated with the accumulation of aggregated proteins, such as Fused in sarcoma (FUS) [7] and Tar DNA-binding protein 43 (TDP-43) [1, 2, 6]. Accordingly, protein aggregates have been proposed to act as key drivers of neurodegeneration and age-related cell death, generally [1, 3, 69, 70]. However, many detailed studies challenge the notion that protein aggregation causes toxicity in models of ageing and neurodegeneration [30–33]. Additionally, the fidelity of protein biosynthesis is clearly a modifier of ageing [45–47] and protein aggregation [49–51], including the aggregation of proteins associated with neurodegenerative diseases [53, 54]. Yeast has been an effective model to study chronological ageing [37, 55, 57, 62], translation errors [50, 51, 59, 71], and the aggregation of neurodegenerative disease-associated proteins [8–10, 14, 18, 35]. Here, we explore the combined influence of ageing and translation errors on the aggregation of FUS and TDP-43 and their toxicity, which closely reflects the accumulation of different proteostasis stressors experienced over the lifespan of neurons in ALS.

Previously described FUS and TDP-43 yeast models expressed very high levels of FUS and TDP-43 [8–10, 14], which have delivered important insights into the aggregation and cytotoxicity of FUS and TDP-43. However, owing to their acute cytotoxicity, these models are not optimal for investigating metabolism and chronological ageing as modifiers of FUS and TDP-43 aggregation and their toxicity. We overcome these limitations by devising yeast models that express low levels of FUS and TDP-43, resulting in only mild growth defects. Although we successfully use these models to uncover a relationship between ageing, protein aggregation, and corresponding toxicity, our model is not without shortcomings. Chiefly, our ageing paradigm yields a mixed population of living and dead cells after eight days, making it challenging to assess the molecular underpinnings associated with end-stage ageing. Thus, we opted to focus our investigation on the earlier prodromal phase of ageing to decipher the molecular changes that precede and plausibly promote the fitness defects we observe after eight days of ageing. Our optimized yeast models also facilitate studies in media containing non-fermentable carbon sources, driving yeast cells to energy production via mitochondrial respiration [72], which mimics the metabolic activity in neurons [73, 74]. Similar to others [24, 25], we show that cells producing energy by oxidative phosphorylation or β-oxidation are sensitized to the expression of both FUS and TDP-43 compared to fermenting yeast. Notably, respiring cells frequently show a reduction in FUS and TDP-43 aggregates, suggesting an uncoupling of FUS and TDP-43 aggregation from their cytotoxicity. Therefore, we speculate that aggregation is not necessarily the key mechanism by which FUS and TDP-43 induce toxicity in respiring cells.

Our study further explores how chronological ageing alters the accumulation of FUS and TDP-43 aggregates and their toxicity. When cells are aged chronologically, we find that FUS and TDP-43 do not induce cell death but instead prevent aged cells from re-entering the cell cycle. Further, we reveal a negative correlation between proteostasis loss caused by ageing and FUS and TDP-43 aggregation. Notably, protein aggregation was not universally inhibited by ageing, as ageing increased the aggregation of a commonly used metastable reporter protein, Firefly Luciferase-GFP (FFL-GFP). Our findings indicate that FFL-GFP aggregation is driven by the inability of cells to maintain properly folded proteins, whereas the formation and maintenance of FUS and TDP-43 inclusions require functional molecular chaperones. Generally, molecular chaperones and protein remodeling complexes can mediate the aggregation of misfolded proteins through protective mechanisms [33, 39, 75, 76]. In addition, our findings may reveal a specific effect of ageing on the liquid-like RNA-containing condensates such as stress granules and P bodies that harbor TDP-43 and FUS [9, 15] compared to the quality control compartments that harbor FFL [77]. Specifically, the ability of TDP-43 and FUS to associate with RNA regulates both their aggregation and toxicity in yeast [8–10, 14] and prevents their conversion to solid, more stable aggregates in other model systems [78, 79].

Furthermore, the solubility of FUS and TDP-43 aggregates plausibly influences their toxicity. Although both FUS and TDP-43 form liquid-like aggregates in yeast, their conversion to more solid and likely stable aggregates can be protective [16]. Concordant with those findings, we show that cells that die during the ageing process consistently retain fewer FUS and TDP-43 aggregates compared to their surviving counterparts. Our findings contribute to the growing consensus that specific types of FUS and TDP-43 aggregates can perform protective functions, and that misfolded oligomeric species may be the drivers of toxicity [16, 80, 81]. Accordingly, we postulate that ageing cells are particularly sensitive to the misfolding of FUS and TDP-43, plausibly due to impaired proteostasis [36, 38, 82] and the associated inability to form or maintain protective aggregates [40]. Yet, our data does not rule out the less plausible possibility that dead or dying cells cannot form or maintain FUS and TDP-43 aggregates. We also assessed whether FUS and TDP-43 interact with proteostasis networks to impair their aggregation in ageing cells. Indeed, FUS and TDP-43 reduce the steady-state levels and the solubility of molecular chaperones, thereby likely reducing their protective functions in proteostasis and compromising longevity [39, 62, 82–86].

We furthermore investigate whether mistranslating tRNA variants exacerbate ageing phenotypes in our FUS and TDP-43 yeast models. Our findings indicate that, while mistranslating tRNA variants do not induce cell death, they prevent aged cells from re-entering the cell cycle. However, mistranslating tRNA variants partially reduce the protective effect of caloric restriction on age-related cell death. Remarkably, although neither mistranslating tRNA variants nor FUS and TDP-43 exacerbate age-related cell death on their own, the combination of mistranslation and FUS or TDP-43 induces synthetic toxicity in ageing cells. Notably, the stress caused by mistranslating tRNA variants closely mimics that of ageing, impairing proteostasis and increasing aggregation of FFL-GFP, while antagonizing the aggregation of FUS and TDP-43. We deduce that ageing, mistranslation, and expression of FUS or TDP-43 synergistically impair proteostasis, preventing formation of protective FUS and TDP-43 aggregates and foiling cellular survival. These findings support our hypothesis that the inability to maintain FUS and TDP-43 aggregates in ageing cells promotes their toxicity.

Conclusions

Most ALS patients accumulate aggregated proteins, like TDP-43 and FUS, in their neurons. The mechanism by which ageing modulates this aggregation, and whether these aggregates cause disease, has remained elusive. Here, we find that ageing, paradoxically but specifically, impairs aggregation of FUS and TDP-43, plausibly due to impaired proteostasis. We also highlight mistranslation as a modulator of FUS and TDP-43 toxicity in ageing cells. Based on our findings, we propose a model, whereby mistranslation and possibly other challenges to proteostasis, combined with ageing, impair aggregation, thus unleashing the cytotoxicity associated with smaller, more soluble misfolded FUS and TDP-43 conformers. Our research underscores the need to determine the molecular and cellular mechanisms underlying protective protein aggregation and explore if this model also applies to other misfolded proteins and neurodegenerative diseases.

Methods

Plasmids

All plasmids used in this study are described in Supplementary Table 1. The plasmids encoding TDP-43, FUS, and FUS-YFP under the control of the MET17 promoter were generated using cut-paste cloning by Analisa Echeverria (this study). The plasmid encoding TDP43-YFP was generated using Gateway cloning. The plasmid encoding FUS under the control of the GAL1 promoter was a kind gift from Dr. Aaron Gitler [9]. The plasmid encoding TDP-43 under the control of the GAL1 promoter was generated by Gateway cloning. The p426MET17 FFL_D50N,G119N-GFP plasmid was a kind gift from Dr. John Glover [87]. The plasmids encoding TDP-43, FUS, and FFL_D50N,G119N-mCherry were all generated using Gateway cloning using a pAG416GPD ccdB mCherry destination vector, which was designed using cut-paste cloning. Plasmids encoding tRNA variants were a kind gift from Dr. Chris Brandl (pSUP17-wt, pSUP17-P > S_weak, pSUP17-P > S_strong, pSUP17-R > S) [60].

Yeast strains, media, and culturing conditions

All experiments were performed using derivatives of the W303 yeast strain (leu2-3,112 trp1-1 can1-100 ura3-1 ade2-1 his3-11,15). Strains deleted for HSP104, or HSP42 and HSP26 were described previously [88]. Yeast cells were transformed with desired plasmids encoding either FUS, TDP-43, or FFL_D50N,G119N using the standard LiAc/PEG protocol [89]. Cells were cultured in and plated on agar plates containing synthetic defined (SD) media supplemented with 2% glucose, 6.7 g/mL yeast nitrogenous bases with ammonium sulfate and amino acids (40 mg/mL L-lysine, 20 mg/mL L-arginine, 10 mg/mL L-threonine, 60 mg/mL L-phenylalanine, 20 mg/mL L-isoleucine, 10 mg/mL L-methionine, 20 mg/mL adenine hemisulfate). Media was supplemented with additional amino acids for selection where necessary (20 mg/mL L-histidine, 60 mg/mL L-leucine, 20 mg/mL uracil, 80 mg/mL L-tryptophan). L-methionine was omitted from the media when necessary for induction of the MET17 promoter. For ageing experiments, cells were grown to saturation in a 3 mL liquid culture for 24 h before cells were pelleted and resuspended in media lacking methionine. Cells were then allowed to age chronologically for eight days. At the indicated time points, aliquots of the culture were taken to perform experiments. For caloric restriction, cells were grown to saturation in a 3 mL liquid culture for 24 h before the cells were pelleted and resuspended in media lacking methionine and containing only 0.2% glucose.

Growth assay

To assess growth defects, we performed growth assays as described by Petropavlovskiy et al. [90]. Briefly, cells were diluted to OD600 = 1, and five-fold serial dilutions were performed of each sample in a 96-well plate. Using a 48-pronged frogger, single spots of each sample were plated onto SD agar plates containing 2% glucose. When stated, cells were plated on agar plates that were modified to contain different carbon sources (2% galactose, 2% glycerol, 2% potassium acetate (KoAc), 0.1% myristic acid solubilized in 0.05% Tween20, or 0.1% oleic acid solubilized in 0.05% Tween20) instead of 2% glucose. Plates were incubated at 30 °C for 24–72 h before plates were photographed and analyzed using ImageJ. The densitometry of the spots at the same dilution factor was later compared. Experiments were repeated at least three times with independent transformants.

Propidium iodide staining and flow cytometry

To assess cell death, we performed propidium iodide staining as described by Chadwick et al. [62]. Briefly, 100 µL of cell culture was aliquoted into a labelled tube, pelleted, and resuspended in 100 µL of 5 µg/mL propidium iodide in PBS. After 10 min, the cells were pelleted and resuspended in PBS. The percentage of propidium iodide-stained cells was determined by flow cytometry using the BD Bioscience FACS Celesta flow cytometer. A 561 nm laser was used to identify cells stained with propidium iodide. A sample of cells that were boiled for five minutes was used to generate a gate corresponding to propidium iodide-positive staining. At least three independent experiments were performed for each sample, with a minimum of 150,000 cells being analyzed.

Fluorescence microscopy

To measure the formation of protein inclusions, we performed fluorescence microscopy using cells expressing either FFL-GFP, FFL-mCherry, TDP43-mCherry, TDP43-YFP, FUS-mCherry or FUS-YFP. Briefly, 100 µL of cell culture was aliquoted into a labelled tube and stained with propidium iodide as previously described. Cells were then imaged using the BioTek Cytation 5 Cell Imaging Multi-mode Reader at 20 × magnification. Images were taken using a GFP filter to visualize and assess inclusion formation of proteins of interest fused to GFP or YFP, and a TexasRed filter to visualize and assess inclusion formation of proteins of interest fused to mCherry or cells stained with propidium iodide. Experiments were repeated at least three times with independent transformants, and at least 400 cells were imaged and later analyzed. Downstream analysis of fluorescence microscopy images was performed in R using the EBImage package in R [91].

Protein extraction and western blot

To measure steady state levels of proteins, protein extraction was performed using a modified alkaline lysis protocol [92]. Briefly, 0.5 mL of OD600 = 2 cells were harvested and resuspended in 50 µL alkaline lysis buffer (0.1 M NaOH, 2% SDS, 50 mM EDTA) and boiled for five minutes. The lysate was then cleared at 21 k × g for 10 min, transferred to a fresh tube, and diluted in loading buffer and boiled again for five minutes. Protein lysates were then resolved by SDS-PAGE and transferred to PVDF membrane followed by immunoblotting with anti-FUS (Bethyl, A300-302A, 1:500), anti-TDP-43 (Novus, H00023435-M01, 1:500), anti-Pgk1 (Origene, AP21371AF-N, 1:5000), anti-Hsp104, anti-Hsp42, and anti-Hsp26 (all kind gifts from Dr. Johannes Buchner). Secondary antibodies used were either conjugated with Alexa Fluor 680 (Thermofisher, 1:5000) or horseradish peroxidase (Thermofisher, 1:5000). Blots were imaged, and the densitometry of bands was assessed and compared in relation to the Pgk1 loading control. Experiments were repeated at least three times with independent transformants.

Sedimentation assay

To assess solubility of proteins, a mild mechanical lysis was used to extract proteins. Briefly, 3 mL of OD600 = 2 cells were harvested and resuspended in 100 µL mild lysis buffer (100 mM Tris pH 7.5, 200 mM NaCl, 5% glycerol, 1 mM DTT, 5 mM EDTA, 50 µM NEM, and 10 µM PMSF). Cell suspension was then transferred to an Eppendorf tube containing 100 µL acid-washed glass beads (Sigma, G8772) and then disrupted for five 1-min cycles interspersed by 1-min incubations on ice. The lysate was then cleared at 200 × g for 2 min before being transferred to a new tube. An aliquot of the total lysate was taken before the remaining lysate was centrifuged at 500 × g for 15 min. The supernatant was aliquoted into a new tube, and the pellet was then resuspended in mild lysis buffer. All protein fractions were diluted with an equal volume of SUMEB (1% SDS, 8 M Urea, 10 mM MOPS, 10 mM EDTA). Equal volumes of each protein fraction were then resolved by SDS-PAGE followed by immunoblotting as described above.

Statistical analysis

All graphical representations of data and statistical analysis were performed using GraphPad Prism. Data were compared using either one or two-way ANOVA followed by Fisher’s LSD. All relevant p-values are stated.

Abbreviations

ALS

Amyotrophic Lateral Sclerosis

FTD

Frontotemporal Dementia

FUS

Fused in sarcoma

TDP-43

TAR DNA-binding protein 43

SOD1

Superoxide dismutase 1

FFL-GFP

Firefly luciferase-GFP

Authors’ contributions

D.W.M.: Investigation, Conceptualization, Formal analysis, Methodology, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing. N.C.: Investigation, Formal analysis. R.S.: Investigation, Formal analysis. M.L.D.: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing—review & editing. All authors read and approved the final manuscript.

Funding

This research was funded through a NSERC Discovery grant (RGPIN-2024-05867) and a McGill-Western Initiative for Translational Neuroscience (ITN) grant to M.L.D., and an ALS Canada Trainee Fellowship to D.W.M. This research was also supported by CIHR Skin Research Training Centre (201903). Funding to pay the Open Access publication charges for this article was provided by: ITN grant to M.L.D.

Data availability

All original western blots are included in supplementary data (Additional File 2). The remaining datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. The R script used to analyze fluorescence microscopy is available at (10.6084/m9.figshare.28688966).

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.