A monocarboxylate transporter rescues frontotemporal dementia and Alzheimer’s disease models

Brains are highly metabolically active organs, consuming 20% of a person’s energy at resting state. A decline in glucose metabolism is a common feature across a number of neurodegenerative diseases. Another common feature is the progressive accumulation of insoluble protein deposits, it’s unclear if the two are linked. Glucose metabolism in the brain is highly coupled between neurons and glia, with glucose taken up by glia and metabolised to lactate, which is then shuttled via transporters to neurons, where it is converted back to pyruvate and fed into the TCA cycle for ATP production. Monocarboxylates are also involved in signalling, and play broad ranging roles in brain homeostasis and metabolic reprogramming. However, the role of monocarboxylates in dementia has not been tested. Here, we find that increasing pyruvate import in Drosophila neurons by over-expression of the transporter bumpel, leads to a rescue of lifespan and behavioural phenotypes in fly models of both frontotemporal dementia and Alzheimer’s disease. The rescue is linked to a clearance of late stage autolysosomes, leading to degradation of toxic peptides associated with disease. We propose upregulation of pyruvate import into neurons as potentially a broad-scope therapeutic approach to increase neuronal autophagy, which could be beneficial for multiple dementias.

pyruvate import into neurons as potentially a broad-scope therapeutic approach to increase neuronal autophagy, which could be beneficial for multiple dementias.

Introduction
The brain is a highly metabolically active organ, using about 20% of the energy consumed by the human body [1]. In neurons glycogen storage is very reduced or non-existent, therefore they rely on a continuous supply of glucose to fuel their high energy requirements. Most of this glucose is used to maintain synaptic activity [2].
Interestingly, dementia progression is associated with a drop in glucose metabolism in the brain. This is well established in Alzheimer's disease (AD), where glucose hypometabolism precedes the onset of clinical symptoms of disease [3], worsens with disease progression and mirrors the pattern of brain atrophy [4]. A reduction in expression of several glucose transporters has been observed in the brains of mouse AD models [5] and of patients [6]. Alzheimer's disease is caused by the accumulation of extracellular plaques composed of Amyloidß (Aß) peptides, derived from the mis-processing of the Amyloid Precursor Protein (APP). Aß 1-42 is a major component of plaques [7], and dominant genetic mutations linked to familial AD, lead to an increase in its production [8]. Increase amounts of Aß 1-42 in animal models lead to pathology and neurodegeneration [9], underlying its causal role in disease development.
Patients with frontotemporal dementia (FTD), the second most common early onset dementia, also present with glucose hypometabolism in the frontal and temporal areas [10,11]. This hypometabolism pre-dates dementia in FTD linked Granulin (GRN) mutant carriers [12], spreads as disease progresses [13] and in Microtubule-Associated Protein Tau (MAPT) -linked FTD patients closely maps to areas of toxic MAPT accumulation [14], suggesting it closely tracks disease development. One study also noted that glucose hypometabolism seemed to precede structural abnormalities [15]. Moreover, carriers of a MAPT haplotype linked to an earlier onset show reduced glucose metabolism in frontal areas [16] and diabetes is an independent risk factor in FTD development [17], suggesting that glucose metabolism might drive disease development. Very recently, an FTD mouse knock-in model carrying a Triggering Receptor Expressed On Myeloid Cells (TREM2) mutation, showed a significant reduction in glucose metabolism, providing further which was one of the first models of disease to report the toxicity associated with C9 repeats [29], with expression of 36 or 100 repeats leading to a dramatic shortening in lifespan and eye degeneration, due to the generation of toxic DPRs [29].
In this work, we over-expressed putative monocarboxylate transporters and found that, bumpel, a member of the conserved family of the SLC5A sodium coupled monocarboxylate transporters, can dramatically rescue the phenotype of C9 expressing flies, and lead to a reduction in the levels of specific DPRs. We show that bumpel is indeed a lactate and pyruvate transporter and that its overexpression is associated with the clearance of late stage autolysosomes, which accumulate in response to C9 epxression.
Importantly over-expression of bumpel can also rescue an AD fly model, suggesting that an increase in monocarboxylate transporters in neurons could provide a broad ranging therapeutic intervention.

Bumpel, a monocarboxylate transporter, rescues C9 flies
In order to examine the relevance of monocarboxylate import to neurodegeneration in vivo, we over-expressed members of each family of transporters, facilitated and active, to determine if monocarboxylate import could have an effect on C9 toxicity in adult flies. Our C9 fly model expresses 36 C9 repeats pan-neuronally only in adult flies thanks to the Gal4-Gene-switch (GS) driver [30]. The gene-switch driver is induced when flies are fed the drug RU486, which is done after flies have eclosed, ensuring the expression of the repeats is limited to post-mitotic adult neurons, thus eliminating any confounding developmental effects. This is especially important in metabolic studies, as neurons are mostly glycolytic during embryonic stages and only switch to oxidative phosphorylation after birth [31]. Flies over-expressing 36 C9 repeats have a very short lifespan [29] and locomotion defects [32].
To test whether monocarboxylate import could ameliorate C9 associated phenotypes, we searched for transporters in the FlyORF collection of overexpression lines, which contains over 2000 lines, all inserted in the same site, as we wanted the lines to be comparably expressed. We identified lines over-expressing three genes homologous to the SLC16A family of transporters, hermes (hrm), outsiders (out) and chaski (chk), and one SLC5 transporter, bumpel. We over-expressed all these in adult neurons of our fly model expressing 36 C9 repeats, to see if they could ameliorate the lifespan shortening phenotype (Fig 1B-E). Whereas over-expression of the homologous genes to the SLC16A family hrm, out and chk did not rescue C9 toxicity (Fig 1B, C and D), the over-expression of bumpel leads to a dramatic increase in lifespan from medians of 20 days, to nearly 30 days medians ( Fig 1E). We confirmed the expression of bumpel by qPCR (Fig S1). Since both bumpel and C9 are driven by the same Gal4-GS driver, it is possible that binding to the UAS-bumpel promoter is titrating out the driver protein, leading to a drop in C9 expression and therefore, a phenotypic rescue. To confirm this is not the case we generated flies carrying an empty vector inserted in the same location as the FlyORF constructs used above, this also contains the binding sites for the driver and therefore could account for any dilution effects. Expression of the "empty" vector has no effect on lifespan of C9 expressing flies (Fig S1B), confirming that the rescue is due to bumpel and not to a non-specific dilution of the driver.
Bumpel was also able to rescue a 49 pure repeat fly model [33] (Fig S2), confirming its ability to rescue a variety of repeat lengths. Bumpel could also rescue motor phenotypes of C9 flies. C9 expressing flies display an early hyperactivity phenotype, with increased activity around day 4, this was brought back down by cooverexpression of bumpel (Fig 1F), showing that bumpel could rescue an early as well as late phenotypes associated with C9 expressing flies.
Bumpel, together with its paralogues kumpel and rumpel, is expressed in glia in flies, where it is thought to promote transport of substrates across the brain [34].

Bumpel reduces DPR accumulation
Bumpel's rescue could be either due to a specific rescue of C9 toxicity or to a general improvement of organismal health. To distinguish between these possibilities we over-expressed bumpel in the neurons of a wild type fly and found that the lifespan was slightly shortened (Fig 2A), suggesting that the rescue is specific to C9 toxicity and is not due to a general improvement in health. C9 toxicity in flies is primarily driven by DPR accumulation [29], we therefore measured the levels of DPRs in C9 flies. We found that both GR and GP (Fig 2B and C) were reduced in C9 fly heads, indicating that the rescue is due to a downregulation of toxic DPR proteins. This was not due to a general drop in transcription from the elav promoter as the levels of the Gal4GS driver were actually increased by the presence of bumpel ( Fig   S3), or to general reduction in translation efficiency as levels of an mCD8-GFP protein, also driven by the elav-Gal4GS driver was not affected by bumpel expression (Fig 2D).
The decrease in DPR levels could either be due to bumpel specifically affecting RAN translation from the C9 repeats or due to an increase in DPR degradation. To distinguish between these possibilities, we checked the ability of bumpel to affect the lifespan of flies expressing and translating GR alone, from a standard ATG start codon, without any underlying repetitive sequence able to undergo RAN translation [29]. GR is the most toxic DPR produced by the C9 sense repeats. We found that bumpel was able to strongly rescue GR toxicity (Fig 2E). We also found that bumpel is able to rescue PR toxicity (Fig 2F), however bumpel over-expression was not able to rescue GA toxicity (Fig 2G), suggesting the rescue is limited to arginine containing DPR proteins.

Bumpel is a pyruvate importer
Based on sequence homology, bumpel is predicted to transport both lactate and pyruvate [34], to confirm this we drove bumpel in larval fly salivary glands expressing lactate sensor Laconic (Fig 3A) or the pyruvate sensor Pyronic (Fig 3B), and exposed isolated tissue to either lactate or pyruvate. In both cases the salivary glands expressing bumpel showed a faster increase of monocarboxylate import (for lactate control vs bumpel: 0.1073 min −1 vs. 0.1717 min −1 ; for pyruvate: 0.0037 min −1 vs. 0.010 min −1 ) and to a higher extent (for lactate over-expression of bumpel a 74% increase and for pyruvate a 152% increase). This demonstrates that bumpel can transport both lactate and pyruvate.
To check whether an increase in pyruvate import was responsible for the drop in DPRs we injected a mixture of sodium pyruvate and ethyl pyruvate, which will cross membranes more readily and therefore the blood brain barrier, and reach the CNS better. Injection of adult 36R expressing flies induced on RU for 1 or 2 days, followed by 1 day recovery, showed a reduction in GP levels ( Fig 3C). These experiments suggest that it is the increased import of pyruvate that leads to a drop in DPR levels.
Pyruvate in the cell can be readily metabolized to Acetyl-CoA by Pyruvate dehydrogenase (Pdha) and enter the TCA cycle, it can also be converted to lactate by lactate dehydrogenate (ldh). We tested if imported pyruvate needed to be metabolized to exert its effect by knocking down ldh and Pdha by RNAi. Knock-down of both ldh and Pdha had no effect on the ability of bumpel to down-regulate GP levels ( Fig 3D, E, S4), suggesting it is the metabolite itself that is leading to DPR degradation, not a down-stream product of its metabolisation.
Upregulation of mitophagy has been shown to lead to a drop in proteins associated with neurodegenerative disease which localise to mitochondria: experiments in worms have shown that pharmacological increase in mitophagy can indeed reduce the levels of Aß [38], which has been found to localise to mitochondria [38]. GR has also been shown to localise to mitochondria [39], therefore an upregulation in mitophagy could lead to the changes in toxic protein levels in C9.
We found that bumpel over-expression led to an increase in Pink and Parkin mRNAs, two key regulators of mitophagy of damaged mitochondria [40] (Fig 4A and   B) and a drop in mitochondria copy number in heads, measured by the ratio of mtDNA to nuclear DNA (Fig 4C, S5A). This was confirmed by looking at levels of mito-GFP, which localises to mitochondria and found it to be reduced in the presence of bumpel, whereas a membrane localised GFP, mCD8-GFP was not affected (Fig 4D), suggesting a potential increase in mitophagy, or a drop in mitochondrial biogenesis in bumpel over-expressing flies.
We checked whether downregulation of Pink1, which regulates stress induced mitophagy [40], could affect DPR levels. Pink1 RNAi had no effect on GP levels in C9 or C9 bumpel expressing flies (Fig 4E), potentially suggesting that mitophagy does not play a role in regulating DPR levels. Pink1 is known to regulate stress induced mitophagy, however it's role in basal mitophagy is less clear [41]. To confirm the role of mitophagy in modulation of DPR levels, we upregulated mitophagy pharmacologically with the potent inducer Kaempferol [42], this led to an increase in GP in 36R expressing flies, indicating that upregulation of mitophagy does not lead to a reduction of DPR levels ( Fig 4F). This demonstrates that mitophagy upregulation is not a mechanism which can lead to a reduction in DPRs and therefore can not be responsible for the rescue observed in the bumpel expressing flies. Import of pyruvate in fly salivary glands has also been shown to lead an increase in autophagy [43] and there are several lines of evidence from the literature to suggest that autophagy induction can indeed modulate DPR accumulation. C9orf72 protein has been shown to regulate endosomal trafficking and autophagy [44,45] and a reduction in basal levels of autophagy has been observed in patient derived cell lines, due to C9orf72 haploinsufficiency [46] and this facilitates the accumulation of DPRs in cells [47]. Treating cells expressing 80 C9 repeats or individual DPRs with compounds inducing autophagy can lead to a reduction in DPRs [47,48].
We decided to monitor autophagy in flies expressing C9 repeats. Autophagy is initiated by the engulfment of cellular material by autophagosomes, which, in later stages go on to fuse with lysosomes to form autolysosomes. We visualised autophagosomes with Atg8-mcherry and lysosomes with lysotracker dye. This allowed us to visualise both autophagosomes, only marked by Atg8-mcherry, and autolysosomes, marked by both Atg8-mcherry and lysotracker [49]. We found that expression of C9 repeats led to an increase in Atg8-mcherry vesicles, in particular autolysosomes, indicating a potential block in late stage autophagy (Fig 5 A, B and C), which could lead to accumulation of DPR proteins. Defects in late state autophagy and lysosomes have often been linked to neurodegenerative diseases [50], including in C9orf72 expansion models [51].
We then checked if modulation of autophagy was able to alter the levels of DPRs.
We over-expressed and downregulated the levels of Atg1, an essential mediator for autophagy initiation [52], in adult flies expressing 36 C9 repeats. We found that upregulation of Atg1 did indeed decrease the levels of GP in C9 expressing flies ( Fig   5D), albeit not to the same extent as bumpel, whereas downregulation of Atg1 increases GP levels ( Fig 5E). This suggests that autophagy can modulate DPR levels.
Modulation of Atg1 when bumpel was co-overexpressed, however, did not affect GP levels ( Fig 5D, E), potentially suggesting that bumpel and Atg1 act in the same pathway and that bumpel modulates autophagy either downstream of Atg1 or independently of Atg1. QPCR confirmed that Atg1 was indeed reduced by the RNAi line used (Fig S6).
To confirm the role of autophagy in the modulation of C9 phenotypes, we downregulated Atg1 in C9 flies and found that it shortened the lifespan of C9 expressing flies, demonstrating how autophagy can modulate C9 phenotypes (Fig 5F). In the presence of bumpel, Atg1 RNAi did not affect the lifespan of C9 expressing flies ( Fig   5G), again suggesting bumpel and Atg1 act in the same pathway, and that bumpel is likely to activate autophagy either downstream or independently of Atg1. If bumpel was acting in the same pathway as Atg1, over-expression of both Atg1 and bumpel together could lead to a synergistic rescue. Atg1 over-expression in C9 flies alone did not affect phenotype ( Fig 5H). However, when we over-expressed Atg1 in the presence of bumpel we found an increased rescue relative to bumpel alone, suggesting the two proteins are synergising together to rescue C9 phenotypes ( Fig   5I).
We then monitored the ability of bumpel to modulate the autolysosome block observed in C9 expressing neurons. The bumepl transgene in the FlyORF stock is inserted into the ZH-86Fb site which carries a 3xP3-RFP construct, giving very strong red fluorescence in the brain. This RFP construct interferes with any imaging analysis. It is flanked by two loxP sites which we could excise using Cre recombinase. For image analysis we therefore used the bumpel stock no longer carrying the 3xP3-RFP.
We found that over-expression of bumpel did not decrease overall autophagosome numbers, but was able to specifically rescue the increase in autolysosomes (Fig 5A, C), suggesting it facilitates the clearance of autolysosomes, leading to degradation of DPRs.
The highly conserved transcription factor EB (TFEB) is a master regulator of lysosomal biogenesis and autophagy [53], and has been shown to facilitate the degradation of toxic proteins associated with neurodegenerative diseases in mammalian disease models [53], and in flies [54]. In C9 models nuclear-cytoplasmic transport defects lead to the mis-localisation of TFEB to the cytoplasm [51], leading to autolysosomal dysfunction. This would lead to a mis-regulation of TFEB target genes. The fly homologue of TFEB, Mitf, has been shown to co-ordinate the expression of a number of genes involved in lysosomal biogenesis, autophagy as well as lipid metabolism [54]. Of 38 genes described to be regulated by the Mitf, 16 are differentially expressed in a RNA seq dataset of fly heads expressing 100GR repeats [32], of these 16, 15 are downregulated ( Table 1), suggesting that Mitf's activity is downregulated in GR100 expressing flies. The toxicity in 36R expressing flies is mostly derived from the GR dipeptide repeat protein [29], and bumpel was able to rescue GR toxicity. We checked whether in C9 there was also a drop in Mitf targets and whether bumpel was able to restore the expression of Mitf target genes when co-overexpressed in C9 flies. Indeed, 4 Mitf targets, known to localise to the lysosome, were downregulated in response to C9 expression and their expression was increased in the presence of bumpel (Fig 5J), however a Mitf target not involved in lysosomal biogenesis, Cyp4d1, was not rescued (Fig S7) suggesting the rescue might be specific to Mitf targets involved in lysosomal biogenesis.
These experiments suggest that bumpel over-expression leads to relief of a late lysosomal block within C9 expressing neurons, possibly via Mitf activation, leading to a drop in the levels of DPR proteins and amelioration of the C9 disease phenotype ( Fig 5K).

Bumpel rescues an Alzheimer's disease fly model
We have found that an increase in pyruvate import is beneficial to an FTD/ALS fly model. Alzheimer's disease is also characterised by a drop in glucose metabolism in patients' brains [3] and late autolysosomal defects [55].
We wondered if increased transport of pyruvate could potentially be a broadspectrum therapy and whether the over-expression of bumpel could be beneficial to an AD fly model. To explore this question, we over-expressed bumpel in a fly model expressing two copies of Aß42 [56], linked to signal peptide ensuring secretion.
These flies express high levels of Aß which can be secreted [57] and aggregate in Thioflavin S stained structures [56]. Similarly to our C9 model, we limit the expression of Aß to adult neurons, these flies have a shortened lifespan and a climbing defect as they age [58]. Interestingly, over-expression of bumpel ameliorated the lifespan and climbing phenotype of Aß expressing flies, and led to a reduction in Aß levels ( Fig 6A, B and C), showing that bumpel can rescue an AD fly model too, and similarly to the C9 models, bumpel reduces toxic proteins associated with disease.

Discussion
Glucose hypometabolism of brain areas affected by disease is a common feature of many neurodegenerative diseases [59]. Glucose metabolism is tightly coupled between neurons and glia in brain, with lactate being shuttled from glia to neurons, to provide fuel for oxidative phosphorylation [23]. We have shown that upregulation of a monocarboxylate importer can ameliorate both C9orf72 repeat expansion toxicity and Aß toxicity in fly models. Both these diseases are caused by the accumulation of toxic peptides and in both cases the amelioration was linked to a drop in the toxic protein levels, suggesting that the phenotypic improvements are due to a downregulation of the toxic peptides.
The rescue was however not ubiquitous, toxicity due to the over-expression of poly(GA), one of the toxic DPRs associated with C9 repeats, was not rescued by bumpel over-expression, suggesting that the reduction of toxic peptides is specific to certain proteins.
Interestingly, over-expression of proton coupled lactate/pyruvate transporters did not rescue C9 phenotypes. It is possible that bumpel, since it is sodium coupled, would allow for constant import of lactate and pyruvate into neurons, even against a lactate/pyruvate concentration gradient, which proton coupled transporters might not allow. On the other hand, it is also possible that bumpel has a lower (or higher) affinity for substrates than other transporters, which could also explain the difference in behaviour. We show that the decrease in DPRs is associated with pyruvate import and that pyruvate itself is likely responsible for the rescue, and not a downstream metabolic process. Pyruvate has been shown to be able to modulate signalling pathways in the brain [26,60], such as HIF1 transcription, or erythropoietin signalling pathway and plays a protective role following cerebral injury [61,62].
Recently, import of pyruvate into cells has been associated with an induction of autophagy and mitophagy in cells [35][36][37] and flies [43]. We did indeed observe a reduction in mitochondrial copy number, when we co-overexpressed bumpel suggesting an increase in mitophagy. However, modulation of mitophagy did not appear to lead to a change in DPR levels, suggesting that it is unlikely to be a primary mechanism by which bumpel leads to DPR degradation.
Autophagic impairments have been associated with a number of AD models and an accumulation of autophagic vacuoles have been observed in patients' brains [63].
Autophagic defects are thought to contribute to disease progression and facilitate the accumulation of amyloid beta [64], possibly because of faulty autolysosome acidification [55]. Several studies have shown that upregulation of autophagy can ameliorate Aß pathology and phenotypes in mice and cells [64][65][66], suggesting that an upregulation of autophagy can indeed lead to the clearance of Aß peptides.
The same has been seen in C9 models, where defects in autophagy, especially linked to C9orf72 protein reduction in disease, can lead to an accumulation of DPR pathology [47], and upregulation of autophagy in cells can help with DPR clearance [47]. Increasingly, neurodegenerative disease have been linked to late autolysosomal defects [67] and these have also been observed in C9 models of disease [51].
We also observe an aberrant accumulation of autolysosomes in our C9 expressing fly neurons and over-expression of bumpel was able to promote autolysosomal clearance, suggesting it is acting late in autophagy to restore a healthy autophagic flux. This could be mediated by TFEB, a transcription factor regulating expression of many lysosomal genes [68], which has been shown to promote clearance of toxic peptide in a number of neurodegeneration models [69].
We confirmed C9 repeat expression impairs TFEB function, as seen in previous studies [51], and we find bumpel can restore the expression of fly TFEB target genes, specifically v-ATPase subunits, which could restore lysosomal function. How bumpel leads to a restoration of TFEB function will require further investigation.
Studies in human cells have linked expression of bumpel homologues and pyruvate import to histone deacetylase inhibition [70,71]. TFEB's activity is regulated by a number of acetylates and de-acetylases [72], so this could be a potential link, but it remains to be established if this is the case here.
We also find that increased autophagy, can reduce DPR levels, and that combined induction of early and late autophagy (by combining Atg1 with bumpel) leads to a synergistic rescue of C9 toxicity, suggesting that boosting autophagy with multiple interventions can be even more effective than a single intervention.
Pyruvate import might also be affecting neurons in other ways. Import of pyruvate into mitochondria could lead to an increase in oxidative phosphorylation and reactive oxygen species (ROS) formation, this, through a hormetic mechanism has been shown to boost cellular defence pathways [73]. An excessive increase in ROS formation could also explain the reduced lifespan when bumpel is over-expressed in a physiological ageing context, without disease.
Over-expression of bumpel is therefore a method to effectively increase autophagy This study overall points to a novel and unexplored role of pyruvate importers as a broad-spectrum modulator of toxic proteins associated with dementia, via neuronal autophagy enhancement, which could provide a novel therapeutic avenue.

Fly husbandry and stocks
All flies were reared at 25ºC on a 12:12 light: dark cycle at constant humidity and on standard sugar-yeast medium (15g/L agar, 50 g/L sugar, 100 g/L autolysed yeast, 100g/L nipagin and 2ml/L propionic acid). Adult-onset, neuron-specific expression of UAS constructs was achieved as described in [29]. The UAS-Aßx2 stock was a gift from Pedro Fernandez Funez (University of Minnesota) [56].
To remove the 3xP3-RFP for image analysis, we crossed a UAS-bumpel, elavGS homozygous fly with a fly carrying Cre recombinase and a third chromosome balancer, and selected flies which had lost the RFP fluorescence in the eye, we confirmed these flies were still long lived.

Lifespan analysis
All stocks used for lifespan analysis were back-crossed into a standard w1118 (for over-expression constructs) or v-w1118 stock (for Trip lines) for 6 generations to ensure homogenous genetic back-grounds. Male and female flies were allowed to mate and lay eggs for 24 hours on agar grape plates with yeast. The eggs were collected and seeded at standard density in 50ml bottles with SYA. After eclosion, flies were allowed to mate for 24-48 hours. At least 110-150 females of the appropriate genotype were split into groups of 15 and housed in vials containing SYA medium with or without drugs. Deaths were scored and flies tipped onto fresh food 3 times a week. Data are presented as cumulative survival curves, and survival rates were compared using log-rank tests. All lifespans were performed at 25ºC.

Analysis of activity and sleep
Individual, 4 day-old, mated, female flies were placed in 65 x 5 mm glass tubes containing standard 1xSYA, and activity was recorded using the DAM system (Drosophila Activity Monitoring System, TriKinetics, Waltham, MA) as described previously [75]. Flies were entrained to a 12:12 hour LD cycle at 25 °C and 65% humidity 24-36 hours before recording. Activity data are represented by mean values with their SEM. 32 flies were scored per genotype.  were then injected into the abdomen of each fly using Drummond "NANOJECT II"

Western Blotting
Automatic Nanoliter Injector (3-000-206A). All the flies were put back onto RU food for another 24 hours after injection and then flash-frozen by liquid nitrogen. Flies were processed for GP western as described above.

Microscopy
Fly brains expressing Atg8-mcherry were dissected in Schneinder's media and week, flies were tapped down to the bottom and allowed to climb. The whole process was recorded by a camera, and the videos were analysed by the automatic quantification system FreeClimber [79]. Vertical velocities of the fly populations were obtained and plotted in GraphPad Prism 9.

Statistical analysis
Lifespans were compared by log-rank test, performed in Excel. Two way comparisons were performed by unpaired t-test and multiple comparisons, following a one way ANOVA, were performed either by Tukey , Šídák's or Dunnett's multiple comparisons test as appropriate. Comparisons between groups were performed by 2-Way ANOVA (Fig 5J). Tests were performed in GraphPad Prism 9.
When multiple replicates of the same experiment were combined, we compared the effect of the experimental replicate and the effect of the treatment by 2 way ANOVA.
since the variability between experimental replicate was not significant, we combined the values from multiple experimental replicates together (this was done for Fig 5B   and 5C).

Acknowledgments:
This We thank Nazif Alic for useful feedback during the preparation of the manuscript.     Autophagy initiates by the formation of a phagophore, engulfing aggregated proteins, including DPRs, to generate a autophagosome, which then fuses with a lysosome to generate an autolysosome, where the internal pH drops and proteases are induced, degrading organelles and proteins, and releasing the components back to the cytoplasm for re-cycling. DPR stall the maturation of the autolysosomes, components are not degraded and there is an increase in stalled autolysosomes, as well as DPRs, leading to a toxic feed-back loop (upper panel). In the presence of bumpel, pyruvate is imported, and this relieves the block in late autolysosomal maturation, allowing components to be degraded, autolysosomes to be cleared and leading to a drop in DPRs (lower panel).