Glucocerebrosidase reduces the spread of protein aggregation in a Drosophila melanogaster model of neurodegeneration by regulating proteins trafficked by extracellular vesicles

Abnormal protein aggregation within neurons is a key pathologic feature of Parkinson’s disease (PD). The spread of protein aggregates in the brain is associated with clinical disease progression, but how this occurs remains unclear. Mutations in the gene glucosidase, beta acid 1 (GBA), which encodes the lysosomal enzyme glucocerebrosidase (GCase), are the most penetrant common genetic risk factor for PD and dementia with Lewy bodies, and also associate with faster disease progression. To explore the mechanism by which mutations in GBA influence pathogenesis of these diseases, we previously created a Drosophila model of GBA deficiency (Gba1b) that manifests neurodegeneration, motor and cognitive deficits, and accelerated protein aggregation. Proteomic analysis of Gba1b mutants revealed dysregulation of proteins involved in extracellular vesicle (EV) biology, and we found altered protein composition of EVs from Gba1b mutants. To further investigate this novel mechanism, we hypothesized that GBA may influence the spread of pathogenic protein aggregates throughout the brain via EVs. We found that protein aggregation is reduced cell-autonomously and non-cell-autonomously by expressing wildtype GCase in specific tissues. In particular, accumulation of insoluble ubiquitinated proteins and Ref(2)P in the brains of Gba1b flies are reduced by ectopic expression of GCase in muscle tissue. Neuronal expression of GCase also cell-autonomously rescued protein aggregation in brain as well as non-cell-autonomously rescued protein aggregation in muscle. Muscle-specific GBA expression rescued the elevated levels of EV-intrinsic proteins and Ref(2)P found in EVs from Gba1b flies. Genetically perturbing EV biogenesis in specific tissues in the absence of GCase revealed differential cell-autonomous effects on protein aggregation but could not replicate the non-cell-autonomous rescue observed with tissue-specific GBA expression. Additionally, we identified ectopically expressed GCase within isolated EVs. Together, our findings suggest that GCase deficiency mediates accelerated spread of protein aggregates between cells and tissues via dysregulated EVs, and EV-mediated trafficking of GCase may partially account for the reduction in aggregate spread. Author’s Summary Parkinson’s disease (PD) is a common neurodegenerative disease characterized by abnormal clumps of proteins (aggregates) within the brain and other tissues which can lead to cellular dysfunction and death. Mutations in the gene GBA, which encodes glucocerebrosidase (GCase), are the strongest genetic risk factor for PD, and are associated with faster disease progression. GCase-deficient mutant flies display features suggestive of PD including increased protein aggregation in brain and muscle. We found that restoring GCase protein in the muscle of mutant flies reduced protein aggregation in muscle and the brain, suggesting a mechanism involving interaction between tissues. Previous work indicated that GBA influences extracellular vesicles (EVs) – small membrane-bound structures released by cells to communicate and/or transport cargo from cell to cell. Here, we found increased aggregated proteins within EVs of mutant flies, which was reduced by restoring GCase in muscle. In addition, we found GCase within the EVs, possibly explaining how GCase in one tissue such as muscle could reduce protein aggregation in a distant tissue like the brain. Our findings suggest that GCase influences proteins within EVs, affecting the spread of protein aggregation. This may be important to understanding PD progression and could uncover new targets to slow neurodegeneration.


Introduction
Parkinson's disease (PD) is the most common neurodegenerative movement disorder, affecting 1-2% of people over 65 years of age [1]. PD is characterized by cardinal motor and non-motor symptoms, including rigidity, slowness of voluntary movements, and cognitive decline [2][3][4]. Intraneuronal Lewy bodies containing ubiquitinated proteins and a-synuclein are a hallmark pathologic finding in PD. The stereotypic spread of Lewy bodies in PD from the rostral brain stem to the midbrain and eventually throughout the neocortex suggests a prion-like mechanism mediating propagation of protein aggregates from neuron to neuron [5]. This temporo-spatial spread of Lewy bodies correlates with clinical progression of PD and has been replicated in several animal models [6][7][8].
Although much work has focused on identifying genes involved in PD, and how perturbation of these genes lead to PD pathogenesis, the mechanisms underlying PD are not yet completely understood.
Mutations in the gene glucosidase, beta acid 1 (GBA), encoding the lysosomal ceramide metabolism enzyme glucocerebrosidase (GCase), are the strongest genetic risk factor for PD and dementia with Lewy bodies, increasing risk of developing PD by approximately 5-fold in GBA mutation carriers compared to non-carriers [9][10][11]. GBA carriers with PD are otherwise clinically similar to idiopathic PD patients, with indistinguishable response to dopaminergic medications, slightly younger age of onset by about 4 years, and higher incidence of cognitive decline [12,13]. While some studies suggest that GBA carriers with PD may have a heavier burden of Lewy bodies than in non-carriers with PD, the neuropathologic features are similar [14]. Importantly, mutations in the gene GBA are common, having been found in 4-5% of all idiopathic PD patients [13,15]. Recent longitudinal clinical studies have revealed that in addition to increased risk, GBA carriers with PD have faster progression of both motor and cognitive symptoms compared to idiopathic PD patients [16][17][18].
To examine how GBA influences PD pathogenesis, we previously developed and characterized a Drosophila model of GBA deficiency (Gba1b) that manifests several phenotypes reminiscent of key features of PD, including neurodegeneration, locomotor deficits, cognitive deficits, and accelerated protein aggregation in multiple tissues including the nervous system and muscle [18]. Our proteomic analysis of Gba1b mutant flies revealed that proteins involved in extracellular vesicle (EV) biology were dysregulated, and EVs isolated from Gba1b Drosophila hemolymph revealed increased levels of aggregate-prone and EV-intrinsic proteins, indicating that GBA deficiency alters the protein composition of EVs [19]. EVs are a heterogeneous group of membrane-bound vesicles secreted by cells that can have multiple functions, including intercellular communication through protein and nucleic acid cargo and discard of cell components outside of the cell [20,21]. EVs can originate from the endosomal system as a result of fusion of the late endosome to the plasma membrane, releasing intraluminal vesicles as EVs into the extracellular matrix (exosomes), or from direct outward budding of the plasma membrane (microvesicles) [20,22,23]. EVs have been implicated in the propagation of misfolded proteins between cells in multiple neurodegenerative diseases, including PD [24][25][26][27][28][29]. a-synuclein has been found within EVs isolated from tissues of PD and dementia with Lewy bodies patients [24,30], and in vitro studies have suggested that a-synuclein within EVs may be more pathogenic than free extracellular a-synuclein [31].
Accordingly, we hypothesized that tissue-specific expression of wildtype (WT) GCase might reduce the protein aggregates that accumulate in Gba1b mutants.
Using our GBA-deficient Drosophila model [18,19], we examined whether GCase could be mediating propagation of protein aggregates from cell-to-cell via EVs. In this study, we found that tissue-specific expression of WT GCase in Gba1b mutants corrected the alterations in protein composition of GBA-deficient EVs, as well as protein aggregation in local and distant tissues. Perturbing two independent EV biogenesis pathways in Gba1b mutants resulted in tissue-specific cell-autonomous effects on protein aggregation and changes in EV protein composition but did not rescue protein aggregation in distant tissues. Finally, we observed ectopically expressed GCase in EVs, suggesting that trafficking of GCase within EVs may contribute to the observed non-cell-autonomous rescue. Our findings suggest that mutations in GBA result in alterations in the protein composition of EVs that promote enhanced cell-to-cell transmission of pathogenic protein aggregates. Moreover, our findings indicate that GCase can be packaged into EVs and trafficked between cells to reduce protein aggregation throughout an organism. These findings suggest a possible mechanism underlying the clinical finding that GBA mutation carriers not only have increased risk of developing PD, but also faster progression of disease.

Results
Protein aggregation in Gba1b mutants can be rescued cell-autonomously and noncell-autonomously by tissue-specific dGba1b expression.
Our prior work revealed accelerated aggregate accumulation and dysregulation of EVrelated proteins in Gba1b mutant flies suggesting that GCase deficiency may influence cell-to-cell spread of protein aggregates. To explore whether tissue-specific GBA expression could reduce aggregation levels in distant tissues, we attempted to rescue the accumulated ubiquitinated protein and Ref(2)P in the brains of Gba1b mutants by expressing WT Drosophila Gba1b (dGba1b) in non-neural tissue. Using the DMef-GAL4 driver to drive expression of WT dGba1b in muscles throughout the fly, insoluble ubiquitinated protein aggregate accumulation was reduced in both the thoraces and heads of Gba1b mutant flies ( Fig 1A&B). This non-cell-autonomous rescue of ubiquitinated protein aggregates was dramatically apparent in whole brains ( Fig 1C).
However, because DMef-GAL4 is also expressed in muscles located in the fly head that control the proboscis, we repeated this experiment using the Act88F-GAL4 driver. Act88F expression is concentrated in the thoracic indirect flight muscles and is not found in the head [32]. Expression of WT dGba1b using the Act88F-GAL4 driver in Gba1b mutants significantly reduced accumulation of insoluble ubiquitinated proteins in the thoraces and heads ( Fig 1E&F) and rescued their shortened lifespan ( Fig 1D).
Ref(2)P, the Drosophila ortholog of mammalian SQSTM1/p62, is important for selectively targeting ubiquitinated proteins for lysosomal degradation. While Ref(2)P is often used as a marker of autophagic flux, we have previously found it to accumulate in Gba1b mutants with no evidence of significantly impaired autophagy [19]. However, Ref(2)P is also required for aggregate formation under normal physiological conditions [33], and it has been shown to be released concomitantly with a-synuclein in EVs [30]; therefore we used its accumulation as a readout of protein aggregation. We anticipated that Ref(2)P accumulation in Gba1b mutants could also be non-cell-autonomously rescued. We indeed found that Ref(2)P accumulation was significantly decreased in both the thoraces and heads of Gba1b mutants expressing WT dGba1b in flight muscle ( Fig   1G&H).
To determine whether there might be a directionality or tissue-specificity to the noncell-autonomous rescue of protein aggregation, we expressed WT dGba1b in the nervous system and assessed for protein aggregates in the body. Neuronal expression of WT dGBA1b using the elav-GAL4 driver decreased both Ref(2)P accumulation and insoluble ubiquitinated proteins in both the heads and bodies of Gba1b mutants (Fig 2). These results suggest that GCase may normally function to reduce the spread of protein aggregates from one cell and tissue to another.

Non-cell-autonomous rescue of protein aggregation in Gba1b mutants is mediated by extracellular vesicles
There are multiple mechanisms that could mediate non-cell-autonomous interactions, including exocytosis of cytoplasmic components into the extracellular matrix, direct cellto-cell contacts, and release of cytoplasmic components via EVs. Given that our prior proteomic analysis of Gba1b mutants found evidence of dysregulation of EVs [19], we hypothesized that GCase influences the trafficking of aggregate-prone proteins in EVs.
We isolated EVs from hemolymph and confirmed that Ref (2)

P was increased in EVs from
Gba1b mutants compared to controls (Fig 3A), as we had seen previously [19]. We also found that proteins associated with EVs, Rab11 and Rab7, were elevated ( Fig 3B&C).
We tested the possibility that muscle-specific expression of WT dGba1b would reduce the increased levels of Ref(2)P and EV-intrinsic proteins from Gba1b mutants. Indeed, we observed a reduction in Ref(2)P, Rab11, and Rab7 levels in EVs isolated from Gba1b mutants expressing WT dGba1b in flight muscle using the Act88F-Gal4 driver (Fig 3A-C).
However, while ubiquitinated proteins were increased in Gba1b mutant whole flies, they were not significantly altered in EVs isolated from Gba1b mutants ( Fig 3D). These results indicate that expression of GCase in the muscles of Gba1b mutants restores normal EV content and EVs may mediate the spread of protein aggregates between tissues.

Gba1b mutants
We previously found that pan-neuronal RNAi knockdown of Endosomal Sorting Complexes Required for Transport (ESCRT) proteins involved in EV biogenesis decreased the accumulation of insoluble ubiquitinated proteins and Ref(2)P cellautonomously in the heads of Gba1b mutants [19]. We hypothesized that if the observed non-cell-autonomous rescue is mediated by EVs, then RNAi knockdown of ESCRT machinery might reduce protein aggregation in distant tissues in Gba1b mutants by decreasing production of dysregulated EVs that promote propagation of protein aggregates. We confirmed that neuronal RNAi knockdown of Multivesicular body subunit 12 (Mvb12), a component of the ESCRT machinery required for EV biogenesis [34,35], cell-autonomously decreased ubiquitinated proteins and Ref(2)P levels in heads ( Fig  4A&B). However, RNAi knockdown of Mvb12 in muscle did not rescue increased ubiquitinated proteins or Ref(2)P levels in the heads of Gba1b mutants (Fig 4C&D). This suggests that knockdown of ESCRT-dependent EV biogenesis may be specifically important for the cell-autonomous regulation of protein aggregation in neurons but is not involved in the non-cell-autonomous rescue of brain protein aggregation by musclespecific dGba1b expression.
We next examined knockdown of neutral sphingomyelinase (nSMase), a lipidmodifying enzyme important for the formation and release of EVs [36,37]. nSMase hydrolyzes sphingomyelin, producing phosphocholine and ceramide. This enzyme has been implicated in multiple cellular functions, including inflammatory responses, reaction to lung and cardiac pathology, synaptic regulation, and release of EVs independent of ESCRT machinery [38]. We again hypothesized that tissue-specific RNAi knockdown of nSMase in Gba1b mutants could reduce protein aggregation in distant tissues by reducing the production of dysregulated EVs promoting protein aggregation. Instead, we observed increased accumulation of ubiquitinated proteins and Ref (2) However, knocking down nSMase in neuronal tissue did not change Ref(2)P or ubiquitinated protein levels in the heads of Gba1b mutants ( Fig 5E&F). These results suggest that nSMase-mediated release of EVs may be important for the cell-autonomous regulation of protein aggregation in muscles but not neurons. We next examined whether EV cargo is altered in Gba1b mutants with RNAi knockdown of nSMase in muscle. We found increased Ref(2)P and ubiquitinated protein levels in EVs isolated from Gba1b mutants with nSMase knocked down in muscle, but no change in the levels of Rab11 or Rab7, which are already increased in EVs from Gba1b mutants (Fig 6). This suggests that nSMase may regulate the protein cargo of EVs but may not alter the abundance of EVs.

Ectopically expressed GCase is trafficked within extracellular vesicles
Our finding that GBA expression in muscle rescues the protein composition in EVs indicates that non-cell-autonomous rescue is mediated at least in part by preventing spread of aggregated proteins trafficked by EVs. However, EV-mediated trafficking of GCase could also be another contributing factor to the non-cell-autonomous rescue of protein aggregation in Gba1b mutants. To test this possibility, we investigated whether ectopically expressed GCase travels to distant tissues via EVs. Because there are no antisera that recognize Drosophila GCase, we examined ectopic expression of human GBA (hGBA) in Gba1b mutants. We previously found ubiquitous expression of WT hGBA to be sufficient to partially rescue the shortened lifespan of Gba1b mutants [18]. Here, we found that expression of WT hGBA in flight muscle also reduced ubiquitinated protein aggregation and Ref(2)P accumulation in Gba1b mutants in both thoraces and heads ( Fig   7A-D). EVs isolated from Gba1b mutants expressing hGBA in flight muscle were found to have decreased Rab11 levels compared to controls (Fig 8A&B), confirming that human GCase (hGCase) can also revert alterations in EVs due to dGba1b deficiency. These results support a functional equivalence between Drosophila and hGCase in reducing the formation of protein aggregates. Interestingly, we were also able to detect hGCase in both the thoraces and heads of Gba1b mutants expressing hGBA in flight muscle (Fig 7E), indicating that hGCase itself is able to travel to distant tissues. Furthermore, we detected hGCase in EVs isolated from hemolymph of flies expressing hGBA in muscle (Fig 8A).
To determine whether hGCase is localized within EVs or associated on the outer surface of EVs, we added Proteinase K to the isolated EV fraction. We found hGCase to be resistant to Proteinase K digestion of the EV fraction but degraded if EVs were first treated with a detergent to disrupt vesicular membranes before Proteinase K digestion (Fig 8C).
These results indicate that GCase can be incorporated into EV cargo, and that WT GCase may be trafficked by EVs to distant tissues to reduce protein aggregation.

Discussion
Many genetic influences of PD have now been identified, and much work has been focused on how these genes lead to protein aggregation through mechanisms such as protein misfolding and autophagy defects. However, none of these genes have been implicated in cell-to-cell spread of pathogenic protein aggregates, which closely correlates with clinical disease progression. Our proteomic analysis and non-cell-autonomous rescue of protein aggregation in Gba1b mutants has led us to hypothesize that GBA mutations may influence the rate of propagation of protein aggregates between neurons.
Our work suggests a link between GBA mutations and faster spread of intracellular protein aggregates via a novel EV-mediated mechanism, possibly explaining the recent clinical finding that GBA mutations accelerate the progression of clinical disease. Using a Drosophila model of GBA deficiency that manifests accelerated protein aggregation, we found that expressing WT GCase in specific tissues of a GBA-deficient fly can not only rescue protein aggregation cell-autonomously and in distant tissues, but also rescue alterations in protein cargo observed in EVs isolated from Gba1b mutant hemolymph.
Interestingly, ectopically expressed WT GCase itself was found within EVs of GBAdeficient flies, suggesting that the non-cell-autonomous rescue due to GCase expression is mediated by both reduction in aggregated proteins in EVs and trafficking of GCase via EVs to distant cells and tissues. Perturbing EV biogenesis through decreased expression of ESCRT machinery or ESCRT-independent nSMase affected protein aggregation in local tissues in a tissue-dependent manner. Together, these findings suggest that mutations in GBA result in the accelerated spread of protein aggregates through dysregulation of proteins trafficked by EVs.
Although our model of GBA mutations promoting spread of protein aggregates via EVs is novel, the idea that proteostasis can be maintained in a non-cell-autonomous fashion is well supported in the literature. For example, in C. elegans, misfolded asynuclein accumulating in endo-lysosomal vesicles was found to be transmitted from muscle to the hypodermis, a nearby tissue, for degradation [39]. It is possible that a noncell-autonomous mechanism is necessary because certain tissues may be more efficient in reducing protein aggregation. This has been previously described, where overexpression of FOXO in Drosophila muscle decreased aging-related protein aggregates in muscle as well as brain and other distant tissues, but FOXO overexpression in adipose tissue was unable to prevent protein aggregation in muscle [40]. In our model, overexpressing dGba1b in Drosophila muscle or neuronal tissue prevented accumulation of protein aggregates throughout the organism, however overexpression of WT GCase in midgut and fat body did not significantly reduce protein aggregation in the brain (S Fig 1). These discrepancies could be due to tissue-specific differences influencing biogenesis of EVs, such as metabolic rate or endovesicular trafficking intrinsic to specific cells.
Our unexpected results from perturbations of EV biogenesis machinery suggest that the EV-mediated regulation of protein aggregation is tissue-specific and complex.
Because an increase in EV-intrinsic proteins and alteration of protein cargo were observed in Gba1b mutants [19], we anticipated that genetic perturbations decreasing the These discrepancies between perturbation of EV biogenesis pathways and tissue types could be due to cell-specific compensatory mechanisms or intrinsic metabolic demands and solicit further investigation.
Our work suggests that GCase deficiency influences EV biogenesis to promote faster propagation of pathogenic protein aggregates throughout the tissues of an organism, which may be a compensatory response to cell-autonomous lysosomal stress. In the initial characterization of our Drosophila GBA-deficient model we found accelerated insoluble ubiquitinated protein aggregates, accumulation of Ref(2)P, and oligomerization of ectopically expressed human a-synuclein in Gba1b mutants, suggesting an impairment in lysosomal degradation [18,42]. A similar GBA-deficient Drosophila model also found evidence of lysosomal dysfunction, including enlarged lysosomes in GBA-deficient brains [43]. However, our proteomic analysis of Gba1b mutants did not support a profound impairment in autophagy, but instead suggested dysregulation of EVs with altered protein cargo which could be suppressed locally with knockdown of genes encoding ESCRT machinery important for EV biogenesis [19]. Based on these results, we believe that our initial observations of increased insoluble ubiquitinated proteins and Ref (2)

Drosophila strains and culture
Fly stocks were maintained on standard cornmeal-molasses food at 25°C. The

Preparation of Triton-soluble and insoluble fractions
Tissues from 10-day-old flies (6 females and 6 males per sample) were homogenized in Triton lysis buffer (50 mM Tris-HCl pH 7.4, 1% Triton X-100, 150 mM NaCl, 1 mM EDTA), and then spun at 15,000 x g for 20 min. The detergent-soluble supernatant was collected and mixed with an equal volume of 2x Laemmli buffer (4% SDS, 20% glycerol, 120 mM Tris-Cl pH 6.8, 0.02% bromophenol blue, 2% β-mercaptoethanol), and the same buffer was used to resuspend the Triton-insoluble pellet. All samples were boiled for 10 minutes.
The Triton-insoluble protein extracts were then cleared of debris by centrifugation at 15,000 x g for 10 minutes, followed by collection of the supernatant. At least three independent experiments were performed.

Western blotting
Proteins were separated by SDS-PAGE on 4%-20% MOPS-acrylamide gels (GenScript  (Fisher, 32106). Densitometry measurements of the western blot images were performed using Fiji software [45]. For homogenates, signal was normalized either to Actin or Ponceau S [46,47]. Normalized western blot data were log-transformed when necessary to stabilize variance before means were compared using Student t-test. Each experiment was performed at least three times.

Extraction of hemolymph and preparation of EV fractions
Hemolymph was obtained from 30 flies (15 males and 15 females, 10 to 11 days old) per sample. All flies were frozen with liquid nitrogen and decapitated by vortexing. Frozen flies were placed in a 1.7-mL tube containing a volume of PBS scaled to the number of flies used (2 μL/fly) and thawed for 5 min at room temperature. The tubes were then centrifuged at 5000 x g for 5 min at 4°C, after which the extracted hemolymph (supernatant) was centrifuged for 30 min at 10,000 x g at 4°C to remove cell debris and the cell-free supernatant was collected. The supernatant was then filtered with Ultrafree 0.22 μm spin filters (Fisher, UFC30GV0S) and centrifuged at 3000 x g for 5 min at 4°C; this was the EV fraction. In order to obtain whole-fly homogenate from the same animals used for collection of hemolymph, the bodies from four flies (2 males and 2 females) were homogenized in RIPA buffer (150mM NaCl, 1% Nonidet P-40, 0.5% Sodium deoxycholate, 0.1% SDS, 50mM Tris pH 8, diluted 1:1 with ddH20), centrifuged at 10,000 x g at 4°C for 5 min, and then the supernatant was transferred to a new tube. An equal volume of 2x Laemmli buffer (4% SDS, 20% glycerol, 120 mM Tris-Cl pH 6.8, 0.02% bromophenol blue, 2% β-mercaptoethanol) was added to the EV fractions and also to the whole-fly protein homogenates, and all samples were boiled for 10 min and then stored at −80°C. The experiment was repeated at least three times.
To determine if proteins were contained within EVs, before adding the Laemmli buffer, the EV fractions were incubated with PBS for 5 min at room temperature and then 0.5 μg/μL Proteinase K was added for 30 min at 4°C to digest external proteins or with 1% sodium dodecyl sulfate (SDS) for 5 min at room temperature and then 0.5 μg/μL Proteinase K was added for 30 min at 4°C to disrupt vesicular membranes and digest proteins. After incubation, Laemmli buffer was added and samples were boiled preparing them for western blotting.  Homogenates were prepared from fly thoraces and heads using 1% Triton X-100 lysis buffer. Western blot analysis was performed on the Triton X-100 insoluble proteins using