Glucocerebrosidase Deficiency in Drosophila Results in α-Synuclein-Independent Protein Aggregation and Neurodegeneration

Mutations in the glucosidase, beta, acid (GBA1) gene cause Gaucher’s disease, and are the most common genetic risk factor for Parkinson’s disease (PD) and dementia with Lewy bodies (DLB) excluding variants of low penetrance. Because α-synuclein-containing neuronal aggregates are a defining feature of PD and DLB, it is widely believed that mutations in GBA1 act by enhancing α-synuclein toxicity. To explore this hypothesis, we deleted the Drosophila GBA1 homolog, dGBA1b, and compared the phenotypes of dGBA1b mutants in the presence and absence of α-synuclein expression. Homozygous dGBA1b mutants exhibit shortened lifespan, locomotor and memory deficits, neurodegeneration, and dramatically increased accumulation of ubiquitinated protein aggregates that are normally degraded through an autophagic mechanism. Ectopic expression of human α-synuclein in dGBA1b mutants resulted in a mild enhancement of dopaminergic neuron loss and increased α-synuclein aggregation relative to controls. However, α-synuclein expression did not substantially enhance other dGBA1b mutant phenotypes. Our findings indicate that dGBA1b plays an important role in the metabolism of protein aggregates, but that the deleterious consequences of mutations in dGBA1b are largely independent of α-synuclein. Future work with dGBA1b mutants should reveal the mechanism by which mutations in dGBA1b lead to accumulation of protein aggregates, and the potential influence of this protein aggregation on neuronal integrity.


Introduction
Gaucher's disease (GD), the most common lysosomal storage disorder, is caused by recessive mutations in the glucosidase, beta, acid 1 (GBA1) gene, which encodes the lysosomal enzyme glucocerebrosidase [1,2]. GD has traditionally been categorized into neuronopathic and nonneuronopathic subtypes. However, recent work has challenged this classification by revealing that both subtypes are associated with a significantly elevated risk of Parkinson's disease (PD) [3][4][5][6], the most common neurodegenerative movement disorder. PD is characterized pathologically by the degeneration of dopaminergic neurons in the midbrain and the accumulation of ubiquitinated intraneuronal protein aggregates called Lewy bodies. More recently, heterozygous mutations in the GBA1 gene were found to be among the most common genetic associations with sporadic PD and dementia with Lewy bodies (DLB) [7][8][9]. The mechanism by which mutations in GBA1 cause these neurodegenerative disorders is currently unclear.
A number of different models have been proposed to explain the influence of GBA1 mutations on PD, DLB and neuronopathic forms of GD, with α-synuclein protein playing a prominent role in many of these models [3,4,10,11]. α-synuclein is a major component of the Lewy body aggregates that define sporadic PD and DLB, and mutations that lead to increased expression of α-synuclein cause heritable forms of PD with a disease severity commensurate with αsynuclein abundance [12,13]. The finding that glucocerebrosidase normally localizes to the lysosome has led to the model that mutations in GBA1 impair the lysosomal degradation of misfolded forms of α-synuclein, resulting in toxic accumulation of α-synuclein aggregates. While previous work offers support for the model that GBA1 mutations trigger the accumulation of α-synuclein aggregates [14,15], the mechanism by which they do so remains controversial. Moreover, the extent to which α-synuclein contributes to the pathogenesis of GBA1associated neurodegenerative diseases also remains unclear.
To explore the mechanism by which mutations in GBA1 lead to neurodegenerative diseases, we created a Drosophila model of glucocerebrosidase deficiency by generating a deletion of the Drosophila GBA1 homolog dGBA1b. Drosophila dGBA1b mutants exhibit shortened lifespan, locomotor, memory and other behavioral deficits, neurodegeneration, and accumulation of insoluble protein aggregates that are normally degraded through an autophagic process. While glucocerebrosidase deficiency mildly enhanced the toxicity of α-synuclein in dopaminergic neurons, and resulted in increased α-synuclein aggregation, α-synuclein expression did not enhance any of the phenotypes of dGBA1b mutants. Together, our findings indicate that mutations in GBA1 lead to an increased abundance of proteins normally degraded by autophagy, but that the phenotypes accompanying glucocerebrosidase deficiency are largely independent of α-synuclein. Future studies of Drosophila dGBA1b mutants should clarify the nature of this autophagic defect, and will provide further insight into the pathogenesis of GD, PD, and DLB.

Glucocerebrosidase deficiency in Drosophila results in shortened lifespan and behavioral deficits
To pursue a loss-of-function analysis of the GBA1 gene, we conducted a BLAST search for Drosophila homologs using the human glucocerebrosidase protein as a query sequence. This search revealed two Drosophila GBA1 homologs (CG31148 and CG31414), each of which shares 32% amino acid identity (50% similarity) to human glucocerebrosidase ( Fig 1A). In accordance with a recent study [16], we hereafter refer to the CG31148 and CG31414 genes as dGBA1a and dGBA1b, respectively. dGBA1a and dGBA1b reside approximately 2 kb apart on the right arm of chromosome 3 with a single gene (CG31413) situated between them ( Fig 1B). Data from the Drosophila modENCODE [17] and FlyAtlas [18] gene expression studies indicate that dGBA1a is expressed primarily or exclusively in the midgut, whereas dGBA1b is broadly expressed throughout development in a wide range of tissues, including the larval and adult brain. The gene situated between dGBA1a and dGBA1b (CG31413) encodes a putative quiescin sulfhydryl oxidase that is expressed exclusively in the male accessory gland, an organ that secretes proteins into the seminal fluid. We confirmed the expression patterns described by the modENCODE [17] and FlyAtlas [18] projects for dGBA1a, dGBA1b and CG31413 using quantitative RT-PCR (qPCR) (S1 and S2B Figs). Given the expression patterns of dGBA1a and dGBA1b, the dGBA1b gene appeared most relevant to our studies, and our work focused on this gene.
To explore the consequences of glucocerebrosidase deficiency in Drosophila, we created deletions of the Drosophila GBA1 homologs using publicly available transposon insertions in dGBA1a and dGBA1b [18]. Among the >200 candidate deletions generated using this approach was a 4.3 kb deletion (designated GBA1 ΔTT ) that removes the first 433 amino-terminal codons of the dGBA1b gene and the carboxy-terminal 33 codons of the dGBA1a gene ( Fig  1B). This deletion also removes CG31413 that is situated between dGBA1a and dGBA1b ( Fig  1B). In addition to the GBA1 ΔTT deletion, we also generated a precise excision of the transposon used to create the GBA1 ΔTT allele (a chromosome in which the transposon was excised, but did not produce a deletion), which was used as an isogenic wild type (WT) control in all of our studies (designated GBA1 +/+ ). The GBA1 ΔTT allele results in the production of a hybrid transcript fusing the amino terminal dGBA1a coding sequence to the carboxy terminal dGBA1b sequences (Figs 1B, S2B and S2C) but does not create an in-frame fusion of the dGBA1a and dGBA1b coding sequences, and therefore would not be expected to yield a chimeric glucocerebrosidase protein. Given that the GBA1 ΔTT deletion eliminates the presumptive promoter and >3/4 of the amino terminal coding sequence of dGBA1b, resulting in~75% reduced expression of a GBA1a transcript lacking the final 33 codons of dGBA1a, we conclude that it likely represents a null allele of dGBA1b and a hypomorphic allele of dGBA1a. The residual expression of this fusion transcript was nearly undetectable in heads from GBA1 ΔTT homozygotes (S2C Fig). GBA1 ΔTT homozygotes were viable, fertile and had no detectable morphological abnormalities ( Fig 1C). However, before proceeding with a full characterization of GBA1 ΔTT phenotypes, we first sought to confirm that this deletion alters glucocerebrosidase activity and to test whether inactivation of the CG31413 gene is likely to contribute to the GBA1 ΔTT phenotypes. We compared glucocerebrosidase activity in 14-day-old control flies and GBA1 ΔTT homozygotes using a standard fluorescent assay (see Materials and Methods). Since dGBA1a is expressed in the midgut and dGBA1b is expressed ubiquitously, we hypothesized that dGBA1b is responsible for most of the glucocerebrosidase activity in fly heads and should be absent or severely reduced in GBA1 ΔTT mutant heads. Consistent with our hypothesis glucocerebrosidase activity was reduced by~60% in GBA1 ΔTT mutant heads compared to control fly bodies ( Fig  1D) However, we did not observe a significant reduction in glucocerebrosidase activity in GBA1 ΔTT mutant bodies (Fig 1E), suggesting that most of the glucocerebrosidase activity in bodies derives from dGBA1a expression in the midgut, and that the function of dGBA1a is retained in GBA1 ΔTT homozygotes.
Although the CG31413 gene is only expressed in the male accessory gland, implying that mutations in this gene should only influence male fertility, we obtained and tested a stock bearing a transposon insertion within the single exon of CG31413, which is predicted to interrupt the coding sequence of this 561-amino-acid protein at amino acid position 119 (S4A Fig). qPCR revealed a reduction in the expression of CG31413 in flies homozygous for this transposon insertion by~90% compared to controls (S4C Fig). This mutation in trans to GBA1 ΔTT did not detectably influence lifespan or behavior (S4D- S4F Fig). Thus, our data does not support involvement of CG31413 in the phenotypes of GBA1 ΔTT homozygotes.
Although GBA1 ΔTT heterozygotes and homozygotes were viable and fertile, GBA1 ΔTT homozygotes were significantly shorter-lived than WT controls and GBA1 ΔTT heterozygotes, with mean survival of 33 days in GBA1 ΔTT homozygotes compared to mean survival of 59 days in controls and 62 days in GBA1 ΔTT heterozygotes (Fig 2A, S1 Table). GBA1 ΔTT homozygotes also exhibited a variety of age-dependent behavioral abnormalities, including a progressive climbing defect that was present as early as 5 days of age ( Fig 2B; S3E Fig), delayed recovery from heat-induced paralysis (Fig 2C), and delayed recovery from temporary paralysis after mechanical stress (bang sensitivity) ( Fig 2D). GBA1 ΔTT heterozygotes also exhibited a mild bang sensitivity phenotype at advanced ages, but did not exhibit phenotypes in the other behavioral assays performed (Fig 2A-2D).
Ectopic expression of dGBA1b using the Actin-GAL4 driver ameliorated the decreased lifespan, climbing defect, heat and bang sensitivity phenotypes of GBA1 ΔTT homozygotes (Fig 2A-2D, S1 Table). Ectopic expression of human WT GBA1 also rescued the lifespan defect of GBA1 ΔTT homozygotes to a greater extent than did expression of a human GBA1 construct encoding the p.N370S missense allele (S4 Fig), one of the most common pathogenic mutations causing GD (Fig 2A, S1 Table). Together, these findings suggest that the phenotypes of GBA1 ΔTT homozygotes result from loss of dGBA1b function, and that Drosophila dGBA1b is functionally equivalent to human GBA1.
Glucocerebrosidase deficiency results in a memory defect and neurodegeneration, but no loss of dopaminergic neurons Because GBA1 mutations are frequently associated with cognitive impairment [3,4], and the phenotypes of the GBA1 ΔTT homozygotes are similar to those of previously characterized Comparison of protein sequence of human GBA1 and Drosophila dGBA1a and dGBA1b. Gray indicates similar residues, whereas black shading indicates identical residues. (B) Genomic organization of the Drosophila GBA1 homologs, dGBA1a and dGBA1b, and the intervening CG31413 gene. Orange and blue boxes represent coding and non-coding sequences, respectively. Black arrows indicate direction of transcription. Red arrows designate the breakpoints of the GBA1 ΔTT deletion allele. (C) There was no significant difference in the percentage of GBA1 ΔTT homozygotes and WT controls (GBA1 +/+ ) that survived the embryo to 1st instar larval transition, the 3 rd instar larval to pupal stage transition, or the pupal to adult stage transition. (D) Relative glucocerebrosidase (GCase) enzyme activity from isolated heads of 14-day-old controls and GBA1 ΔTT homozygotes. (E) Relative GCase enzyme activity from bodies excluding heads of 14-day-old controls and GBA1 ΔTT homozygotes. Error bars represent standard error of the mean (s.e.m.), ns indicates p>0.05, **p<0.005 by Student t test in all results shown in this figure. Drosophila models of neurological dysfunction [19,20], we tested whether GBA1 ΔTT homozygotes also exhibited a cognitive defect. To perform this analysis, we used a simple courtship assay of learning and memory. After mating, Drosophila females will not mate again for at least 24 hours, and during this time will vigorously reject mating advances from males. Virgin males placed with recently mated females learn that their advances are likely to be rejected, and demonstrate this learned behavior with an increased latency to initiate courtship. The latency to initiate courtship was similar in 14-day-old untrained GBA1 ΔTT homozygotes and controls ( Fig  3A), indicating that GBA1 ΔTT homozygotes do not have a motor impairment that adversely influences courtship. The GBA1 ΔTT homozygotes also displayed normal learning, as the increase in latency to initiate courtship 1 hour following training was the same in both Drosophila WT dGBA1b using the Actin GAL4 driver (Actin-GAL4>UAS-GBA1b;GBA +/+ ), GBA1 ΔTT homozygotes ectopically expressing Drosophila WT dGBA1b using the Actin GAL4 driver (Act-GAL4>UAS-GBA1b;GBA1 ΔTT/ΔTT ), GBA1 ΔTT homozygotes ectopically expressing human WT GBA1 using the Actin GAL4 driver (Act-GAL4>UAS-hGBA1 WT ;GBA1 ΔTT/ΔTT ), and GBA1 ΔTT homozygotes ectopically expressing human GBA1 harboring the p.N370S mutation using the Actin GAL4 driver (Act-GAL4>UAS-hGBA1 N370S ;GBA1 ΔTT/ΔTT ). (B) Climbing index of 5-day-old flies of indicated genotypes as described in A. (C) Recovery time from mechanical stress (bang sensitivity) of flies of given genotypes as described in A at given adult ages. (D) Recovery time from heat stress of flies of given phenotypes as described in A at given adult ages. Error bars represent s.e.m., *p<0.05, **p<0.005 by Student t test in all results shown in this figure. doi:10.1371/journal.pgen.1005944.g002 GBA1 ΔTT homozygotes and controls ( Fig 3B). However, while control flies still showed increased latency to initiate courtship 24 hours after training, courtship latency in GBA1 ΔTT homozygotes and heterozygotes had returned to the level seen in untrained animals ( Fig 3B). These findings indicate that GBA1 ΔTT heterozygotes and homozygotes are capable of remembering for short, but not long, periods of time.
To test whether the phenotypes of GBA1 ΔTT homozygotes are caused by neurodegeneration, we examined brain sections from paraffin-embedded fly heads for gross neuropathology. We found that 30-day-old GBA1 ΔTT homozygotes exhibited substantially increased brain vacuolization relative to age-matched controls ( Fig 3C). Similar vacuoles have been characterized in other Drosophila neurodegenerative models, including Drosophila models of tauopathy [21], Alzheimer's disease [22], and amyotrophic lateral sclerosis [23], indicating that GBA1 ΔTT homozygotes have a neurodegenerative phenotype. As there is dopaminergic neuron loss in PD and DLB, and prior work demonstrated significant dopaminergic neuron loss in the protocerebral posterior lateral 1 (PPL1) cluster of dopaminergic neurons (outlined in Fig 3D) in several Drosophila models of PD, including parkin mutants [24] and α-synuclein transgenic flies [25], we tested whether GBA1 ΔTT homozygotes also exhibit dopaminergic neuron loss. Despite the increased presence of vacuoles in 30-day-old GBA1 ΔTT homozygote brains compared to age-matched controls (Fig 3C), there was no significant loss of dopaminergic neurons in the PPL1 cluster in 30-day-old GBA1 ΔTT homozygotes ( Fig 3E).
Glucocerebrosidase deficiency leads to the accumulation of protein aggregates that are normally degraded through autophagy Previous work has shown that insoluble ubiquitinated protein aggregates accumulate naturally with age in Drosophila, and that genetic perturbations of autophagy factors influence the formation of these aggregates, suggesting that they are normally degraded in the lysosome [26][27][28]. We postulated that these insoluble ubiquitinated protein aggregates would be more abundant in GBA1 ΔTT homozygotes if glucocerebrosidase deficiency impairs lysosomal protein degradation. Consistent with our hypothesis, we found that insoluble ubiquitinated protein aggregates were far more abundant in GBA1 ΔTT homozygotes in protein extracts from whole flies ( Fig 4A) as well as isolated fly heads (Fig 4B and 4C). Soluble ubiquitinated proteins were also more abundant in GBA1 ΔTT homozygotes, suggesting that their ability to degrade ubiquitinated proteins is generally impaired (Fig 4A and 4B). These effects were reversed upon ectopic expression of dGBA1b, confirming that this phenotype is a consequence of dGBA1b deficiency (Fig 4B and 4C). Ubiquitinated aggregates were also visualized in fixed thoracic muscle (Fig 4D  and 4E) and whole brains ( Fig 4F), with increased aggregate formation seen in GBA1 ΔTT homozygotes compared to age-matched controls. Interestingly, GBA1 ΔTT homozygote brains included both punctate (arrowhead) and filamentous (arrow) staining patterns suggestive of aggregates within both cell bodies and cell projections, whereas only the punctate staining pattern was observed in age-matched controls.  To determine whether increased ubiquitinated protein aggregation in GBA1 ΔTT homozygotes is due to impaired lysosomal protein degradation, we examined whether Ref (2)P, the Drosophila homolog of SQSTM1/p62, is more abundant in GBA1 ΔTT homozygotes compared to WT controls. SQSTM1/p62 is important for selective targeting of ubiquitinated proteins for lysosomal degradation and accumulates when autophagic flux is impaired [29,30]. Similar to the autophagy mutant, Atg7, we found a dramatic increase in Ref (2)P in GBA1 ΔTT homozygotes compared to age-matched controls and GBA1 ΔTT homozygotes ectopically expressing dGBA1b, (Fig 4G and 4H). These findings support the model that glucocerebrosidase deficiency impairs the lysosomal degradation of ubiquitinated protein aggregates, and that the phenotypes of GBA1 ΔTT homozygotes are a consequence of defective flux of autophagic vesicles. However, there was no significant difference in Cathepsin D or hexosaminidase enzyme activity, other lysosomal enzymes, between controls and GBA1 ΔTT mutants (Fig 4I and 4J), suggesting that glucocerebrosidase deficiency does not generally impair lysosome enzymatic function.
Glucocerebrosidase deficiency enhances α-synuclein aggregation, but has little effect on α-synuclein toxicity α-synuclein protein aggregation and Lewy body formation are hallmark pathologic characteristics of GBA1-associated neurodegenerative diseases [31]. Moreover, duplications or triplications causing an overabundance of α-synuclein protein cause heritable forms of PD [32] and α-synuclein protein is at least partially degraded in the lysosome [33]. Recent work in mammalian models, cell culture and post-mortem tissue from individuals with PD indicates that a reciprocal relationship exists between glucocerebrosidase deficiency and the abundance of αsynuclein aggregates [14,31,34,35]. Together, these findings raise the possibility that GBA1 mutations act by influencing the abundance of toxic α-synuclein aggregates. However, it is unclear from previous work how much α-synuclein contributes to the pathogenesis of disorders caused by GBA1 mutations [14,31]. Our Drosophila model of glucocerebrosidase deficiency provides a unique opportunity to address this issue, as there is no Drosophila ortholog to human α-synuclein.
To test whether the lack of dopaminergic neuron loss in GBA1 ΔTT homozygotes (Fig 3E) is explained by the absence of a Drosophila ortholog to human α-synuclein, we ectopically expressed human α-synuclein in a GBA1 ΔTT homozygous background. In previous work we showed that abundant expression of human α-synuclein using two copies of the Tyrosine Hydroxylase (TH) GAL4 driver, and two copies of a UAS-α-synuclein transgene resulted in mild degeneration of dopaminergic neurons [25]. Using this identical α-synuclein expression paradigm, we observed a trend towards mild loss of dopaminergic neurons in the PPL1 cluster of WT controls that did not reach statistical significance (Fig 5A). This trend was further enhanced and reached statistical significance in GBA1 ΔTT homozygotes ectopically expressing α-synuclein (Fig 5A), indicating that glucocerebrosidase deficiency mildly enhanced α-synuclein toxicity in dopaminergic neurons. However, expression of α-synuclein in dopaminergic neurons did not significantly enhance the shortened lifespan, climbing deficits, sensitivity to mechanical stress or heat stress of GBA1 ΔTT mutants (Fig 6A-6D).
To test the hypothesis that glucocerebrosidase deficiency impairs lysosomal degradation of α-synuclein protein and leads to the accumulation of toxic α-synuclein aggregates, we compared the abundance of α-synuclein aggregates in GBA1 ΔTT homozygotes and controls. We were unable to detect aggregation of WT α-synuclein by Western blot analysis. However, expression of an α-synuclein variant harboring the pathogenic missense mutation p.A53T (αsynuclein A53T ), which causes autosomal dominant early-onset PD and accelerated insoluble fibril formation [36], resulted in the appearance of high molecular weight (HMW) α-synuclein Western blot analysis using antiserum to α-synuclein (α-syn) of Tritoninsoluble protein fractions from heads of 10-day-old WT controls (GBA1 +/+ ), GBA1 ΔTT homozygotes (GBA1 ΔTT/ΔTT ), WT controls ubiquitously expressing αsynuclein with the p.A53T mutation (Act-G4>UAS-syn A53T ;GBA1 +/+ ), and GBA1 ΔTT homozygotes ubiquitously expressing α-synuclein with the p.A53T mutation (Act-G4>UAS-syn A53T ;GBA1 ΔTT/ΔTT ). βAct was used as a loading control. (C) Densitometric quantification of α-syn signal in the Triton-insoluble fraction from 10-day-old heads of flies of indicated genotypes as described in B. The level of HMW α-syn signal in the Triton-insoluble fraction in GBA1 ΔTT/ΔTT was normalized to its βAct loading control, and this ratio was in turn normalized to the insoluble α-syn/βAct ratio of GBA1 +/+ , N = 3. (D) Western blot analysis using an antiserum to Ubiquitin (anti-Ub) of Triton-soluble (S) and-insoluble (I) protein extracts from heads of controls ectopically expressing WT α-synuclein using the TH-GAL4 driver (TH-G4>UAS-syn WT ;GBA1 +/+ ) and GBA1 ΔTT homozygotes ectopically expressing WT α-synuclein using the TH-GAL4 driver (TH-G4>UAS-syn WT ;GBA1 ΔTT/ΔTT ). βAct was used as a loading control. (E) Densitometric quantification of anti-Ub levels in Triton-insoluble protein fractions from heads of flies of indicated genotypes as described in D. Levels of Ub signal per genotype were normalized to respective βAct loading controls, and the ratio of Ub/βAct in Th-G4>UAS-syn WT ;GBA1 ΔTT/ΔTT was normalized to the ratio of α-syn/βAct in TH-G4>UAS-syn WT ;GBA1 +/+ . Error bars represent s.e.m. unless indicated, *p<0.05, **p<0.005 by Student t test in all results shown in this figure. species (>50kD) in the Triton-insoluble protein fraction (Fig 5B). We further found that the abundance of these HMW α-synuclein species were increased in GBA1 ΔTT homozygotes ubiquitously expressing α-synuclein A53T relative to control flies expressing human α-synuclein A53T (Fig 5B and 5C). These findings indicate that glucocerebrosidase deficiency impairs degradation of HMW α-synuclein aggregates, or promotes the kinetics of formation of these  aggregates. However, ubiquitously expressing α-synuclein A53T in GBA1 ΔTT homozygotes had no significant influence on the phenotypes of the GBA1 ΔTT homozygotes (S5C and S5D Fig) indicating that the enhanced α-synuclein aggregate formation conferred by glucocerebrosidase deficiency was not matched by a detectable increase in α-synuclein toxicity.
Since α-synuclein is a major component of Lewy bodies and recent work suggests that αsynuclein fibrils may contribute to PD pathogenesis in a non-cell-autonomous fashion by seeding protein aggregation in a prion-like mechanism [37][38][39], we hypothesized that ectopically expressed α-synuclein might increase the pool of insoluble ubiquitinated proteins seen in GBA1 ΔTT homozygotes. Surprisingly, there was no significant enhancement of total insoluble ubiquitinated protein levels with ectopic expression of either WT α-synuclein or the A53T variant of α-synuclein in either WT controls or GBA1 ΔTT homozygotes relative to controls and GBA1 ΔTT homozygotes without α-synuclein expression (Fig 5D and 5E; S5A and S5B Fig).
These results indicate that α-synuclein expression does not independently or synergistically influence the increased protein aggregation resulting from dGBA1b deficiency, even when expressing a mutant form of α-synuclein prone to forming fibrils.

Discussion
To gain insight into the molecular mechanisms underlying PD, DLB and neuronopathic forms of GD, we developed a Drosophila model of glucocerebrosidase deficiency. Glucocerebrosidase deficiency in Drosophila results in shortened lifespan, a variety of age-dependent behavioral phenotypes, neurodegeneration and the accumulation of insoluble proteins that are normally degraded through an autophagic mechanism. While these phenotypes are reminiscent of αsynucleinopathies [40,41], glucocerebrosidase deficiency only mildly influenced the neuronal toxicity and aggregation of α-synuclein, and ectopic expression of α-synuclein did not significantly enhance the glucocerebrosidase deficient phenotypes. Together, our findings indicate that the pathological consequences of glucocerebrosidase deficiency in Drosophila are largely independent of α-synuclein, and that glucocerebrosidase deficiency is the major contributor to pathology in diseases associated with GBA1 mutations.
Our finding that α-synuclein is not a central participant in the pathogenesis associated with glucocerebrosidase deficiency is consistent with recent studies in two different fish species [42,43]. The inverse correlation between glucocerebrosidase activity and α-synuclein aggregation in Drosophila is also consistent with previous studies in rodent models, vertebrate cell culture, post-mortem brain tissues from PD patients, and a recent study in Drosophila [14,31,34,35]. Although we were only able to observe an influence of glucocerebrosidase deficiency on the aggregation of the p.A53T variant of α-synuclein, this finding may simply reflect the fact that this variant is more aggregation-prone, thus allowing us increased sensitivity to detect aggregation. However, our work showing the dGBA1b gene is the predominant Drosophila GBA1 homolog expressed in the fly head contrasts with a recent report showing that neuronal inactivation of the Drosophila dGBA1a gene exacerbated the toxicity of α-synuclein in dopaminergic neurons and in the fly eye [16]. Additional studies will be required to fully address the role of dGBA1a, which appears to remain largely functional in our mutant, and to definitively rule out a role for the CG31413 gene situated between dGBA1a and dGBA1b on the phenotypes of GBA1 ΔTT homozygotes.
Although our work indicates that glucocerebrosidase deficiency has little influence on the toxicity of α-synuclein, the association of GBA1 mutations with PD and DLB frequently involves heterozygous carriers of GBA1 missense alleles. This finding has led to the suggestion that the GBA1 mutations act through a dominant toxic gain-of-function mechanism to cause PD and DLB, perhaps by seeding α-synuclein aggregates via a prion-like mechanism [37][38][39]44]. Because our work involved a putative null allele of dGBA1b, we were unable to address this potential model of pathogenesis. Previous work also indicates that ectopic expression of human α-synuclein in Drosophila confers only mild phenotypic consequences [25,45], so it is also possible that the influence of glucocerebrosidase deficiency on α-synuclein toxicity is not readily evident in Drosophila. While we fully acknowledge these potential confounds, several compelling observations suggest that a loss-of-function mechanism best explains the influence of GBA1 mutations on PD and DLB. For example, many different GBA1 mutations are associated with α-synucleinopathies, including putative null alleles [46,47], and the molecular severity of a GBA1 allele correlates with the risk of developing an α-synucleinopathy in heterozygous carriers [48]. Perhaps the strongest evidence for a loss-of-function mechanism is the finding that individuals with biallelic GBA1 mutations have a substantially elevated risk for developing PD relative to heterozygous GBA1 mutation carriers [3,4,49]. Together these findings offer support for the relevance of animal models bearing null alleles of the GBA1 gene [16,42,43], including our fly model of glucocerebrosidase deficiency, on our understanding of the influence of GBA1 mutations in PD and DLB.
Our glucocerebrosidase deficient fly model should be a valuable tool in future work aimed at understanding the mechanisms underlying the neurodegenerative diseases associated with mutations in GBA1. Although glucocerebrosidase deficiency does not result in dopaminergic neuron degeneration in Drosophila, this finding does not necessarily challenge the utility of our fly model to understand the role of glucocerebrosidase deficiency in PD. Previous work has established that mutational inactivation of Drosophila homologs of genes involved in heritable forms of PD often results in phenotypes that appear discordant with those seen in humans [50][51][52][53]. For example, null mutations of the PINK1 or parkin genes in Drosophila result in dramatic muscle degeneration and germ line defects that are not evident in humans bearing null mutations in these genes [52,54]. However, substantial insight into the roles of PINK1 and Parkin in mitochondrial quality control derived directly from studies of PINK1 and Parkin in the Drosophila musculature and germ line [55]. We anticipate that similarly important insight into the mechanisms underlying neuronopathic GD, PD and DLB will come from studies of the phenotypes of our fly model of glucocerebrosidase deficiency.
Our work suggests at least two general mechanisms by which glucocerebrosidase deficiency triggers neuropathology. First, glucocerebrosidase deficiency may impair autophagy, resulting in increased protein aggregation. Our work in GBA1b mutant flies showing accumulation of Ref(2)P, HMW α-synuclein aggregates, and protein aggregates that are normally degraded through an autophagic mechanism supports this model. Glucocerebrosidase is an important lysosomal enzyme in lipid metabolism [56][57][58], and a deficiency in this enzyme could influence lysosome membrane fluidity, vesicular dynamics, and the biogenesis of lysosomes [57,59,60]. These effects could impair the trafficking of misfolded proteins to the lysosome and/or fusion of autophagic vacuoles. As we did not observe a decrease in Cathepsin D activity in GBA1b mutant flies, lysosomal function may not be impaired by glucocerebrosidase deficiency. Alternatively, glucocerebrosidase deficiency may promote the formation of protein aggregates, rather than impair their degradation. Lipid composition has been shown to influence the kinetics of formation of protein aggregates and α-synuclein fibrilization [61,62], suggesting that an alteration in lipid composition resulting from glucocerebrosidase deficiency could accelerate the accumulation of protein aggregates. These aggregates might subsequently seed further aggregation in a prion-like mechanism. In support of this model, lipid composition has been shown to affect the kinetics of amyloid-β aggregation [63], and recent studies suggest that nonautonomous spreading of α-synuclein fibrils may contribute to PD pathogenesis (39)(40)(41). While it remains unclear whether the increased protein aggregates that we observed in GBA1b mutant flies are due to impaired degradation or accelerated formation of misfolded proteins, α-synuclein expression did not enhance the abundance of protein aggregates, arguing against an additive influence of α-synuclein on protein aggregation metabolism. Future experiments will be required to distinguish between these models and reveal the underlying mechanism of GBA1-mediated accumulation of protein aggregates.

Drosophila strains and cultures
Fly stocks were maintained on standard cornmeal-molasses food at 25°C. The human GBA1 wild type and p.N370S transgenes were created by Alex Whitworth (MRC Mitochondrial Biology Unit) and were targeted to the attP landing site of Bloomington stock 24482-M {3xP3-RFP.attP'}ZH-51C (with M{vas-int.Dm}ZH-2A). Full length human α-synuclein was cloned into pUASt. pUAS-synuclein WT was introduced into the Drosophila genome by injection in the Pallanck lab using standard P-element-mediated transformation techniques [64]. The full length dGBA1b transcript was cloned into pUASTattB and targeted to the attP landing site of Bloomington stock 36304-y[1] v [1]; P{y[+t7.7] = CaryP}attP40, and Bloomington stock 8622-y[1] w[67c23]; P{y[+t7.7] = CaryP}attP2. pUASTattB-dGBA1b was introduced into the Drosophila genome by Duke University Model System Genomics. All other stocks were obtained from the Bloomington Stock Center except for the TH-GAL4 driver [65]. The TH-GAL4 transgene was randomly transpositioned to the second chromosome in the Pallanck lab.
Generation of GBA1 ΔTT and isogenic wild type GBA1 + alleles CG31148 Forward ACACCGGAGCAGTGGAATAC Reverse GTCGGAGCTTTTTGAACTCG CG31414 Forward CGCAACAATTCGCTGAACTAC Reverse TGTTTGGGTCAAAAACAGCA Whole genome sequencing using 100-bp paired-end reads on the Illumina HiSeq2000 platform was used to map the exact breakpoints of the GBA1 ΔTT allele. Specifically, reads were aligned and processed against r5.43 of the D. melanogaster genome using BWA and Samtools. Breakpoints were confirmed by visual inspection using the Integrated Genome Viewer.

Detection of possible fusion transcript due to GBA1 ΔTT allele
The following primers were used to amplify and quantitate by qPCR a possible fusion transcript from cDNA isolated from 20 whole GBA1 ΔTT homozygotes: Primer 1: TCTCCAAGTGGGTTCCAGAG Primer 2: TGAAGTTGTGCGTCAGATCC Primer 3: GACCGGACAACAAAATCACC Primer 4: CGGAGCATCCACAAAGTTCT

Gene expression analysis
RNA was extracted from 20 whole bodies, 20 bodies without heads, or 60 heads from 5 day old male flies (unless otherwise indicated) of the indicated genotype, using TRIzol (Life Technologies) and Direct-zol RNA MiniPrep extraction kit (Zymo Research). cDNA was generated using the iSCRIPT cDNA Synthesis kit (Bio-Rad). Quantitative real-time PCR (qPCR)-based quantification was performed for the following genes using the indicated primers: Forward CCGCTGAAGGTAATGCCTTG Reverse CGTTTATCCGATCCTTTGCAGA qPCR was performed using iTaq Universal SYBR Green Supermix (Bio Rad) and a LightCycler 480 (Roche) machine. The delta-delta log 2 method was used to calculate fold change in expression levels. Rap2L was used as the internal control, as the expression of this gene has been reported as the most invariant across different genotypes and ages [66]. Each experiment was performed at least three times.

Hydrolase activity assays
Glucocerebrosidase and hexosaminidase activities were determined using the artificial substrates 4-methylumbelliferyl (4-MU)-β-D-glucoside or 4-MU-N-acetyl-β-D-glucosaminide as previously described (31). Briefly, twenty bodies without heads or 60 heads of 5-7-day-old flies were homogenized in 300 ul of KP buffer (50 mM potassium phosphate buffer, 0.25% Triton X-100, pH 6.5) and centrifuged at 4,000 g for 5 minutes. Supernatants were transferred to a fresh tube and assayed for the hydrolase activity in 0.1M sodium acetate buffer (pH 4.5). Lysates were incubated with Conduritol B epoxide (50 μM, 45 minutes, Toronto Research Chemicals, Inc) to subtract for non-lysosomal glucocerebrosidase mediated activity. Hydrolase reactions were stopped (1M Glycine-NaOH buffer, pH 12.5) and fluorescence quantified using a Spectra Max M2 fluorescent plate reader (excitation 365 nm, emission 445 nm, Molecular Devices). Protein levels were determined using the micro-BCA kit (Pierce, Rockford, IL). Each experiment was performed at least three times.

Lifespan analysis
Longevity assays were conducted at 25°C. Groups of 10-20 age-matched flies were collected at 0-24 hours old and transferred to fresh standard food every 2-3 days. The number of dead flies was recorded during each transfer. Transfers were continued until all flies died. Kaplan-Meier lifespan curves were generated using Stata (StataCorp, College Station, TX), and analyzed by Cox proportional hazard models for statistically significant differences in survival between tested genotypes.

Bang sensitivity assay
Vials containing 5 flies each were vortexed at maximum speed for 10 seconds, then observed until flies recovered. The time for full recovery with purposeful movements was recorded. Assays were videotaped for scoring. Recovery longer than 3 minutes was recorded as 3 minutes.

Heat stress assay
Vials containing 10 flies each were submerged in a 39°C water bath for 6 minutes, then removed to room temperature and the time for full recovery with purposeful movements was recorded. Assays were videotaped for scoring. Recovery longer than 6 minutes was recorded as 6 minutes.

Climbing assay
Climbing behavior was assayed as previously described [50]. Briefly, 20-30 flies were placed into the first chamber of a countercurrent apparatus, tapped to the bottom, and given 30 seconds to climb a distance of 10 cm. Flies that climbed 10 cm in 30 seconds were shifted to another chamber, and the experiment was repeated. After five trials, the flies in each chamber were counted, and the climbing index was calculated as the weighted average of the number of flies in each chamber divided by five times the number of flies in the assay.

Learning and memory assay
Virgin females aged 13 days were placed with virgin males aged 13 days for 24 hours for mating. These females were then used as "trainers" for the subsequent conditioning assay. Naïve (virgin) 14-day-old males were placed with a recently mated "trainer" female for 1 hour in a conditioning/mating chamber. These conditioned males were returned to individual vials for 1, 6, or 24 hours, and then placed in a mating chamber with a freeze-killed virgin female as a courtship target and videotaped. A blinded investigator reviewed videos, and latency time was recorded as the number of seconds until the male extended one wing.

Image analysis
DA neurons were analyzed in situ as previously described [24]. An Olympus Fluoview-1000 confocal microscope was used to acquire optical sections of fixed brains at 1-μm intervals with identical parameters. The number of TH-positive neurons within the PPL1 dopaminergic neuron cluster was counted by visual inspection of individual confocal Z-series images by an investigator blinded to genotype.
For quantification of ubiquitin aggregates within thoracic muscle, each Z-stack of thoracic muscle was acquired with identical parameters (34 x 0.3 μm thick slices) using an Olympus Fluoview-1000 confocal microscope and deconvolved using Huygens Professional 14.10.0-2p4 software (Scientific Volume Imaging, Netherlands), with a signal to noise ratio of 20, maximum iterations of 140. The number of ubiquitinated protein aggregates per stack was calculated using the Advanced Object Analyzer of the Huygens software, using a threshold of 2 times the standard deviation of the image. The seed threshold for objects was set at 10% and the garbage threshold for objects at 35 voxels or smaller. Objects at the margins were also discarded, and watershed segmentation was used to differentiate between objects in close proximity to each other. At least 9 Z-stacks were analyzed per condition.

Cathepsin D assay
The bioactivity of Cathepsin D was measured in 10 ng of protein from homogenized whole flies by a fluorescence-based assay using MCA-GKPILFFRLK(DNP)-dR-NH2 as a synthetic substrate (CTSD Activity Fluorometric Assay Kit; BioVision, Mountain View, CA). Fluorescence was measured at an excitation/emission = 328/460 nm using a Synergy H1 Hybrid Reader (BioTek, Winooski, VT). Each experiment was performed at least three times.