Autophagic lysosome reformation dysfunction in glucocerebrosidase deficient cells: relevance to Parkinson disease

Glucocerebrosidase (GBA1) gene mutations increase the risk of Parkinson disease (PD). While the cellular mechanisms associating GBA1 mutations and PD are unknown, loss of the glucocerebrosidase enzyme (GCase) activity, inhibition of autophagy and increased α-synuclein levels have been implicated. Here we show that autophagy lysosomal reformation (ALR) is compromised in cells lacking functional GCase. ALR is a cellular process controlled by mTOR which regenerates functional lysosomes from autolysosomes formed during macroautophagy. A decrease in phopho-S6K levels, a marker of mTOR activity, was observed in models of GCase deficiency, including primary mouse neurons and the PD patient derived fibroblasts with GBA1 mutations, suggesting that ALR is compromised. Importantly Rab7, a GTPase crucial for endosome-lysosome trafficking and ALR, accumulated in GCase deficient cells, supporting the notion that lysosomal recycling is impaired. Recombinant GCase treatment reversed ALR inhibition and lysosomal dysfunction. Moreover, ALR dysfunction was accompanied by impairment of macroautophagy and chaperone-mediated autophagy, increased levels of total and phosphorylated (S129) monomeric α-synuclein, evidence of amyloid oligomers and increased α-synuclein release. Concurrently, we found increased cholesterol and altered glucosylceramide homeostasis which could compromise ALR. We propose that GCase deficiency in PD inhibits lysosomal recycling. Consequently neurons are unable to maintain the pool of mature and functional lysosomes required for the autophagic clearance of α-synuclein, leading to the accumulation and spread of pathogenic α-synuclein species in the brain. Since GCase deficiency and lysosomal dysfunction occur with ageing and sporadic PD pathology, the decrease in lysosomal reformation may be a common feature in PD.


Introduction
Gaucher disease (GD), the most common lysosomal disease, is caused by homozygous mutations in the glucocerebrosidase 1 (GBA1) gene which encodes the lysosomal enzyme glucocerebrosidase (GCase). GCase is responsible for the conversion of glucosylceramide to glucose and ceramide in the lysosome. Deficiency in GCase activity leads to accumulation of its substrate in the lysosome and compromised lysosomal activity (1,2).
The pathology of Parkinson disease (PD) is characterized by loss of substantia nigra pars compacta dopaminergic neurons with the formation of a-synuclein-rich Lewy bodies in surviving neurons (3). A subset of GD patients and GBA1 mutation carriers are at high risk of developing PD (4)(5)(6)(7). Mutations in the GBA1 gene are numerically the most important genetic risk factor for developing PD (7,8). The patients with PD with GBA1 mutations have similar clinical features and pathology to sporadic PD (8,9), except there is an earlier age of onset (about 5 years) and increased cognitive decline (8)(9)(10). The biochemical link between GBA1 mutations and PD was established with the identification of GCase deficiency in PD brains positive for these mutations (11). This deficiency was most pronounced in the substantia nigra and other midbrain areas, sites of the greatest pathology in PD. Of particular importance was the demonstration that there was evidence of significant GCase deficiency in sporadic PD brains (11)(12)(13) suggesting loss of enzyme activity is central to PD pathogenesis. Loss of GCase activity has been shown to promote a-synuclein accumulation (14)(15)(16)(17) and mitochondrial dysfunction (18,19). However, the mechanism by which GCase deficiency leads to a-synuclein accumulation is still unclear.
GCase deficiency and the subsequent accumulation of its substrate affect lipid metabolism and trafficking, decreases lysosomal degradation capacity and impairs autophagy (5). The autophagy -lysosomal pathway (ALP) is important in the clearance of a-synuclein and the impairment of both macroautophagy and chaperone mediated autophagy (CMA) have been linked to PD pathogenesis (20)(21)(22)(23).
Since lysosomes are the primary degradative compartment of the cell, their biogenesis and recycling are vital for cellular function. Lysosome biogenesis requires the endosomal system to direct newly synthesized lysosomal proteins to the lysosome. Lysosomal targeted substrates pass through different endosomal intermediaries to reach their final destination. Endosomes mature from early endosomes (Rab5 positive) to late endosomes (Rab7 positive) which then fuse with the lysosome, delivering the cargo. This process is important for delivery of essential proteins for lysosomal function (24). Recently Yu et al. (25) described lysosomal reformation following autophagy termination; this pathway is used to recycle autolysosomes in order to maintain the cellular pool of dense and functional lysosomes. After the degradation of the autolysosomal products by autophagy, mTOR is reactivated leading to the attenuation of autophagy and to the formation of proto-lysosomal tubules and vesicles that are ultimately excluded from the autolysosomes and mature into functional lysosomes, in a process dependent of Rab7 (25). This process, termed autophagy lysosome reformation (ALR), was found to be inhibited in fibroblasts derived from patients with the lysosomal storage disorders Scheie syndrome and Fabry disease (25).
The relationship between loss of GCase activity in the lysosome and PD is not yet fully understood. Here we describe for the first time the importance of GCase in lysosomal reformation. We show that GCase deficient cells exhibit altered lysosomal recycling, and propose that the accumulation of defective lysosomes contributes to autophagy impairment and accumulation of a-synuclein.
The lysosomal dysfunction detected could be related to the accumulation of dysfunctional lysosomes due to impaired lysosomal recycling via ALR. The recycling process is regulated by activation of mTOR, and is measured via phosphorylation of its substrate p70S6Kinase (phopho-S6K) (25).
Analysis of phopho-S6K levels in a human patient-derived fibroblasts with homozygous and heterozygotic GBA1 mutations, confirmed that GBA1 deficient human fibroblasts had significantly reduced phopho-S6K levels under basal conditions (55% decrease in GBA deficient cells, P < 0.001) ( Figure 2B) similar to Gba1 deficient MEFs.
Alterations in lysosomal pH can lead to the changes in lysosomal function and in the lysosomal reformation process. The amount of acidic/functional lysosomes in the Gba1 deficient cells was analysed using the Lyso ID probe. A significant decrease in acidic vesicles in both Gba1 KO (67% 6 20.99, P < 0.01) and Gba1 HET (40.9% 6 13.57, P < 0.001) was detected, when compared to Gba1 WT MEF. This was not accompanied by alteration in lysosomal morphology or localization (data not shown). Treatment with Cerezyme resulted in a tendency to increase the acidic vesicles in the case of Gba1 WT and Gba1 KO (110% 6 16.96 and 82% 6 29.39) and Gba1 HET (77% 6 20.93, P < 0.01) suggesting that GCase deficiency may be responsible for the lysosomal dysfunction ( Figure 2D and E).
ALR impairment could be a consequence of the decreased lysosomal acidification, thus, we treated Gba1 WT, KO and HET with 100nM of bafilomycin A1 (Baf A1) for 2h, which inhibits vacuolar H þ ATPase (V-ATPase) and increases lysosomal pH. Upon treatment with Baf A1, Gba1 WT cells had a trend to decrease phopho-S6K levels whereas, in both Gba1 HET and Gba1 KO cells Baf A1 treatment significantly decreased phospho-S6K, when compared to untreated controls (Fig. SI 2A). This suggests that lysosomal reformulation is affected by the alkalization of lysosomes.
To analyse whether decreasing cholesterol levels in Gba1 HET and KO restored ALR, we treated cells with Lovastatin (10lM for 48h), an HMG-CoA reductase enzyme inhibitor known to inhibit cholesterol production. Lovastatin decreased cholesterol levels by 43% in Gba1 WT cells, 27% in Gba1 KO and 26% in Gba1 HET. This was the maximal dose and time before cell morphology and proliferation became affected. This treatment did not appear to restore ALR as phopho-S6K levels were not significantly changed in HET or KO following cholesterol decrease (Supplementary Material, Fig. SI 2B).

Impairment of protein degradation via chaperonemediated autophagy in GCase deficient MEFs
Impaired autophagy has also been linked to GCase deficiency, therefore we measured macroautophagy flux. LC3B-II protein level was analyzed in Gba1 WT, KO and HET MEFs. There were We also assessed CMA activity. By measuring the turnover of the well-characterized CMA substrate, GAPDH, following protein synthesis inhibition with cycloheximide (CHX) for 24h, we detected that GAPDH protein levels were significantly decreased by 49.6% in Gba1 WT MEF following CHX treatment, whereas GAPDH protein levels in Gba1 HET and Gba1 KO MEF were similar to non-treated cells ( Figure 3D, P < 0.001). This suggests that in Gba1 KO and HET, degradation via CMA is compromised. The protein levels of two key mediators of CMA, hsc70 and Lamp2a, were unchanged when compared to Gba1 WT (Hsc70: Gba1 HET, 108% 6 28.50; Gba1 KO, 103% 6 27.70; Lamp2a: Gba1 HET, 73% 6 15.00; Gba1 KO, 78% 6 27.50).

Knockdown of GCase in SH-SY5Y cells affects lysosomal content and lysosomal recycling
To assess ALR and the turnover of a-synuclein we knockeddown (KD) GCase by siRNA in parental SH-SY5Y cells and SH-SY5Y over expressing human a-synuclein. GCase protein expression was decreased by 56 6 6% following 10 days of KD with GBA1 siRNA, when compared to cells treated with scrambled siRNA ( Figure 4A). This resulted in a significant 70% decrease in GCase activity ( Figure 4A). The activities of the lysosomal Gba1 WT (n ¼ 4). Cerezyme treatment significantly increased acidic functional lysosomes in Gba1 HET MEFs but not Gba1 KO. Results were expressed as % of respective control. # P < 0.01 for HET þ cerezyme compared to HET. Scale bar 10mm. (E) Lyso ID fluorescence measurements using a plate reader detected a significant decrease in functional/acidic lysosomes in both Gba1 HET and Gba1 KO compared to Gba1 WT (n ¼ 4). All data represent mean 6 SD, *P < 0.05; **P < 0.01, ***P < 0.001 vs. respective control. enzymes b-hexosaminidase and b-galactosidase and the protein expression levels of mature cathepsin D protein and LAMP1 were all significantly increased in GBA1 KD cells ( Figure 4B-D).
Under basal conditions (CTR), phospho-S6K was significantly decreased by 25.6% in GBA1 KD cells ( Figure 5A; P < 0.05, n ¼ 5). Starvation-induced autophagy for 2 h (ST) inhibited mTOR, resulting in loss of S6K phosphorylation. Culturing of these cells with media containing serum for 1 h (REC) recovered S6K phosphorylation, but was less in GCase KD cells ( Figure 5A). To confirm that this was due to GCase deficiency, S6K phosphorylation was assessed in SH-SY5Y treated with the GCase inhibitor conduritol-bepoxide (CBE; 10 lM). This concentration of CBE inhibited GCase activity by 92.0 6 0.0% after 10 days of treatment (replenished every 3 days). Phospho-S6K was significantly decreased by 37% under basal (CTR) conditions in CBE-treated cells, and did not recover as much as untreated cells (SH-SY5Y) once starvationinduced autophagy had been removed for 1 h (REC; Figure 5B).
The acidic/functional lysosomes were quantified in this model using the Lyso-ID probe. GCase KD SH-SY5Y cells presented a decreased number of functional/acidic lysosomes compared to control ( Figure 5C; 51% 6 23.58, P < 0.05). Changes in lysosomal morphology and cellular localization were not detected (data not shown).
Analyses of GCase KD cells indicated that the 70% decrease in GCase activity did not result in an increase in the substrate glucosylceramide or other sphingolipids such as ceramide and sphingomyelin (data not shown). Total cholesterol levels were increased by 135 6 13% (P < 0.05, n ¼ 7) in GCase KD cells ( Figure 5D).  Gba1 HET and Gba1 KO presented more punctate Rab7 staining and increased co-localization with cathepsin D. Graph represents co-localization quantification using Pearson's Coefficient (n ¼ 3). (C) Total cholesterol levels were increased in Gba1 KO and Gba1 HET compared to Gba1 WT (n ¼ 3). (D) GAPDH protein levels decreased in Gba1 WT after 24h of protein synthesis inhibition with cycloheximide (CHX) (n ¼ 3). No difference was observed in Gba1 KO and Gba1 HET MEF (n ¼ 3). All data represent mean 6 SD, *P < 0.05; **P < 0.01, ***P < 0.001 vs respective control. impaired in GBA1-siRNA treated cells. Protein levels of the autophagic protein p62 were also significantly increased in GBA1 siRNA-treated cells ( Figure 6B; P < 0.05), further confirming an inhibition of macroautophagy flux in these cells. Protein levels of CMA proteins Lamp2a and Hsc70 were unchanged (data not shown).

Knockdown of GCase in SH-SY5Y cells inhibits macroautophagy flux and increases a-synuclein protein levels
GCase KD in SH-SY5Y cells over expressing exogenous WT asynuclein tagged with hemagglutinin (HA) resulted in a significant increase in a-synuclein protein levels ( Figure 6C; scram siRNA, 0.65 6 0.18; GBA1 siRNA, 1.01 6 0.06; P ¼ 0.05, n ¼ 5). HA-tagged asynuclein was also found to be increased in these cells by immunofluorescence following treatment with GBA1 siRNA ( Figure 6C).

Impaired lysosomal recycling and a-synuclein accumulation in GCase deficient neurons
We treated WT neurons with CBE and analysed phopho-S6K levels to detect defects in the ALR in GCase deficient cortical neurons. After 10 days of treatment, CBE inhibited GCase activity by 97.0 6 0.0%. Phospho-S6K was significantly decreased by 35% under basal (CTR) conditions in CBE-treated neurons. Once starvation-induced autophagy had been removed for 1 h (REC), phospho-S6K levels increased but levels were still lower than CTR ( Figure 7A). Rab7 protein levels were also significantly increased in CBE-treated neurons ( Figure 7B). To verify if ALR was also compromised in Gba1 KO mice neurons, we compared phospho-S6K levels in Gba1 KO and Gba1 HET to WT Gba1 cortical neurons ( Figure 7C). We found that the basal levels of phospho-S6K were lower in Gba1 KO and Gba1 HET neurons compared to Gba1 WT neurons (Gba1 KO 51.0% 6 19.55, P < 0.01; Gba1 HET 39% 6 11.25, P < 0.001, n ¼ 4).
Acidic/functional lysosomes were quantified in GCase deficient cortical neurons using the Lyso-ID probe. There was a trend for a decrease in the number of functional lysosomes in GCase deficient neurons compared to control ( Figure 7D; 80.0% 6 24.5, n ¼ 4). Changes in lysosomal morphology and localization were not detected (data not shown).
Similar to MEFs and SH-SY5Y, loss of GCase activity resulted in an increase in total cholesterol levels by 138 6 23% (P < 0.05, n ¼ 3) in CBE-treated neurons compared to untreated neurons ( Figure 7E).
Under basal conditions, LC3B-II levels were significantly increased by 159% in CBE-treated neurons ( Figure 8A). Treatment with Baf A1 increased LC3B-II to a similar amount in CTR and CBE-treated cells suggesting the increase in LC3B-II under basal conditions was due to inhibition of macroautophagy flux ( Figure 8A). In support of this, the autophagic cargo protein p62 was also significantly increased by 134% ( Figure 8A). Levels of the CMA proteins hsc70 and Lamp2a were unaffected (Supplementary Material, Fig. SI 4A, B).
Inhibition of macroautophagy significantly increased intracellular total monomeric a-synuclein levels by 147% as measured by western blotting ( Figure 8B). Phosphorylation of asynuclein at Ser129 has been reported in a-synuclein aggregates and Lewy bodies. Western blotting indicated that phospho-Ser129 a-synuclein was significantly increased by 165% in CBE-treated neurons ( Figure 8B). Furthermore, probing of intracellular lysates with an antibody that recognizes all types of amyloid oligomers (e.g. a-synuclein, huntingtin or Ab42) indicated that 4/7 independent CBE-treated neuron cultures exhibited higher molecular weight protein species (>50 kDa) that were not detected in control neurons (Supplementary Material, Fig. SI 4C). These bands were increased by co-treating neurons Protein levels of mature cathepsin D and LAMP1 were also increased in GBA1 siRNA KD cells (n ¼ 5). All data represent mean 6 S.E.M. * P < 0.05, ** P < 0.01. with 25 nM bafilomycin A1 for 48 h suggesting these amyloid oligomeric species were due to inhibition of lysosomal activity.
Measurement of monomeric a-synuclein in cell lysates by ELISA confirmed that intracellular levels were significantly increased in GCase-deficient neurons ( Figure 8C). The release of asynuclein into cell media from these same neurons was also significantly increased by 159%, when compared to untreated neurons ( Figure 8C). The amount of a-synuclein detected in media (CTR, 0.70 6 0.10 ng/mg protein; CBE, 1.116 0.29 ng/mg protein) is a small fraction compared to intracellular a-synuclein levels (CTR, 1109 6 146 ng/mg protein; CBE, 1237 6 163 ng/mg protein).

Discussion
We report that the inhibition of autophagy following GCase deficiency is due to loss of lysosomal biogenesis via ALR and the maturation of endosomes. These events occurred in several cellular models of heterozygous and homozygous GCase deficiency suggesting that this is a mechanism by which pathogenic a-synuclein species could accumulate in PD (Figure 9).
ALR is a vital process for the cell to restore its pool of functional lysosomes. This process is initiated upon attenuation of autophagy due to reactivation of mTOR, which was measured by the increased phosphorylation of S6K (25). In all cases of GCase deficiency, phospho-S6K was decreased suggesting mTOR was less active in these cells. Upon starvation followed by recovery, phospho-S6K levels in GCase deficient cells remained lower than in controls. Therefore over time, repeated cycles of autophagy followed by ALR are going to result in less functional lysosomes in GCase deficient cells as evidenced by the decreased number of acidified lysosomes and expression and activity of other lysosomal enzymes. Neurons differentiated from iPSC with homozygote or heterozygote GBA1 mutations have shown lysosomal dysfunction and impaired autophagy, with lysosomal cell content either increasing (29) or decreasing (15,30). Notably, we report that ALR impairment and loss of acidic lysosomes could be reversed by treating cells with recombinant GCase (Cerezyme).
The GTPase Rab7 has an important role in endosomelysosome maturation (27,28,31) and is also required to dissociate from autolysosomes for ALR to proceed (25). In GCase deficient cells, we detected both an accumulation of Rab7 and increased co-localization with cathepsin D suggesting Rab7 does not dissociate from the autolysosomes as quickly as in control cells. Inhibition of ALR has been shown to increase Rab7 protein levels and have increased association with autolysosomes (32). Furthermore, these same changes in Rab7 and the accumulation of cholesterol in late endosomes have also been associated with impaired maturation of lysosomes (28,33). Dysregulation of Rab7 has been linked with mutations in the Leucine-rich repeat kinase 2 (LRRK2) gene, which cause autosomal dominant PD (34)(35)(36), suggesting that impaired maturation of lysosomes and ALR could be a common mechanism in PD pathogenesis.
Specific lipids have been implicated in the maturation of endosomes and ALR (32,33). Increased cholesterol also impairs the maturation of late endosomes to lysosomes (33,37). Therefore, changes in glucosylceramide or other lipid and sterol species as a result of GCase deficiency may contribute to the decrease in lysosomal maturation and ALR. ALR inhibition has also been reported to occur in fibroblasts with saposin c deficiency (38). Saposin c is a cofactor for GCase and saposin c deficiency in these cells resulted in accumulation of glucosylceramide (38). Inhibition of GCase with CBE, or administration of exogenous glucosylceramide, increases cholesterol, inhibits endosomal degradation and alkalizes endosomes/lysosomes (37,39). Corroborating our data, two studies showed cholesterol accumulation in patient derived skin fibroblasts bearing GCase mutations (40,41).
As cholesterol accumulated in our models we studied whether decreasing cholesterol levels, using Lovastatin, a drug widely used to treat hypercholesterolaemia, would restore ALR (42). We did not detect a restoration of ALR in Gba1 KO and HET cells when cholesterol levels were lowered, although the maximum non-toxic dose of Lovastatin was unable to restore levels back to normal. Statins such as Lovastatin have wide range of effects in the cells, not only decreasing cholesterol levels but affecting other pathways including activation of AKT in neurons. (43). AKT is known to be upstream of mTOR. While total cellular cholesterol levels are reduced by Lovastatin, we cannot exclude the possibility that cholesterol is still increased in endosomes and lysosomes. Therefore, we do not exclude the possible importance of cholesterol/lipid dysregulation in ALR impairment following GCase deficiency.
Substantial evidence suggests a role for autophagy impairment and lysosomal depletion in PD and ageing (20). This not only results in increased intracellular a-synuclein levels, but also increased release and transmission of a-synuclein to other cells (44,45). Our neuronal GCase deficiency models exhibited inhibition of macroautophagy, increased intracellular total and phosphorylated (S129) monomeric a-synuclein levels and release of a-synuclein, co-incident with impaired ALR. In line with our findings, a recent report showed increased release of asynuclein in iPSC-derived dopamine neurons from the patients with PD with heterozygous GBA1 mutations (29). The amount of a-synuclein in condition media was a small fraction compared to intracellular levels (< 1%). We measured release in to media for 72 h, so it is likely that some a-synuclein had been degraded or taken back up by neurons during this time.
A proportion of CBE-treated neurons (4 out of 7 cultures) were also positive for amyloid-like deposits. The antibody cannot distinguish whether these oligomers contain a-synuclein or other amyloidgenic proteins such as b-amyloid or huntingtin. Since phospho Ser129 a-synuclein, which is typically found in asynuclein aggregates, is also increased in these cells, and a-synuclein aggregates have previously been reported in other GCase deficient models (14,17,19), it is likely that these amyloid deposits contain a-synuclein. However, further work is required to identify the composition of these putative oligomers.
We did not find impairment in macroautophagy flux in our Gba1 mutant MEFs. In previous studies no evidence of macroautophagy inhibition was found in GD fibroblasts (38,46). However, macroautophagy has been shown to be impaired in mixed cultures of neurons and astrocytes from the same mouse model (19).
Loss of ALR appeared to be similar in our heterozygote or homozygote GBA1 models. Autophagy inhibition and a-synuclein accumulation have previously been reported to be similar in iPSC-derived neurons from GD patients and the patients with PD with heterozygous GBA1 mutations (15). Genetic studies also support these observations, as both GD and heterozygotic GBA1 gene carriers present a similar risk of developing PD (47).
We propose that loss of functional lysosomal GCase leads to impairment of lysosomal recycling and endosome maturation, possibly due to alterations in lipid and/or cholesterol homeostasis. Over time, repeated cycles of autophagy followed by impaired ALR cause a decrease in the pool of functional lysosomes and thus autophagy deficiency, diminishing the cellular quality control mechanisms which results in a-synuclein accumulation and increased a-synuclein release, contributing to PD pathology. Since recombinant GCase can reverse ALR impairment, we anticipate that strategies to restore GCase activity in the brains of both sporadic patients with PD and those with GBA1 mutations will improve autophagy lysosomal pathway, preventing the accumulation of a-synuclein and spread of pathology.

Ethical approval
Dermal fibroblasts derived from controls (n ¼ 2) or patients with GD or PD were generated following local ethical approval from the Hampstead Research Ethics Committee: reference number 10/H0720/21. All individuals provided written informed consent.
The Gba1 knockout mouse model colony was covered by project licence 70/7685 issued by the United Kingdom Home   Office. All animal procedures were carried out in accordance with the United Kingdom Animals (Scientific Procedures) Act of 1986 (Schedule 1). All efforts were made to reduce the number of animals by following the 3R's.
WT mice and heterozygotic Gba1 transgenic mice (26) were used to generate primary cortical neurons as follows. Cortexes from day 15 embryos were dissected, meninges removed, homogenized by passing through a 23G needle and centrifuged at 1000rpm for 5 min. Cell pellets were resuspended in neurobasal media (Invitrogen) supplemented with B27 (Invitrogen), glutamax (Sigma) and antimycotic/antibiotic solution (Sigma) and seeded onto poly-ornithine coated plates (Sigma). Neurons were left in culture for no longer than 15 days.
Conduritol-b-epoxide (CBE; Universal Biologicals, Cambridge, UK) was used to inhibit GCase activity in SH-SY5Y and cortical neurons. Cells were continuously treated with 10lM of CBE every 3 days for 10 days.
Starvation and recovery assay SH-SY5Y, MEFs and neurons were plated in 6 well plates. Cells were treated with either complete media or media lacking FBS (starvation media). After 2h of starvation, cells were either harvested, or for recovery studies, media were replaced with complete media and cells harvested after 1h incubation.

Western blotting
Cells were harvested with trypsin, washed in phosphate buffered saline (PBS), and lysed in 1% (v/v) Triton X-100 in PBS supplemented with protease and phosphatase inhibitors. Nuclei and cell debris were removed by centrifugation at 17, 000g. For intracellular detection of total monomeric a-synuclein levels in neurons, cell pellets were lysed in 0.1% SDS, 150 mM NaCl, 10mM Tris, pH7.5 supplemented with RQ1 DNase (Promega) and protease and phosphatase inhibitors and incubated at 37 C for 1 h. For detection of oligomeric species, cells were lysed with RIPA buffer on ice (150 mM NaCl, 1% (v/v) NP-40, 1% (w/v) sodium deoxycholate, 50 mM Tris, pH 8.0) supplemented with protease inhibitors. Debris was removed by centrifugation as above. Protein (5-30 lg) was separated on 4-12% or 12% Bis-Tris NuPAGE gels (Invitrogen) and transferred to a Hybond P membrane ( GAPDH turnover GAPDH turnover was assessed using cycloheximide (25mg/ml) assay. Samples were harvested 24h after treatment. Samples were loaded according to cell number calculated using a fluorescent DAPI assay of DNA content using a plate reader (excitation, 360nm; emission 460 nm). Western Blot was performed as described previously using anti-GAPDH primary antibody (ab8245, 1:1000 Abcam).

Lysosomal enzyme assays
Cell lysates were prepared as above, sonicated and GCase activity determined in samples (20 lg protein) by hydrolysis of 5 mM 4-methylumbelliferyl-b-D-glucopyransoside in McIIvaine buffer (pH 5.4) in the presence of 22 mM sodium taurocholate at 37 C for 1 h. The reaction was stopped by the addition of 0.25M glycine (pH 10.4) and 4-methylumbelliferone fluorescence measured at excitation 360 nm, emission 460 nm.

Lysosomal function: Lyso-ID green detection kit
Acidic vesicles were detected in live cells using the Lyso-ID Green detection kit (Enzo Lifesciences, Farmingdale, NY, USA). Cells were seeded in 96-well plates or 12 well plates (90% confluent) and incubated with 4 ll/ml Lyso-ID in assay buffer supplemented with 2% (v/v) FCS at 37 C for 30 min. Cells were washed with assay buffer and fluorescence measured on a plate reader at excitation 488 nm, emission 520 nm. Following reading, the buffer was aspirated, cells lysed overnight in 0.25M NaOH and protein concentration measured. Cell fluorescence was expressed as fluorescent units/mg protein.
For live cell imaging, cells were seeded on glass bottom dishes (ibidi, Germany) and incubated with 1ll/ml Lyso-ID in HBSS for 15 min. Cells were then washed with HBSS and imaged in HBSS.

Image acquisition and processing
Imaging was performed in a Nikon Eclipse Ti-E inverted microscope system using a 60x/1.4 oil objective and a Hamamatsu ORCA-Flash 4.0 camara. Images were acquired using NIS Elements AR software. Lyso ID signal intensity was measured using Image J software. Rab7, Cathepsin D co-localization was measured using Image J plugin JACoP (52).
Measurement of a-synuclein by ELISA Neurons were cultured in poly-ornithine-coated 12-well plates for 14 days. Neurons were treated with 10 lM CBE on days 4, 9 and 11. Conditioned media were removed from the neurons on day 14 and debris/floating cells removed by centrifugation. Neurons were harvested with trypsinisation and cell pellets and their respective conditioned media frozen. Total cell pellets were lysed in RIPA buffer and protein content calculated using a BCA protein assay kit (ThermoScientific). The amount of intracellular a-synuclein and a-synuclein released in to the media was measured by ELISA (Sensolyte; AnaSpec) as per manufacturer's instructions. The data were expressed as pg a-synuclein/ mg protein from at least three separate wells per prep. The percent increase of a-synuclein in CBE-treated neurons was then calculated from four independent neuronal cultures.

Lipidomic analysis
Cellular lipids were extracted in methanol/chloroform (2:1, v/v). The lipids were analyzed by liquid chromatography-mass spectrometry (LC-MS) using a Thermo Exactive Orbitrap mass spectrometer (Thermo Scientific, Hemel Hempstead, UK), equipped with a heated electrospray ionization probe and coupled to a Thermo Accela 1250 UHPLC system. The samples were analyzed in positive and negative ion modes over m/z 200-2000. Injections (1 ml) were made onto a Thermo Hypersil Gold C18 column (1.9 lm; 2.1 mm Â 100 mm). Mobile phase A consisted of water containing 10 mM ammonium formate and 0.1% (v/v) formic acid. Mobile phase B consisted of 90:10 isopropanol/acetonitrile containing 10 mM ammonium formate and 0.1% (v/v) formic acid. The initial conditions for the analysis were 65%A/35%B. The percentage of mobile phase B was increased to 100% over 10 min and held for 7 min before re-equilibration with the starting conditions for 4 min. The LC-MS data were processed with Progenesis QI software v2.0 (Non-Linear Dynamics, Newcastle upon Tyne, UK) and searched against LIPID MAPS (www.lipidmaps.org/) and the Human Metabolome Database (http://www.hmdb.ca/) for identification. The mean abundance of identified molecular species of glucosylceramide within the study groups was calculated from the normalised positive ion data sets.

Cholesterol measurements
Total cholesterol levels were analysed using the Amplex Red Cholesterol Assay kit from Invitrogen. Manufacturer instructions were followed and cholesterol levels were expressed in mg of total cholesterol/mg of total protein.

Lovastatin treatments
Cells were treated with 10mM of Lovastatin (Sigma) for 48H and cholesterol was measured as described before.

Statistical analysis
Statistical analyses were performed by Student's t test. Values with P < 0.05 were considered significant.

Supplementary Material
Supplementary Material is available at HMG online.