Ambroxol reverses tau and α-synuclein accumulation in a cholinergic N370S GBA1 mutation model

Abstract Cognitive impairment is a common non-motor complication of Parkinson’s disease (PD). Glucocerebrosidase gene (GBA1) variants are found in 10–15% of PD cases and are numerically the most important risk factor for PD and dementia with Lewy bodies. Accumulation of α-synuclein and tau pathology is thought to underlie cognitive impairment in PD and likely involves cholinergic as well as dopaminergic neurons. Neural crest stem cells were isolated from both PD patients with the common heterozygous N370S GBA1 mutation and normal subjects without GBA1 mutations. The stem cells were used to generate a cholinergic neuronal cell model. The effects of the GBA1 variant on glucocerebrosidase (GCase) protein and activity, and cathepsin D, tau and α-synuclein protein levels in cholinergic neurons were examined. Ambroxol, a GCase chaperone, was used to investigate whether GCase enhancement was able to reverse the effects of the GBA1 variant on cholinergic neurons. Significant reductions in GCase protein and activity, as well as in cathepsin D levels, were found in GBA1 mutant (N370S/WT) cholinergic neurons. Both tau and α-synuclein levels were significantly increased in GBA1 mutant (N370S/WT) cholinergic neurons. Ambroxol significantly enhanced GCase activity and decreased both tau and α-synuclein levels in cholinergic neurons. GBA1 mutations interfere with the metabolism of α-synuclein and tau proteins and induce higher levels of α-synuclein and tau proteins in cholinergic neurons. The GCase pathway provides a potential therapeutic target for neurodegenerative disorders related to pathological α-synuclein or tau accumulation.

are numerically the most important risk factor for the development of PD, and for Dementia with Lewy Bodies (DLB) (4,5). Cognitive impairment is reported to occur earlier and progress more rapidly in PD subjects with GBA1 variants, including the N370S variant (6).
Most GBA1 variants reduce GCase activity and this in turn is associated with elevated levels of α-syn (7). We have previously reported on the biochemical consequences of the N370S GBA1 variant in dopaminergic neurons derived from patients with PD (8). For the first time, we report the effects of the common N370S PD-associated GBA1 variant in stem cell derived cholinergic neurons to provide further insight into the potential mechanisms of cognitive dysfunction in GBA1-linked PD.
Cholinergic neuronal cell models have been generated from Alzheimer's disease patients (9)(10)(11). Although different procedures and growth factors were used in the protocols, there was a common step in all these procedures, which is turning stem cells into neurospheres, then inducing neurospheres into cholinergic neurons. Neural crest stem cells (NCSCs) derived from adipose tissue have all the properties to form neurospheres (8), suggesting NCSCs may act as an alternative cell resource for generation of cholinergic neuronal models.
Tau is a member of microtubule-associated protein family and is involved in several neurodegenerative diseases. Tau pathology in neurodegenerative diseases is characterised by pathological tau aggregation in neurofibrillary tangles. The aggregation and deposition of tau were observed in approximately 50% of PD brains. Ambroxol (ABX) has been used for several years for the treatment of airway mucus hypersecretion and hyaline membrane disease in new-born babies. A drug screen then identified ABX as small molecule chaperone of GCase (12). Treatment of fibroblasts containing GBA1 mutations with ABX resulted in increased level of GCase protein and its activity (13) and ABX has also been shown to increase GCase activity to reduce GCase substrate in macrophages with GBA1 mutations (14). We employed ambroxol as a GCase enhancement agent to treat GBA1 mutant (N370S) cholinergic neurons to examine whether increased GCase protein and activity affect the metabolism of α-syn and tau proteins in cholinergic neurons.

Formation and characterisation of neurospheres
Adipose-derived NCSCs were converted to neurospheres as previously described (8) ( Figure 1A-C). A bromodoxyuridine (BrdU) incorporation assay showed that 6 days after conversion cells were still proliferating, as also reflected by the increase in size of the neurospheres ( Figure 1D-F).

Induction of neurospheres to medial ganglionic eminence cells
The first step for differentiation of neurospheres to cholinergic neurons is to induce transition to medial ganglionic eminence (MGE) cells (19). A fibronectin-coated plate was used to make suspended neurospheres ( Figure 2G) attach to the plate surface ( Figure 2H).
Cells were further incubated with neurobasal medium supplemented with B27, FGF2 and leukaemia inhibitory factor for up to7 days. MGE cells were formed and migrated from the neurospheres ( Figure 2I). The homeobox protein NKX2-1 is an MGE cell marker (19). A higher level of NKX2-1 expression was seen at 7 days of incubation ( Figure 2K) compared to

Characterisation of cholinergic neurons
After 31 days of cholinergic neuronal differentiation most cells showed neuronal cell morphology ( Figure 3A). The vesicular acetylcholine transporter (VAChT) is a neurotransmitter transporter responsible for transferring acetylcholine into secretory vesicles (20). VAChT is regarded as a specific marker for cholinergic neurons and has been widely used for the study of cholinergic transmission in experimental models of Alzheimer disease and other disorders involving cholinergic neurons (21). Immunostaining showed that the neuronal marker β-III tubulin was expressed in the majority of the NCSC-derived cholinergic neurons after 31 days of differentiation ( Figure 3B, C and D, and Suppl. Figure 1), whereas VAChT was expressed in 45-60% of the cells ( Figure 3B and Suppl. Figure 1).
Immunoblotting demonstrated that VAChT levels progressively increased during the 31-day differentiation protocol ( Figure 3E  receptors have been used to identify cholinergic neurons in previous studies (22,23).
Immunostaining revealed that 35-45% of the differentiated cells expressed ChAT and GABA-B receptors, while >85% expressed the neuronal marker β-III tubulin ( Figure 3C and D, and Suppl. Figure 1).
GCase protein and activity decreased in parallel suggesting that the low GCase activity is due to a lower level of enzyme protein. Compared with controls, tau and phospho-tau (S396) levels were significantly increased, respectively 164% and 120% higher in N370S/WT cholinergic neurons ( Figure 4B and C), while α-syn levels were also significantly increased by 105% in N370S/WT cholinergic neurons ( Figure 4D).

Effects of the GBA1 N370S mutation on Cathepsin D and macroautophagy pathways in cholinergic neurons
Our previous study reported that the N370S/WT GBA1 variant reduced Cathepsin D (CTSD) protein and activity in dopaminergic neuronal cells (29). We therefore compared CTSD protein levels between control and N370S/WT cholinergic neurons. CTSD levels were significantly lower in N370S/WT cholinergic neurons compared with control ( Figure 5A).
Tau and α-syn turn over through the autophagy pathway. Therefore, we examined  Figure   5D). Likewise, levels of the autophagy receptor protein p62, encoded by the SQSTM1 gene, were not significantly different in the control and N370S/WT cholinergic cultures ( Figure   5E).

Ambroxol (ABX) treatment increases GCase protein level and activity
ABX is a GCase pharmacological chaperone and has been reported to increase the expression of the TFEB transcription factor (30) and increase GCase protein and activity in human dopaminergic neurons (8). N370S/WT GBA1 variant cholinergic neurons were treated with ABX for 6 days. The treatment resulted in significant increases of GCase protein level and activity by 50% and 55%, respectively ( Figure 6A, B and C). ABX treatment significantly reduced tau levels to 44% of basal levels ( Figure 6D and E). Phospho-tau (S396) levels varied in the different vehicle-treated N370S/WT GBA1 cholinergic cultures ( Figure   6D) but treatment with ABX resulted in an 8-to 20-fold decrease of phospho-tau (S396) levels ( Figure 6D and F). Levels of α-syn were significantly decreased to 59% of basal levels following ABX treatment ( Figure 6G and H). and correlated with both reduced GCase activity and GBA1 gene expression (38). We have previously published the effects of the N370S GBA1 variant on dopaminergic neurons and found GCase protein and activity significantly decreased (8), with a significant reduction of CTSD protein and activity, and an increase in -syn levels (29). Reduced expression of CTSD has also been reported in GBA1-PD (39). The mechanism for why mature CTSD expression is reduced in GBA1-PD neurons is unclear. Cathepsins are expressed as proproteins, maturing as they reach the acidified environment of endolysosomes. Loss of GCase activity likely affects the lipid profile of cells that might influence the transport of proteins to lysosomes and/or the pH of lysosomes. The imbalance of both sphingolipids and phospholipids has also been implicated in the mishandling of -syn (40). While we have not measured lipids in this study, ABX treatment has been reported to lower GCase substrate in macrophages treated with ABX (14), and might contribute to the reduction in at least -syn in our models following chaperone treatment. In addition to autophagy, soluble -synuclein and tau can be degraded by the ubiquitin proteasome system (UPS) (41). The UPS has been reported to be decreased in the brain of Gba1 knockout mice (42) and might also contribute to the accumulation of a-syn and/or tau in this cholinergic model.

Discussion
It has previously been suggested that both α-syn and tau proteins play an important role in the pathogenesis of cognitive dysfunction in PD (38). Our observation that N370S GBA1 cholinergic cells have increased levels of these two proteins may help provide an explanation for the earlier onset and more rapid progression of dementia associated with this mutation (43). This is supported by the observation of both α-syn and tau aggregates in the hippocampus of a homozygous Gba1 mutant mouse model, which were coincident with memory deficits (44). Notably, the increased tau we observe in our cholinergic cells, was phosphorylated at the same residues as tau aggregates associated with the pathology of Alzheimer disease and tauopathies.
ABX has been shown to reverse the biochemical consequences of GBA1 mutations, increasing GCase enzyme activity, and reducing α-syn levels in a range of cell types, including human dopaminergic neurons with the N370S GBA1 variant, and in vivo models (8,30,45,46). ABX has most recently been used in the first clinical trial of a personalised medicine for PD (47). It has been shown to be brain penetrant and to increase GCase protein levels in cerebrospinal fluid in PD patients with and without GBA1 mutations. ABX has been proposed as a potential disease-modifying drug to slow onset and progression in PD. The results of the present study are important in terms of the potential for ABX to reduce tau, phospho-tau and -syn levels in cholinergic neurons and potentially influence the onset and progression of cognitive decline in PD. Its ability to reduce tau, phospho-tau and α-syn in cholinergic cells, as well as α-syn in dopaminergic cells, could represent an important protective mechanism against dementia in PD, and GBA1-PD in particular. DNA from all participants was studied by whole-exome analysis.

Sample collection and cell isolation
Subcutaneous fat was collected by skin biopsy. The previous published procedures were followed for the sample preparation and NCSC isolation (8). The six individual subjects were divided into two groups according to their genotype (WT/WT healthy, N370S/WT PD).

Neurosphere formation and cholinergic neuronal differentiation
NCSCs were cultured as described previously (8), harvested in stem cell growth medium and centrifuged at 200× g to collect cell pellets. Cell pellets were re-suspended in neurosphere formation medium and plated in non-coated (low adhesion) culture dishes. For cholinergic neuronal differentiation, neurospheres were transferred to a fibronectin-coated 6well plate with neurosphere formation medium. Following 24-hour culturing, during which neurospheres attached to the surface of plate, medium was replaced with fresh neurosphere formation medium. The next procedure was divided into 3 stages. Stage 1: neurospheres were cultured in neurosphere formation medium for a further 6 days, medium was changed on day 3; 1 ml of medium was removed, 2 ml of freshly made medium was added to each well during medium change. Stage 2: neurosphere formation medium was removed from each well, 2 ml of fresh pre-cholinergic neuronal differentiation medium was added to each well.
Cells were cultured in pre-cholinergic neuronal differentiation medium for 12 days; medium was changed every 4 days; 1 ml of medium was removed; 2 ml of fresh medium was added to each well during medium change. Stage 3: pre-cholinergic neuronal differentiation medium was removed from each well, 2 ml of fresh cholinergic neuronal differentiation medium was added to each well. Cells were cultured in cholinergic neuronal differentiation medium for 12 days; medium was changed every 4 days; 1 ml of medium was removed; 2 ml of fresh medium was added to each well during medium change.

ABX treatment
Medium was removed; fresh cholinergic neuronal differentiation medium supplemented with 10 µM ABX was added to each well (2 ml/well, 6-well plate); medium was changed every 48 hours. Treatment lasted for 6 days. Control cells were treated with vehicle (dimethylsulfoxide, DMSO) instead of ABX.

GCase enzyme activity assays
Cell pellets were lysed with 1% Triton X-100 in PBS. GCase activity was determined in cell lysates of ~1 μg protein as previously reported (7). Enzyme activities were calculated by subtracting the background fluorescence from the mean fluorescence measured for a given cell lysate and then divided by the standard to calculate the activity in nmol/h/ml. This result was then divided by the total protein concentration, as determined using bicinchoninic acid assay method, to calculate the enzymatic activity in nmol/h/mg.

Immunochemistry
Cells were washed twice with phosphate-buffered saline (PBS), each wash lasting 5 minutes. Cells were fixed with 4% paraformaldehyde in PBS for 15 minutes at room temperature and subsequently permeabilised with 0.25% Triton X-100 in PBS for 15 minutes.
Following three PBS washes, cells were blocked with 10% goat serum in PBS for 30 minutes and incubated with primary antibodies (Suppl. Immunoblotting Cells were harvested, washed with PBS, and processed as previously described (8).
Primary antibodies are given in Suppl. Table 1.