Effects of ambroxol on the autophagy-lysosome pathway and mitochondria in primary cortical neurons

Glucocerebrosidase (GBA1) mutations are the major genetic risk factor for Parkinson’s Disease (PD). The pathogenic mechanism is still unclear, but alterations in lysosomal-autophagy processes are implicated due to reduction of mutated glucocerebrosidase (GCase) in lysosomes. Wild-type GCase activity is also decreased in sporadic PD brains. Small molecule chaperones that increase lysosomal GCase activity have potential to be disease-modifying therapies for GBA1-associated and sporadic PD. Therefore we have used mouse cortical neurons to explore the effects of the chaperone ambroxol. This chaperone increased wild-type GCase mRNA, protein levels and activity, as well as increasing other lysosomal enzymes and LIMP2, the GCase transporter. Transcription factor EB (TFEB), the master regulator of the CLEAR pathway involved in lysosomal biogenesis was also increased upon ambroxol treatment. Moreover, we found macroautophagy flux blocked and exocytosis increased in neurons treated with ambroxol. We suggest that ambroxol is blocking autophagy and driving cargo towards the secretory pathway. Mitochondria content was also found to be increased by ambroxol via peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC1-α). Our data suggest that ambroxol, besides being a GCase chaperone, also acts on other pathways, such as mitochondria, lysosomal biogenesis, and the secretory pathway.


Results
Ambroxol modifies lysosomal content and function. The optimum concentration of ambroxol for the treatment of neurons was established by treating primary wild-type mouse cortical neurons with three different doses of ambroxol: 10, 30, and 60 µM for 5 days. Cellular viability was measured with the Live/Dead assay. The 60 µM dose caused cell death (63% decrease in the number of live cells, p = 0.003, Fig SI 1A), while 10 or 30 μM had no significant effect on cell death. Consequently, we excluded the highest dose from further studies. The morphology of the neurons was further analysed by measuring the neurite length upon ambroxol treatment at doses 10 and 30 µM. There was no evidence of significant alterations in neurite length at these two dosages (Fig SI 1B), thus we continued our study treating neurons with 10 and 30 µM of ambroxol for 5 days.
Firstly, we looked at the effect of ambroxol on lysosomal content and function. As ambroxol is a GCase chaperone 16 , we started by analysing endogenous wild-type GCase activity upon ambroxol treatment. We detected a similar increase in GCase activity both in the 10 µM (39%, p = 0.05) and 30 µM (47%, p = 0.05) doses (Fig. 1A). To assess if a lower dosage could be used, GCase activity was measured upon treatment with 5 µM of ambroxol, however at this dosage no significant increase in GCase activity was observed, thus we continued our studies using 10 µM and 30 µM dosages.
Ambroxol affects the autophagy pathway. To analyse autophagy flux, we detected LC3B-II protein levels in neurons upon ambroxol treatment. We found that ambroxol significantly increased LC3B-II to similar levels at both doses (at 10 µM: 199%, p = 0.02; at 30 µM: 203%, p = 0.003, Fig. 2A). Upon treatment with 100 nM of Bafilomycin A1 (BAF) for 6 h, which increases lysosomal pH and impairs autophagy flux by inhibiting the fusion between the lysosomes and autophagosomes, LC3B-II levels were increased in control cells as expected, and were similar to levels in cells treated with ambroxol alone. However, in ambroxol treated cells, LC3B-II levels were not significantly increased following BAF treatment, compared to just ambroxol treatment alone ( Fig. 2A, upper panel). This was also seen when the LC3B-II data was expressed as a ratio of + BAF/-BAF ( Fig. 2A, middle panel). This suggests that autophagy flux is blocked by ambroxol. To investigate further, we measured P62/ SQSTM1 levels. P62 binds proteins identified for degradation by macroautophagy and helps recruit them to the phagophore, which is then sequestered inside the autophagosome with the cargo and ultimately degraded following fusion with a lysosome. We found that P62 protein levels tended to increase following ambroxol treatment, although this was not significant (Fig. 2B). We also assessed the protein levels of two key mediators of chaperone Scientific RepoRts | (2018) 8:1385 | DOI:10.1038/s41598-018-19479-8 mediated autophagy, hsc70 and Lamp2a. Lamp2a was unchanged upon ambroxol treatment (data not shown) whereas hsc70 was increased when neurons were treated with 30 µM of ambroxol (10 µM: no differences, 30 µM: 195%, p = 0.008; Fig. 2C).
Ambroxol activates transcription factor EB (TFEB). Since we found alterations in lysosomal content and autophagy pathways in neurons treated with ambroxol, and it is well described that TFEB is a master regulator of the lysosomal and autophagy pathway 21,28 , we assessed the activation of TFEB upon ambroxol treatments. TFEB activation is measured by its translocation to the nucleus. We analysed TFEB levels in Triton X-100 cytosolic lysates of neurons treated with ambroxol and detected a decrease in TFEB levels upon treatment (10 µM: 10%, NS; 30 µM: 38%, p < 0.05, Fig. 3A). This might suggest translocation of TFEB to the nucleus. So next we separated cytosolic and nuclear fractions by subcellular fractionation and quantified the amount of TFEB in the nucleus of neurons treated with ambroxol and detected a modest increase in nuclear TFEB both at 10 and 30 µM doses (10 µM: 181%, p = 0.006; 30 µM: 137%, p = 0.01, Fig. 3B). Unfortunately, we were unable to reliably detect endogenous mouse TFEB moving to the nucleus in neurons by immunofluorescence.
We also measured the mRNA levels of Cathepsin D, which is one of the lysosomal genes under the regulation of TFEB, and found it significantly increased at both dosages (10 µM: 114%, p = 0.01; 30 µM: 138%, p = 0.01, Fig. 3C). This finding in conjunction with the increased Gba1 mRNA levels and lysosomal protein expression ( Fig. 1) suggest that ambroxol is activating TFEB.

Effect of ambroxol on exocytosis.
Since we found an increase in the extracellular α-synuclein levels upon ambroxol treatment, and α-synuclein is known to be secreted via exosomes, we studied the possibility of ambroxol being able to induce exocytosis. BAF treatment was used as a positive control for exosomal release 30 . CD63 and flotillin, two proteins present in exosomes, were quantified in the extracellular media of neurons. Both 10 and 30 µM of ambroxol increased the levels of these exosome markers. The lower 10 µM dose of ambroxol led to a 44% increase of CD63 (p = 0.04) and a 24% increase in Flotilin1 (p = 0.03) whereas 30 µM of Ambroxol increased CD63 by 62% (p = 0.008) and Flotillin1 by 53% (p = 0.04) compared to control (Fig. 6A,B) We also analysed the levels of LAMP1, which is found in exosomes. We detected again that both doses of ambroxol increased the extracellular presence of LAMP1 in the extracellular neuronal media. The lower ambroxol dosage (10 µM) increased LAMP1 levels by 38% (p = 0.02), whereas 30 µM of ambroxol increased the release of LAMP1 by 172% (p = 0.03) (Fig. 6C). We also assessed the intracellular levels of Rab11, as it is a protein required for exocytosis and recycling endosomes. We found that the levels of Rab11 were increased with both doses compared to control (10 µM: 163%, p = 0.02; 30 µM: 178%, p = 0.02, Fig. 6D) which corroborates the hypothesis that ambroxol is inducing exocytosis.

Effect of ambroxol on Gba1-deficient neurons.
To understand whether some of the effects of ambroxol are dependent on GCase expression and activity we assessed the effect on Gba1 −/− (GBA KO) and Gba1 −/+ (GBA HET) primary cultures. GBA HET neurons showed a significant decrease in GCase activity (58% decrease, p = 0.003), whereas GBA KO neurons had negligible GCase activity (KO: 1.19 nmol/h/mg vs WT: 159 nmol/h/ mg). Treatment with 30 µM of ambroxol increased WT neurons GCase activity by 37% (p = 0.004) and GBA HET neurons by 50% (p = 0.001) whereas no changes were observed in GBA KO (Fig. 7A). No differences were observed in Hex A activity in WT, HET or KO GBA neurons (Fig. 7B). Extracellular release of α-synuclein was also increased in GBA HET and GBA KO upon treatment with 30 µM ambroxol (Fig. 7D). Together our data indicate that some of the effects of ambroxol in neurons occur independently of GCase activity.

Discussion
Small molecule chaperones that increase GCase activity are being studied for their potential to be disease-modifying therapies for PD with GBA1 mutations [18][19][20]22,23,[31][32][33] . Since GCase activity is also decreased in sporadic PD 8,24 and ambroxol has been shown to increase WT GCase and other lysosomal proteins in control fibroblasts [18][19][20] we have investigated the effect of ambroxol on mouse neurons containing WT GCase.
In a recent study from our laboratory WT mice were treated with ambroxol at a concentration of 4 mM in drinking water resulting in significantly increased brain GCase activity of 16-22% 23 . However, the amount of ambroxol that actually reaches the brain, and neurons in particular, is unknown. Human studies have reported ambroxol concentrations in blood or CSF of up to 3 μM, although this will be dependent on the dosing regimen used 34,35 . We report that 10 μM is sufficient to increase GCase activity in primary neurons. Previous reports in fibroblasts, SH-SY5Y and neural crest stem cell-derived dopaminergic neurons have used 60 µM of ambroxol 18,19,31 but we found this dose to have cytotoxic effects on primary mouse cortical neurons.
Similar to what was reported in fibroblasts 18,19 we have found that ambroxol increased Gba1 mRNA, protein levels and activity, as well as, several lysosomal proteins such as cathepsin D, LAMP1 and the GCase transporter LIMP2. Nuclear translocation of TFEB, a master regulator of lysosomal biogenesis 21 was increased upon treatment with ambroxol which probably contributes to the increased lysosomal content and number of acidic vesicles in these cells.
Activation of TFEB is commonly associated with activation of macroautophagy, particularly following starvation and inhibition of mTOR 28,36 , but is also increasingly associated with other physiological responses ranging from lysosomal exocytosis, lipid catabolism, ER stress and the immune response 26,37-40 . Macroautophagy flux appeared to be inhibited in ambroxol treated mouse cortical neurons as LC3-II, a marker of autophagosome number, was increased under basal levels but was not greatly increased in the presence of bafilomycin A1. This was not expected as ambroxol has been shown to increase macroautophagy in human neuronal cells containing GBA1 mutations 31 . The lower dose or different type of neuron might account for the discrepancy. It should also be noted that mouse cortical neurons were treated at day 4 or 5 in vitro, whereas the human neurons were differentiated for several weeks. Autophagy inhibition appears to play a role in axon development and branching in developing mouse cortical neurons 41,42 , whereas lysosomal content increases significantly as cultured human neurons mature 9,31 . Either of these observations might explain the differing susceptibility of these neurons to ambroxol. The mechanism by which TFEB is activated following ambroxol treatment in mouse cortical neurons requires further investigation. For example, is it a compensatory mechanism due to the inhibition of macroautophagy following ambroxol treatment? Or perhaps a direct consequence of ambroxol treatment affecting the phosphorylation status of TFEB by activating/inactivating a kinase or phosphatase, and thus it's propensity to translocate to the nucleus.
Given the observed increase in exocytosis following ambroxol treatment, perhaps primary mouse neurons at these lower doses respond by increasing the secretory pathway, rather than autophagy. Ambroxol is a mucolytic used primarily to treat respiratory diseases, where it acts as a secretory agent. Fois et al. have shown that ambroxol in concentrations greater than 1μM accumulates in lamellar bodies, which are the acidic Ca 2+ stores in pneumocytes, much like lysosomes, leading to the release of Ca 2+ and increased exocytosis 34 . Notably release of lysosomal Ca 2+ activates calcineurin, which can then dephosphorylate and thus activate TFEB 43,44 . TFEB activation has been implicated in regulating lysosomal exocytosis 37,45 , by increasing the pool of lysosomes in proximity to the plasma membrane and promoting the fusion of lysosomes with the plasma membrane. In lysosomal storage diseases, lysosomal exocytosis has been described as a beneficial event that relieves the cells from storage material and degradation products 46,47 . Lysosomal exocytosis has also been linked to plasma membrane repair, neurite outgrowth, improved phagocytosis and the release of signalling molecules and digestive enzymes 46,48,49 . Lysosomal exocytosis has been reported to generate specific plasma membrane domains containing LC3-II 49 . If this occurs in neurons, it might contribute to the higher LC3-II levels observed following ambroxol treatment that were not increased further by bafilomycin A1 treatment.
The decrease in α-synuclein phosphorylated at Ser129 in neurons treated with ambroxol is similar to that observed in mice over expressing human a-synuclein treated with ambroxol 23 . It is unclear if this decrease is due to increased exocytosis of α-synuclein or via another mechanism.
The observation that Gba1 KO neurons exhibit a similar response to ambroxol with regards to α-synuclein release as WT neurons indicates that this effect is independent of chaperoning GCase.  Fig SI 3. Data presented in % of CTR. All data represent mean ± S.E.M. *p < 0.05, **p < 0.01, ***p < 0.001. We proceeded to investigate the effect of ambroxol on PGC1-α levels as this protein is known to be regulated by TFEB 26,27 . PGC1-α has been described to be involved in mitochondria biogenesis 29 . We found that Pgc1-α mRNA levels were increased upon ambroxol treatment in neurons. Coincident with this, mitochondrial proteins were also found increased in neurons treated with ambroxol, pointing to an increase in mitochondria content in treated neurons. Mitochondrial defects have been found in cell and animal models of GCase deficiency [50][51][52][53][54] and in idiopathic PD [55][56][57] . mTOR, a master regulator upstream of TFEB and PGC1-α, has also been found to be decreased in GCase deficiency models 11,58 . It is interesting to speculate that ambroxol's effects on TFEB and PGC1-α would be beneficial to restore both mitochondrial function and rebalance lysosomal metabolism in PD.
In summary, our data suggest that ambroxol, has several actions beyond its chaperone activity; it influences lysosomal and mitochondrial function, and increases protein exosomal release. Its ability to reduce cellular phosphorylated α-synuclein in in vitro and in vivo models highlights its potential application to PD to modify the progression of this disease. Clinical trials are currently underway to test this hypothesis.

Methods
Ethical Approval. The Gba1 knockout mouse model and wild type (WT) colony was covered by project licence 70/7685 issued by the United Kingdom Home Office. All animal procedures were carried out as described in the above project licence and the United Kingdom Animals (Scientific Procedures) Act of 1986 (Schedule 1). All efforts were made to reduce the number of animals by following the 3 R's.
Primary Cortical Mouse Neurons. WT mice and Heterozygotic Gba1 transgenic mice in which a loxp-neo-loxp (lnl) cassette was inserted into exon 8 of one Gba1 allele in all cell types 59 were used to generate mouse embryonic primary cortical neurons as follows. Cortex from day 15 embryos were dissected, meninges removed, homogenized by passing through a 23 G needle and centrifuged at 1000 rpm for 5 min. Cell pellets were resuspended in neuronal media which consisted of neurobasal media (Invitrogen) supplemented with B27 Nuclear extracts. To prepare nuclear extracts of neurons treated with ambroxol we used the EpiQuik Nuclear Extraction kit (Epigentek), as per manufacturer's instructions. Then, western blotting was performed on the cytoplasmic and nuclear fractions. Dot blotting. We used the Bio-Dot SF blotting apparatus from Bio-Rad to analyse media collected from neuronal cultures treated with ambroxol for 5 days. Before loading the media in the apparatus, debris and floating cells were removed by centrifugation. Bio-Dot SF blotting apparatus was assembled following the instruction manual and 100μl of media was loaded in the respective well. Dot blots were then washed with PBS and then blocked with Block Ace (BioRad) ON at 4 °C, followed by incubation with antibodies against CD63 (Santa Cruz Biotechnology, 1:100), Flotillin1 (Abcam, 1:1000), Lamp1 (Abcam 1:1000). β-hexosaminidase A was assayed in lysates (2 µg protein) using the fluorogenic substrate 4-methylumbelliferyl-N-acetyl-glucosaminide (2 mM), in sodium citrate buffer (pH 4.2) at 37 °C for 30 minutes 60,61 . The reaction was stopped by addition of 0.25 M glycine (pH 10.4) and fluorescence measured as above.

Lysosomal Enzyme
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 12 well plates (90% confluent) and incubated with 4 µl/ml Lyso-ID in assay buffer at 37 °C for 30 minutes. Cells were washed with assay buffer and fluorescence measured on a plate reader at excitation 488 nm, emission 520 nm. Following reading, buffer was aspirated, cells lysed overnight in 0.25 M NaOH and protein concentration measured. Cell fluorescence was expressed as fluorescent units/mg protein. Quantitative Real-Time PCR. RNA was extracted from neuronal cells using RNeasy kit (Qiagen) and converted to cDNA using nanoScript2 RT kit (PrimerDesign). qPCR was performed in a STEP One PCR machine (Applied Biosystems) using the Power SYBR Green PCR master mix (Applied Biosystems). Primer used are listed in Table 1. β-Actin mRNA levels were used to normalize the data. Relative expression was calculated using the ΔCT method.
Measurement of α-synuclein by ELISA. Conditioned media was removed from neurons at day 5 of ambroxol treatment and debris/floating cells removed by centrifugation. Neurons were harvested and cell pellets frozen. The amount of α-synuclein released in to media was measured by ELISA (Sensolyte; AnaSpec) as per manufacturer's instructions. Cell pellets were lysed and protein content calculated using BCA protein assay kit (ThermoScientific). Data were expressed as pg α-synuclein/mg protein.