Beneficial Effects of Acetyl-DL-Leucine (ADLL) in a Mouse Model of Sandhoff Disease.

Sandhoff disease is a rare neurodegenerative lysosomal storage disease associated with the storage of GM2 ganglioside in late endosomes/lysosomes. Here, we explored the efficacy of acetyl-DL-leucine (ADLL), which has been shown to improve ataxia in observational studies in patients with Niemann-Pick Type C1 and other cerebellar ataxias. We treated a mouse model of Sandhoff disease (Hexb-/-) (0.1 g/kg/day) from 3 weeks of age with this orally available drug. ADLL produced a modest but significant increase in life span, accompanied by improved motor function and reduced glycosphingolipid (GSL) storage in the forebrain and cerebellum, in particular GA2. ADLL was also found to normalize altered glucose and glutamate metabolism, as well as increasing autophagy and the reactive oxygen species (ROS) scavenger, superoxide dismutase (SOD1). Our findings provide new insights into metabolic abnormalities in Sandhoff disease, which could be targeted with new therapeutic approaches, including ADLL.


Introduction
The GM2 gangliosidoses [1] are lysosomal storage diseases that include Tay-Sachs disease, Sandhoff disease and GM2 activator deficiency. Sandhoff disease is caused by mutations in HEXB, a gene that encodes the beta subunit of β-hexosaminidase, leading to storage of GM2 ganglioside and other glycoconjugate substrates in the central nervous system (CNS), resulting in progressive neurodegeneration, CNS inflammation and premature death [2]. Symptom onset can occur in infancy or during childhood/adolescence or adulthood [3]. Early-onset disease results from more disabling mutations that lead to attenuated residual enzyme activity and rapid disease progression, while later-onset forms with some residual enzyme activity progress more slowly [3]. Currently, there are no approved disease-specific treatments for Sandhoff disease. However, several experimental approaches have been evaluated over the years, including bone marrow transplantation, substrate reduction therapy (SRT) and gene therapy [4][5][6][7]. For example, the SRT drug, miglustat, reduced levels of GM2 and GA2 in a mouse model of Sandhoff disease (Hexb -/-), resulting in functional benefit [4,8].
In observational clinical studies, miglustat also provided some benefit to juvenile onset Sandhoff patients [9,10].
Acetyl-DL-leucine (ADLL) is a derivative of the branched chain amino acid leucine, which has been used since 1957 as a treatment for acute vertigo and vertiginous symptoms (Tanganil ® ). It is

Mice
Sandhoff disease mice [20] (Hexb -/-) were bred as heterozygotes (Hexb +/-) to generate affected mice (Hexb -/-) and wild-type controls (Hexb +/+ ) and were housed under non-sterile conditions, with food and water available ad libitum. All experiments were conducted using protocols approved by the UK Home Office Animal Scientific Procedures Act, 1986. All animal works were conducted under the UK Home Office licensing authority, and the project license number is P8088558D. The ethical review committee with oversight of the research that was conducted using animals (mice) is the Committee on Animal Care and Ethical Review (ACER). No human subjects were involved in this study.

Behavioural Analysis
Weekly behavioural tests (9-14 weeks) were conducted in a blinded manner for the first cohort of mice. Gait analysis was performed using the CatWalk 10.5 system (Noldus Information Technology) according to the manufacturer's instructions and five compliant runs were recorded per animal at each time point. The camera was set to 40 cm below the walkway, the walkway was approximately 4 cm wide, and detection settings were set to 14.05 camera gain and 0.12 green intensity. Motor function of mice was measured using a Ugo Basile RotaRod NG (Italy), starting from 1 to 10 rpm, and accelerating every 30 s by one rpm.
Bar crossing experiments were conducted to measure motor strength and coordination [21]. Briefly, a metal bar (1.2 mm in width and 26 cm in length) was suspended horizontally between two wooden supports, 30 cm in height, over a cushioned surface, and animals were allowed to grasp the centre of the bar with forepaws only, the tail was released, and the clock was started. Latency to cross or fall from the bar was scored, with a 180 s maximum to termination of each trial. A score was given for each animal between +180 and −180. If crossing time (CT) was greater than 0, then score = 180 − CT; if falling time (FT) was greater than 0, then score = FT − 180.

Tissue Handling
A second cohort of mice was sacrificed at 86 days of age (late symptomatic stage) and perfused with phosphate-buffered saline (PBS, pH 7.4) (Gibco #10010023). Tissues for biochemical analysis were snap frozen in ice-cold isopentane (−80 • C). Biochemical analyses were performed on water-homogenized tissues (mg/mL) and protein content was determined using a BCA protein assay (Thermo Fisher #23227) according to the manufacturer's instructions.

Image Acquisition and Quantification
Western blot image acquisition was conducted with the Odyssey Infrared Imaging system (Model No. 9120, LI-COR, Cambridge, UK). Fluorescence and luminescence quantifications were performed with Fiji Version 1.51g software (Image J) (http://fiji.sc/Fiji) (W. Rasband, NIH, USA).

Statistical ANALYSES
Differences between groups were identified either by one-way or two-way analyses of variances (ANOVA); comparisons of groups were made after two-way ANOVAs with Tukey's multiple comparison tests. Statistical tests were performed using Prism 6 software (Graphpad v6, La Jolla, San Diego, CA, USA).

The Effects of ADLL in Hexb -/-Mice on Motor Function and Lipid Storage
The Hexb -/mouse model of Sandhoff disease has a life span of approximately 16 weeks of age and is pre-symptomatic until approximately 6-8 weeks of age, and then progressively develops tremor and motor function deficits [8]. In the later stages of the disease (12-15 weeks), these mice become increasingly inactive and are unable to complete motor function tests, such as bar crossing [8]. They progressively accumulate GSLs, in particular GM2 and GA2, in the CNS and visceral organs, including the liver.
Hexb -/animals were either untreated or treated from weaning (3 weeks of age) with ADLL (0.1 mg/kg/day, equivalent to the dose used in observational clinical studies [13]). We observed a statistically significant extension in survival with ADLL (median increase of 8.5% (9.5 days), p = 0.0247) (Figure 1a). Bar-crossing performance (time to cross or fall) also improved significantly in the late-symptomatic stage of the disease (12 weeks: p = 0.0343; 13 weeks: p = 0.0058), which was indicative of slower disease progression (Figure 1b). Hexb -/-animals were either untreated or treated from weaning (3 weeks of age) with ADLL (0.1 mg/kg/day, equivalent to the dose used in observational clinical studies [13]). We observed a statistically significant extension in survival with ADLL (median increase of 8.5% (9.5 days), p = 0.0247) (Figure 1a). Bar-crossing performance (time to cross or fall) also improved significantly in the late-symptomatic stage of the disease (12 weeks: p = 0.0343; 13 weeks: p = 0.0058), which was indicative of slower disease progression (Figure 1b).
A second cohort of treated mice were sacrificed at 86 days of age (late-symptomatic stage) to measure changes in GSL levels in the forebrain, cerebellum and liver [8]. Although total GSL levels did not change with ADLL treatment in these organs (Supplementary Figure S1); some specific gangliosides were significantly reduced. For example, levels of GA2, the most accumulated ganglioside in the forebrain, were 18.5% lower in ADLL-treated Hexb -/mice (p < 0.0001) relative to untreated Hexb -/mice ( Figure 1c). Similarly, GA2 levels were 10% lower in the cerebellum (p = 0.0002) ( Figure 1d) and 15.8% lower in the liver in ADLL-treated Hexb -/mice (p = 0.0121) (Figure 1e). Levels of other GSL species did not change significantly upon ADLL treatment in Hexb -/mice (Figure 1c-e).

ADLL Increases Autophagy in Hexb -/-Mice Cerebellum
We conducted a series of experiments to explore the mechanism of action of ADLL in the CNS. Given that lipid reductions in the brain and the cerebellum showed similar trends (significant GA2 reduction) in response to ADLL, we focused on the cerebellum, as it is important for motor function, which is impaired in Sandhoff disease mice. To better understand how ADLL might mediate these disease-modifying effects in Hexb -/mice, we first explored nutrient sensing and autophagy. The mechanistic target of rapamycin (mTOR) is a master regulator of anabolic metabolism and it suppresses catabolic pathways such as autophagy [23]. Leucine increases phosphorylation of mTOR and subsequent activation, as recently reported in NPC fibroblasts [24]. Phosphorylated mTOR (p-mTOR) levels were significantly decreased in Hexb -/cerebellum (48.1%, p = 0.0456) compared to Hexb +/+ cerebellum (Figure 2a,b). However, the ratio between phosphorylated and total non-phosphorylated enzyme (p:T) did not differ between Hexb +/+ and Hexb -/- (Figure 2a,b). ADLL treatment reduced the protein levels of both p-mTOR and mTOR levels in the cerebellum. However, their levels, as well as the p-mTOR: total mTOR ratio did not reach statistical significance when compared to Hexb -/untreated mice (Figure 2a,b). To determine the status of macroautophagy in the mouse brains, we measured changes in the autophagy marker, LC3, both in the cytosolic form (LC3-I) and autophagic vacuole (AV) bound form (LC3-II), along with the LC3 adaptor protein, p62. Both LC3-I and LC3-II were increased in Hexb -/--untreated mice, compared to wild-type mice (LC3-I: 36.6%, p = 0.0011; LC3-II: 50.3%, p < 0.0001) (Figure 2c,e). However, the internal ratio of LC3-II:LC3-I (a more reliable indicator of autophagic events [25]) did not vary between Hexb +/+ and Hexb -/- (Figure 2c,e). ADLL therapy resulted in a 19.5% reduction of LC3-I levels, but was not statistically significant (p = 0.0804), and no change in LC3-II levels compared to untreated Hexb -/mice was observed (Figure 2c,e). The ratio of LC3-II:LC3-I increased significantly by 27.1% (p = 0.0483), suggesting an increased presence of AVs in ADLL-treated Hexb -/cerebellum (Figure 2c,e). Expression of the LC3 adaptor protein, p62, did not differ between Hexb +/+ and Hexb -/--untreated cerebellums, and remained unaltered in ADLL-treated Hexb -/mice (Figure 2d,e), suggesting that the ADLL-induced increase in AVs was not due to impaired clearance, but more likely an elevation of autophagic activity in Hexb -/cerebellum.

ADLL's Effect on Branched Chain Amino Acid and Glutamate Metabolism
In addition to the known effects of the branched chain amino acid leucine on mTOR [26] and autophagy [27,28], leucine also plays a role in glutamine and glutamate utilisation when glucose metabolism is deficient/impaired [29,30]. Leucine catabolism provides a source of acetyl CoA [31]. Therefore, we investigated levels of branched chain keto acid dehydrogenase E1 alpha polypeptide (BCKDHA), which regulates the final step of leucine catabolism to acetyl CoA [32]. No difference in BCKDHA protein levels were found between Hexb +/+ and Hexb -/in the absence or presence of ADLL treatment (Figure 3a,d). There were also no statistical differences between healthy and diseased mice with regard to the inactive phosphorylated form of BCKDHA, or its p:T ratio (Figure 3a,d). Although ADLL decreased p-BCKADH levels significantly (24.3%, p = 0.0494), the absence of a change in its p:T ratio indicated that ADLL did not directly impact on BCKDHA activity (Figure 3a,d). Therefore, although there was a trend towards BCKADH activation with ADLL treatment, indicating enhanced ADLL catabolism, it failed to reach statistical significance.
Leucine can affect mitochondria-mediated energy production through its effect on glucose utilisation [34]. Therefore, we tested the potential effect of ADLL treatment on energy metabolism by measuring the expression of the master regulator of mitochondrial biogenesis, peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC1-α). No detectable differences in PGC1-α protein expression were observed in any group of mice evaluated (Figure 3c,d), which implied no alteration of mitochondrial biogenesis occurred with ADLL treatment.

ADLL Promotes Anaerobic Glycolysis via PDK4 Up-Regulation
We then tested ADLL's effect on glucose utilisation by examining enzymes that regulate aerobic glycolysis, via pyruvate dehydrogenase (PDH), and anaerobic glycolysis, via lactate dehydrogenase Leucine can affect mitochondria-mediated energy production through its effect on glucose utilisation [34]. Therefore, we tested the potential effect of ADLL treatment on energy metabolism by measuring the expression of the master regulator of mitochondrial biogenesis, peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC1-α). No detectable differences in PGC1-α protein expression were observed in any group of mice evaluated (Figure 3c,d), which implied no alteration of mitochondrial biogenesis occurred with ADLL treatment.

Combination Treatment of Miglustat and ADLL Results in an Additive Extension of Lifespan in Hexb -/-Mice
Finally, we combined ADLL with the substrate reduction therapy drug, miglustat (600 mg/kg/day), initiating both treatments from weaning (3 weeks of age). In agreement with previous studies [8], Hexb -/mice had a median survival of 111.5 days, while miglustat therapy significantly increased the life expectancy by a median of 30 days (median survival 141.5 days, p = 0.0088) ( Figure 5). ADLL monotherapy extended lifespan by a median of 9.5 days. However, miglustat and ADLL combination therapy resulted in 150 days survival time, i.e., 38.5 days longer than the untreated group (in median, p = 0.0014) (Figure 5), suggesting an additive effect of ADLL on miglustat therapy.

Combination Treatment of Miglustat and ADLL Results in an Additive Extension of Lifespan in Hexb -/-Mice
Finally, we combined ADLL with the substrate reduction therapy drug, miglustat (600 mg/kg/day), initiating both treatments from weaning (3 weeks of age). In agreement with previous studies [8], Hexb -/-mice had a median survival of 111.5 days, while miglustat therapy significantly increased the life expectancy by a median of 30 days (median survival 141.5 days, p = 0.0088) ( Figure  5). ADLL monotherapy extended lifespan by a median of 9.5 days. However, miglustat and ADLL combination therapy resulted in 150 days survival time, i.e., 38.5 days longer than the untreated group (in median, p = 0.0014) (Figure 5), suggesting an additive effect of ADLL on miglustat therapy.

Discussion
L-Leucine not only serves as a building block for protein synthesis but is also a potent activator of the mammalian target of rapamycin (mTOR), involved in many cellular processes, including protein synthesis, nutrient sensing, cell growth, metabolism, and glucose, lipid and glutamate utilisation [26,34,38]. In this study, we explored the modified leucine amino acid, ADLL, for its potential therapeutic effects in a mouse model of Sandhoff disease, as it had previously provided beneficial effects in another lysosomal storage disease, NPC1 (observational clinical study) [16] and in Tangier disease patient cells [17]. Here, we found that ADLL modestly but significantly extended lifespan and significantly slowed disease progression in Hexb -/mice. Furthermore, ADLL treatment reduced the most prominently accumulated glycosphingolipid, GA2, in both mouse liver and brain. In addition, ADLL increased autophagy, restored aerobic (PDH-dependent) and enhanced anaerobic (LDH-dependent) glycolysis, and returned the glutamate-metabolizing enzyme, GDH, to levels observed in Hexb +/+ mice.
GA2 is the most abundant stored GSL in the Hexb -/model ( Figure 1). Only this lipid showed a statistically significant reduction in response to ADLL treatment. As other GSLs (e.g., GM2) were not reduced, these data suggest that ADLL is not acting as a substrate reduction therapy drug and may indicate instead that the GA2 reducing effect of ADLL occurs through an indirect mechanism. The precise mechanism will require further studies to elucidate.
Although 12-week-old Sandhoff cerebellum had no difference in their p:T mTOR ratio, ADLL treatment caused a slight reduction in total mTOR levels, and significantly elevated autophagy in Hexb -/cerebellum. These findings may explain how ADLL induced the clearance of stored lipids through autophagy induction [40]. It is possible that the N-acetylation of leucine in ADLL blocks the interaction of leucine with the intracellular amino acid sensor, Sestrin 2, causing reduced mTOR activation, as was reported in HeLa cells [41,42], which could enhance autophagy.
Evidence that ADLL did not affect leucine catabolism was seen from the lack of alteration in BCKADH activity (p:T) in Hexb -/cerebellum. However, an elevated level of glutamate dehydrogenase (GDH) is in agreement with a recent metabolomics study of Hexb -/mice [43].
Mitochondrial energy metabolism is a major source of cellular ROS production [44], which reacts with lipids, proteins and DNA, and contributes to cell death [44]. Elevated oxidative stress as a result of ROS is a crucial pathology that contributes to neurodegeneration in Alzheimer's disease, Parkinson's disease and many other neurodegenerative conditions [45]. Similarly, Sandhoff disease has been found to manifest oxidative damage upon inflammation, which fosters neural death [46]. Leucine has been proposed to have differential effects on metabolism, depending on the catabolic and anabolic states. Several studies have observed that leucine-treated cells exhibited improvement in mitochondrial function and oxygen consumption [30]. However, another study found that leucine enhanced anaerobic glycolysis and therefore reduced OXPHOS-dependent ROS burden [38]. Similarly, leucine-treated rats on a high-fat diet (HFD) exhibited improved insulin sensitivity, reduced gluconeogenesis, improved lipid oxidation and mitochondrial function in obese/diabetic stage [47]. However, ADLL was found to contribute to mitochondrial dysfunction in high-fat diet (HFD)-fed rats at the early stage of insulin resistance [34,47], suggesting context-dependent effects.
Recent metabolomic studies of Sandhoff disease mice and patient samples reported alterations in energy metabolism, where high levels of ROS and markers of oxidative stress were interpreted as increased respiratory chain activity [43]. In accordance with this, we found that PDH, a key indicator of aerobic respiration [48], was more active (55.9% reduction of p:T PDHE1-α levels) in Hexb -/compared to Hexb +/+ mice. This finding was consistent with a downregulation of PDK2, which phospho-inhibits PDH. ADLL treatment promoted deactivation of PDH by increasing PDK4 levels [49,50] and increasing LDH levels, thus normalizing glycolytic pathways to levels observed in Hexb +/+ mice. Previous studies reported that PDK4 overexpression increased autophagy (elevated LC3-II levels) [51], and increased lactate production [36], which may explain the elevation in autophagy and LDH levels in Hexb -/cerebellum with ADLL therapy. In a previous study, non-acetylated leucine had a similar effect in porcine intestinal epithelial cells, where it reduced the PDH-dependent OXPHOS pathway, and increased LDH-dependent glycolysis, thus reducing the OXPHOS-derived ROS burden [38]. A leucine-mediated reduction in ROS levels is consistent with the upregulation of the ROS scavenger, SOD1, which we observed with ADLL treatment. However, how ADLL upregulates PDK4, autophagy and lipid oxidation, and how this may link to a reduction of glycosphingolipid storage, remains unclear. Overall, changes in metabolic pathways with ADLL may be linked to leucine's activatory effect on homeostatic AMP-activated protein kinase (AMPK), which regulates the change in nutrient utilisation, anabolic/catabolic shift and glycolysis preference [52]. AMPK can promote glycolysis as a survival mechanism upon oxidative stress [53]. AMPK can also promote oxidative metabolism via increasing mitochondrial biogenesis and OXPHOS capacity [54], correlating with a dual metabolic response upon leucine treatment in different systems. Furthermore, β-hydroxy-β-methyl butyrate, a minor metabolite of leucine, has been reported to stimulate AMPK activation insulin sensitivity and fat oxidation [55]. Therefore, the molecular mechanism of ADLL and its enantiomers, and their effects on AMPK merit further investigation in order to develop novel pharmacological approaches for lysosomal storage diseases. For a summary, see Scheme 1.
[54], correlating with a dual metabolic response upon leucine treatment in different systems. Furthermore, β-hydroxy-β-methyl butyrate, a minor metabolite of leucine, has been reported to stimulate AMPK activation insulin sensitivity and fat oxidation [55]. Therefore, the molecular mechanism of ADLL and its enantiomers, and their effects on AMPK merit further investigation in order to develop novel pharmacological approaches for lysosomal storage diseases. For a summary, see Scheme 1. Scheme 1. Summary of ADLL treatment effects on the Hexb -/mouse cerebellum at 12 weeks of age.
Investigated pathways that leucine can affect are displayed. Red arrows indicate the alterations in untreated Hexb -/-mice, while green arrows indicate the effect of pre-symptomatic ADLL treatment.
Miglustat reversibly inhibits glucosylceramide synthase, the enzyme that catalyses the first committed step in GSL biosynthesis, and was previously shown to slow disease progression in Sandhoff disease mice [8][9][10]. When ADLL and miglustat were combined, there was an additive effect in relation to extension in life expectancy, extending the life span by almost 5 weeks in total.
In summary, we have found that ADLL treatment of Hexb -/-mice partially reduced lipid storage, modestly improved behavioural parameters, significantly extended lifespan and provided additive benefit when combined with miglustat. This study provides mechanistic insights into novel pathological alterations in Sandhoff disease brain metabolism-some of which were corrected by ADLL treatment, highlighting the potential of this drug for treating Sandhoff disease and other lysosomal disorders. Miglustat reversibly inhibits glucosylceramide synthase, the enzyme that catalyses the first committed step in GSL biosynthesis, and was previously shown to slow disease progression in Sandhoff disease mice [8][9][10]. When ADLL and miglustat were combined, there was an additive effect in relation to extension in life expectancy, extending the life span by almost 5 weeks in total.
In summary, we have found that ADLL treatment of Hexb -/mice partially reduced lipid storage, modestly improved behavioural parameters, significantly extended lifespan and provided additive benefit when combined with miglustat. This study provides mechanistic insights into novel pathological alterations in Sandhoff disease brain metabolism-some of which were corrected by ADLL treatment, highlighting the potential of this drug for treating Sandhoff disease and other lysosomal disorders.