Dual Therapy with Liraglutide and Ghrelin Promotes Brain and Peripheral Energy Metabolism in the R6/2 Mouse Model of Huntington’s Disease

Neuronal loss alongside altered energy metabolism, are key features of Huntington’s disease (HD) pathology. The orexigenic gut-peptide hormone ghrelin is known to stimulate appetite and affect whole body energy metabolism. Liraglutide is an efficient anti-type 2 diabetes incretin drug, with neuroprotective effects alongside anorectic properties. Combining liraglutide with the orexigenic peptide ghrelin may potentially promote brain/cognitive function in HD. The R6/2 mouse model of HD exhibits progressive central pathology, weight loss, deranged glucose metabolism, skeletal muscle atrophy and altered body composition. In this study, we targeted energy metabolism in R6/2 mice using a co-administration of liraglutide and ghrelin. We investigated their effect on brain cortical hormone-mediated intracellular signalling pathways, metabolic and apoptotic markers, and the impact on motor function in HD. We here demonstrate that liraglutide, alone or together with ghrelin (subcutaneous daily injections of 150 µg/kg (ghrelin) and 0.2 mg/kg (liraglutide), for 2 weeks), normalized glucose homeostatic features in the R6/2 mouse, without substantially affecting body weight or body composition. Liraglutide alone decreased brain cortical active GLP-1 and IGF-1 levels in R6/2 mice, alongside higher ADP levels. Liraglutide plus ghrelin decreased brain insulin, lactate, AMP and cholesterol levels in R6/2 mice. Taken together, our findings encourage further studies targeting energy metabolism in HD.

Peripheral and central energy metabolism are affected by gut peptide hormones, such as glucagon-like peptide-1 (GLP-1) and ghrelin (reviewed in 13 ). GLP-1 is a key determinant of blood glucose homeostasis, due to its ability to slow down gastric emptying and enhance pancreatic insulin secretion 14 . Ghrelin exerts metabolic effects throughout the body and increases body weight by enhancing appetite 15 .
GLP-1 mimetics (such as exendin-4 and liraglutide) have been shown to exert neuroprotective effects, being beneficial in Parkinson's disease (PD) 16 , Huntington's disease (HD) 17 and Alzheimer's disease (AD) 18 mouse models, and in clinical studies involving PD and AD patients 19 . Similarly, ghrelin and its analogues have been shown to have neuroprotective effects in PD and AD mouse models 20,21 .
Since HD is associated with weight loss and GLP-1 mimetics have anorectic properties, it is challenging to use GLP-1 mimetics in HD. We therefore hypothesize that peripheral injection of liraglutide, together with the orexigenic peptide hormone ghrelin, would maintain body weight and protect against brain/cognitive dysfunction in HD.
In this study, we investigated metabolic alterations in the R6/2 mouse model of HD. We used a pharmacological approach to determine the effects of subcutaneous (s.c.) co-injection of liraglutide and ghrelin on brain cortical hormone-mediated intracellular signalling pathways, metabolic and apoptotic markers, and the impact on motor-cognitive function in the R6/2 mouse model of HD.

Results
Liraglutide alone and in combination with ghrelin normalize peripheral glucose homeostasis in R6/2 mice. Ghrelin has been shown to increase body weight 15 , while liraglutide has been shown to exert the opposite effect 22 . The R6/2 mouse model displays a progressive weight loss. We therefore monitored body weight twice weekly, starting at 9 weeks of age. Subcutaneous administration of vehicle (NaCl), liraglutide alone or in combination with ghrelin was conducted for 2 weeks, starting at 10 weeks of age. As detailed in Material and Methods, at the end of treatment 12-weeks old mice were fasted for ~6 h (starting late in the evening) before euthanasia, and blood and brain cortices collected for remaining measurements. Similar to previous studies, weight loss was present in 12-week old R6/2 mice compared to wildtype (WT) littermates ( Supplementary Fig. 1a). Except for the 1.4-fold lower fat mass composition induced by liraglutide per se, the 2-week treatment strategy chosen did not affect R6/2 mouse body weight or body composition at 12 weeks of age ( Supplementary Fig. 1).
Food consumption was assessed daily during 14 days, from 11 weeks of age in an additional treatment group. Food consumption was not different comparing R6/2 treated with liraglutide in combination with ghrelin with R6/2 mice treated with only liraglutide. Food consumption differed compared to WT mice treated with vehicle the first 2 days of the 2-week treatment period and on day 12 and 14 (see Supplementary Fig. 2 for body weight and food consumption).
Both liraglutide and ghrelin have been shown to affect glucose homeostasis 13 . We therefore evaluated the effect of liraglutide alone and in combination with ghrelin on glucose homeostasis parameters in 12-week old R6/2 mice and WT littermates (Table 1). In this study, R6/2 mice exhibited increased serum glucose levels, in accordance with previous findings 23 . Interestingly, both liraglutide alone and together with ghrelin normalized serum glucose levels in 12-week old R6/2 mice (by 1.7-and 1.5-fold, with P = 0.0019 and P = 0.0136, respectively). HOMA-IR and HOMA-β indexes (homeostasis models to assess insulin resistance and β-cell function, respectively) 24,25 were also calculated ( Table 1). HOMA-β was significantly decreased in 12-week old R6/2 mice compared to WT littermates, which was rescued after treatment with liraglutide alone and in combination with ghrelin (by a 2.7-and 3.5-fold increase, with a P = 0.0452 and P = 0.004, respectively) ( Table 1). HOMA-IR was significantly increased in 12-week old R6/2 mice compared to WT littermates, which was normalized after treatment with liraglutide alone and in combination with ghrelin (by a 1.7-and 1.4-fold increase, respectively, with a P = 0.0074 and P = 0.0298) ( Table 1).
Co-administration of liraglutide and ghrelin appeared to exert distinct effects, since the even lower brain insulin levels in R6/2 mice (by 1.8-fold, P = 0.045) compared to vehicle treated mice (Fig. 1b) were not followed by significant changes in the remaining hormones levels (Fig. 1a,c).
Liraglutide in combination with ghrelin moderately affects glycolysis and energy metabolism in R6/2 mouse brain cortex. Brain energy metabolism has been shown to be altered in R6/2 mice 26 .
We therefore evaluated the possible effects of liraglutide and ghrelin on tricarboxylic acid cycle-and energy metabolism-related parameters in cortex from R6/2 mice and WT mice.
Liraglutide in combination with ghrelin normalize brain triglyceride, and cholesterol levels in R6/2 mice. The cholesterol biosynthetic pathway has been shown to be impaired in HD, with alterations in cholesterol and triglycerides levels 27 . Regarding the possible use of triglycerides, free fatty acids (FFA) and cholesterol as alternative brain metabolic substrates, we observed that, although not significant, there was a trend towards a change in brain cortical triglyceride, cholesterol and FFA levels. Cortical triglyceride and cholesterol levels were ~3-and ~5-fold higher (P = 0.1555 and P = 0.0985) in saline-treated R6/2 mice, whereas their brain FFA were ~3 times lower (P = 0.8611) than in WT littermates ( Fig. 3a-c). Liraglutide per se or upon co-administration with ghrelin normalized brain triglyceride and cholesterol levels to nearly those of WT mice (Fig. 3a,c). These results suggest that liraglutide, alone or co-administrated with ghrelin, may attenuate the use of triglycerides and cholesterol as brain alternative metabolites.

Effect of liraglutide in combination with ghrelin on R6/2 mouse brain cortical apoptotic pathways.
Low levels of brain caspase activity have been shown to affect axonal function 28 , and cognitive deficits 29 , whereas increased caspase activity may lead to inhibition of autophagy 30,31 . We therefore investigated the effects of liraglutide and ghrelin on caspase activities in R6/2 and WT mice brain cortex.
A 2-fold increase in brain cortical Caspase-10-like activity was found in vehicle treated R6/2 compared to WT mice, while no significant changes were observed for the other initiator Caspases-1-, -2-, -8-, -9-like activities ( Fig. 4a-f). These may account for unchanged Caspase-3 mRNA and Caspase-3 enzyme activity (Fig. 4g,h), and Caspase-6-like activity ( Fig. 4i) under these conditions. The 2-fold stimulation of Caspase-6-like activity induced by liraglutide per se in R6/2 mouse brain ( Fig. 4i) was not accompanied by significant changes in the remaining Caspase-like activities or mRNA under these conditions.
Co-administration of liraglutide plus ghrelin resulted in a 1.3-fold stimulation of brain cortical Caspase-12-like activity in R6/2 mice (Fig. 4c,f) that, nonetheless, had no significant impact on the activities of the effector Caspases-3 and -6-like ( Fig. 4h,i).
Liraglutide in combination with ghrelin does not affect motor and exploratory activities in R6/2 mouse. Motor deficits along with a clasping behaviour phenotype are main features of HD mouse models, which progress with the disease 32,33 . However, in this study we could not find any significant differences between experimental groups in terms of the locomotor and exploratory activity parameters evaluated during the open-field test ( Fig. 5a-e), nor in R6/2 mice after liraglutide and ghrelin co-administration (Fig. 5a,c). However, an increase in paw clasping to a score of 2 was observed in vehicle-treated R6/2 mice compared with WT mice ( Table 2) that, nonetheless, was not significantly changed by liraglutide alone or in combination with ghrelin ( Table 2).

Discussion
Accumulating evidence over the last decade supports the concept of HD being a metabolic disorder. Animal studies have demonstrated that targeting peripheral energy metabolism might have beneficial effects on both central and peripheral pathology in HD 3,17 . HD is associated with progressive weight loss and a lower body mass index (BMI), and a higher BMI has been shown to correlate with slower disease progression 34 . As a catabolic state is present in human HD 35,36 and HD mouse models 32 , this suggests that normalization of energy metabolism might be beneficial.
Metabolic effect of drug treatment can of course be related to possible alterations in food intake. It has previously been shown that R6/2 mice display increased food intake in comparison to WT littermates 37 . In our 2-week study there was no change in food intake comparing treated groups, however a decreased food intake in comparison to WT littermates could be noted during a few days of treatment period. Since food consumption was similar the majority of days, food consumption-induced differences should be marginal. However, of course this should be considered when interpreting results.
Although hyperglycaemia along with reduced insulin levels has been shown previously in R6/2 mice 23 , in the present study we found that elevated levels of serum glucose were accompanied by normal insulin levels, increased HOMA-IR and decreased HOMA-β in R6/2 mice compared to WT littermates. This may be explained by the fact that our R6/2 mice presented an early to middle stage disease phenotype and, thus, insulin production was not dramatically affected yet. This was in line with previous studies demonstrating a variation in disease progression depending on the CAG repeat size 41,42 .
Activation of the GLP-1 receptor by liraglutide and other analogues, such as exendin-4, has been shown to exert anti-diabetic effects, by improving pancreatic β-cell function and glucose regulation in HD mouse models 17,43 . Conversely, studies involving the effect of ghrelin administration on insulin secretion β-cell in other rodent and in vitro models have given contradictory results, with both inhibitory 44 and beneficial effects 45 . In the present study, liraglutide injected alone or together with ghrelin normalized peripheral basal glucose levels and both HOMA-β and HOMA-IR (two mathematical models widely used to evaluate β-cell function and insulin resistance, respectively, from fasting glycemia and insulin or C-peptide levels 24,25 ) in R6/2 mice, suggesting a beneficial effect on glucose homeostasis, insulin resistance and pancreatic β-cell function. Mitochondrial dysfunction and oxidative stress have been shown to contribute to R6/2 mouse central pathology, probably resulting from elevated insulin/IGF-1 signalling pathways and subsequent progression of neurodegeneration 46 . Conversely, others have suggested that a compensatory increase in brain insulin levels may overcome systemic insulin resistance in R6/2 mice 47 . In line with the observed peripheral insulin resistance, elevated brain insulin levels in our R6/2 mice were normalized after 2 weeks of co-administration of liraglutide and ghrelin. Decreased brain cortical active GLP-1 and IGF-1 levels after liraglutide administration in R6/2 mice suggested that an activation of compensatory mechanisms could occur to overcome their lower brain hormone levels and maintain downstream cAMP-and/or ERK1,2-mediated signalling pathways.
The unchanged brain cortical ATP levels in vehicle treated R6/2 mice was in line with previous studies in HD mice [48][49][50] and early stage HD patients 51 . This reinforced the idea that, though mutant HTT may hamper brain glycolysis, it may not directly affect mitochondrial oxidative phosphorylation, thus preserving ATP synthesis 48,49 .
Since the later depends mostly on ADP levels, the maintenance of ATP levels in vehicle treated R6/2 mouse may occur at the expense of brain ADP and AMP levels 52 . This may be accompanied by an increment in total creatine levels (creatine plus phosphocreatine, an alternative source of ATP) 52 . Of note, Mochel et al. 24 found that striatal ATP and phosphocreatine levels were inversely correlated with the number of CAG repeats in HD mice. Alternatively, we cannot exclude that the decreased brain ADP and AMP in vehicle treated R6/2 mice may arise from the maintenance of their cAMP and/or of adenosine levels, a critical molecule in neurotransmission and energy metabolism, whose increased extracellular content was beneficial in R6/2 mouse brains 53 . More recently, Kao et al. (2016) 54 suggested that a compensatory biphasic regulation of striatal adenosine tone in HD rodents may arise with disease progression. This may be accompanied by an increased expression and activity of striatal adenosine kinase in late-stage R6/2 mice, ultimately converting adenosine into AMP or inosine and thereby decreasing its levels 54 . Additionally, the tendentiously lower brain cortical lactate and ATP levels, and higher ADP levels associated with subcutaneous liraglutide treatment in R6/2 mice suggested that it may not only partially inhibit their brain glycolysis and energy metabolism, but the lower amount of ATP formed may be also used to further yield ADP and cAMP, thus maintaining their intracellular pools.
Caspases are pivotal players in both apoptosis and autophagy, controlling the turnover of protein aggregates and elimination of damaged cells 55 . However, Caspases-3 and -6 were also found to play other non-apoptotic roles in the CNS 56 . The significant increase in brain Caspase-10-like activity of vehicle treated R6/2 mice compared to WT littermates was not significantly affected by either liraglutide alone or in combination with ghrelin. Despite no significant changes in Caspase-6-like or Caspase-12-like activities in vehicle treated R6/2 mice, liraglutide alone significantly increased Caspase-6-like activity, and in combination with ghrelin stimulated Caspase-12-like activity in R6/2 mice. This could indicate that both liraglutide alone or co-administered with ghrelin may exert some neuroprotection in HD. Unchanged Caspase-3 mRNA expression, and Caspase-3 and -6-like activity suggests that caspase-mediated apoptosis may not be involved in R6/2 mouse neuronal death and locomotor deficits.
Although R6/2 mice display progressive motor dysfunction 30,31 , there were no effects on their motor dysfunction in the present study, evaluated by open field and paw clasping behaviour testing. This could have several reasons but, most likely, the short treatment time was not long enough to translate into functional motor-cognitive recovery. Hence, further studies using longer treatment strategies, possibly starting at an earlier age, are encouraged.
In conclusion, our data demonstrate beneficial effects of liraglutide alone and in combination with ghrelin on peripheral glucose homeostasis in R6/2 mice. Co-administration of liraglutide and ghrelin decreased brain cortical insulin, lactate, AMP and cholesterol levels in R6/2 mice, while liraglutide alone decreased brain cortical active GLP-1 and IGF-1 levels in R6/2 mice, alongside higher ADP levels.
Our results support further studies targeting (brain) energy metabolism, which might exert beneficial effects on HD progression. Evaluating underlying effects of liraglutide and ghrelin administration in HD are warranted.

Material and Methods
Animals. All experimental procedures performed on mice were carried out in accordance with the approved guidelines in the ethical permit approved by The Malmö/Lund Animal Welfare and Ethics Committee (ethical permit number: M5-15).
At the end of the 2-week treatment, mice were fasted for ~6 h (starting late in the evening) and afterwards body composition was measured using the Lunar Prodigy dual energy x-ray absorptiometry (DEXA; GE Lunar Corp., Madison, WI) and thereafter mice were euthanized, blood was collected and brain was dissected. Tissue samples were snap-frozen in liquid nitrogen and stored at −80 °C until further use. Serum obtained from blood samples was collected by centrifugation at 2000 × g, for 10 min, at 4 °C, and immediately frozen to −80 °C. Serum Analyses. Serum glucose levels were measured using the glucose oxidase method (Infinity Glucose Hexokinase Kit, Thermo Scientific, Middletown, VA, USA), and levels of insulin were determined by the Mercodia Mouse Insulin ELISA Kit (Uppsala, Sweden). The homeostatic model assessment (HOMA) was calculated to measure insulin resistance (IR) and pancreatic beta-cell function (β). HOMA-IR was calculated as follows: (Insulin X Glucose)/22.5, and HOMA-β as follows: (20 × Insulin)/(Glucose − 3.5) 24 .

Isolation and Preparation of Brain Cortical Homogenates.
Mice were weighed and euthanized by decapitation, and brains were immediately removed. Brain cortices were immediately dissected and snap-frozen for further studies. Immediately before the experiments, brain cortices were homogenized at 0-4 °C in lysis buffer, containing (in mM) the following: 25 HEPES, 2 MgCl 2 , 1 EDTA, 1 EGTA, (pH 7.4), supplemented with 2 mM DTT, 100 μM PMSF, and commercial protease and phosphatase inhibitors cocktails. The homogenate was centrifuged at 17,968 × g for 10 min, at 4 °C, in a Sigma 2-16 K centrifuge to remove the nuclei, and the resulting supernatant was collected. The pellet was further resuspended in supplemented buffered solution and centrifuged again at 17,968 × g for 10 min, at 4 °C. The supernatant was added to the previously obtained one and protein content was measured using Pierce BCA Assay Kit (Thermo Scientific, Rockford, IL, USA) according to manufacturer's protocol.
Active GLP-1 levels were measured in 5 μL of each brain cortical homogenate (working dilution of 1:5) by the Fluorescent ImmunoAssay kit for Active GLP-1 (7-36) Amide (Human, Rat, Mouse, Porcine, Bovine, Canine, Ovine) Ultra-sensitive (Phoenix Pharmaceuticals, Inc; Karlsruhe, Germany). Fluorescence was determined using an excitation and emission wavelengths of 325 and 420 nm, respectively, in a SpectraMax Gemini EM multiplate fluorescence reader. Results were expressed as pg/mL/mg protein.
Brain cortical insulin levels were measured in 10 μL of each brain cortical homogenate, by using the above mentioned Mercodia Mouse Insulin ELISA Kit. Results were expressed as pg/mL/mg protein.
Brain cortical IGF-1 levels were measured in 5 μL of each brain cortical homogenate (working dilution of 1:20) by the Rat IGF-1 ELISA kit (Biosensis Pty Ltd; Thebarton, South Australia). Absorbance was read at 450 nm, in a SpectraMax Plus 384 microplate reader. Results were expressed as pg/mL/mg protein.
cAMP levels were determined in 7.5 μL of each brain cortical homogenate (working dilution of 1:10) with the cAMP Direct Immunoassay Kit (Colorimetric) (BioVision; Milpitas, CA, USA). Absorbance was read at 450 nm, in a Biochrom Asys Expert 96 UV Microplate Reader (Cambourne, Cambridge, UK). Results were expressed as pmol/mg protein.
Relative PKA activity was determined in 5 μL of each brain cortical homogenate (working dilution of 1:6) by the PKA kinase activity kit (Enzo Life Sciences, Farmingdale, NY, USA). The absorbance was determined at 450 nm, in a SpectraMax Plus 384 multiplate reader. Results were expressed as ng/µL/mg protein.
Western Blot Analyses. Samples containing denatured brain cortical homogenates (25 μg per lane) were subjected to sodium dodecyl sulphate (SDS)/polyacrylamide gel electrophoresis (SDS/PAGE) (8-15%) and transferred onto polyvinyl difluoride (PVDF) membranes. Then, membranes were blocked for 1 h at room temperature in Tris-buffered saline (TBS, pH 7.4) plus 1 or 5% nonfat dry milk or bovine serum albumine (BSA), plus 0.05% Tween 20. Membranes were then incubated overnight at 4 °C with rabbit IRβ (4B8) (1:1000) or rabbit WT +NaCl R6/2 +NaCl R6/2 +Liraglutide R6/2 +Liraglutide +Ghrelin Paw clasping score 0.5 2** 2 1 Table 2. Effect of peripheral liraglutide plus ghrelin co-injection on paw clasping scores in R6/2 mice. Paw clasping phenotype was scored upon mouse suspension by the tail for 180 s. A score of 0 represents no clasping behaviour, 1 occurs when the hind paws touch each other for at least 1 s, and 2 mean that hind paws clasp for 5 s or more. R6/2 mice showed a paw clasping phenotype at 12 weeks of age compared to WT littermates. However, no significant improvement was seen with 2 weeks administration of either liraglutide alone or together with ghrelin. Data are medians of 10 animals/group, as described in Materials and Methods. **P < 0.01 vs. salinetreated WT mice, by the Kruskal-Wallis test, with Dunn post-test (non-Gaussian distribution).