Abstract
Neurodegeneration refers to the selective and progressive loss-of-function and atrophy of neurons, and is present in disorders such as Alzheimer’s, Huntington’s, and Parkinson’s disease. Although each disease presents with a unique pattern of neurodegeneration, and subsequent disease phenotype, increasing evidence implicates alterations in energy usage as a shared and core feature in the onset and progression of these disorders. Indeed, disturbances in energy metabolism may contribute to the vulnerability of neurons to apoptosis. In this review we will outline these disturbances in glucose metabolism, and how fatty acids are able to compensate for this impairment in energy production in neurodegenerative disorders. We will also highlight underlying mechanisms that could contribute to these alterations in energy metabolism. A greater understanding of these metabolism-neurodegeneration processes could lead to improved treatment options for neurodegenerative disease patients.
Keywords
Introduction
Neurodegenerative disorders, including Alzheimer’s disease (AD), Huntington’s disease (HD) and Parkinson’s disease (PD) are characterised by a progressive loss of specific neurons that results in both functional and cognitive deficits. 1 All three diseases display a unique pattern of neuronal loss, which is associated with different symptoms and pathological features. In AD, pyramidal neurons in the Ammon’s horn of the hippocampus are highly vulnerable to damage, and their loss leads to gradual deterioration of cognitive function and memory. 2 This degeneration is associated with abnormal extra-neuronal deposition of amyloid-β peptides (Aβ) in senile plaques and the intraneuronal accumulation of hyper-phosphorylated tau in neurofibrillary tangles (NFTs). 2 HD is an autosomal dominant genetic disorder caused by a longer than usual CAG expansion in exon 1 of the HTT gene, encoding an expanded polyglutamine (polyQ) tract near the N-terminus of the huntingtin protein. 3 Neuronal loss in HD is observed in the medium-sized projection neurons of the dorsal striatum, with symptoms including involuntary movements (chorea), cognitive decline and psychiatric symptoms. 3 PD is characterised by motor abnormalities such as tremors, muscular rigidity and bradykinesia, caused by the progressive loss of dopaminergic neurons in the substantia nigra pars compacta (SNpc) and the degeneration of projecting nerve fibres to the striatum. 4 The exact pathophysiological mechanisms underlying neuronal loss in these neurodegenerative disorders are poorly understood, although perturbations in glucose metabolism appear to be a common feature in all of these diseases. 1 The degenerating neurons in all three diseases are reported to be particularly vulnerable to metabolic stress for a variety of reasons, as previously reviewed by Muddapu et al. 1 In this review, we provide an overview of the common disturbances in both glucose uptake and metabolism in the aforementioned neurodegenerative disorders, and how increased fatty acid metabolism can restore energy supply. We also discuss the long-term disadvantages for the brain of relying on fatty acids as an alternate fuel source, and present hypotheses for other pathways that could be therapeutically targeted to restore energy homeostasis in these neurodegenerative disorders.
Glucose uptake and metabolism in neurodegeneration
Initially it was believed that the reduction of glucose uptake and metabolism was a secondary response to the atrophy and neuronal loss that occurs in these neurodegenerative disorders. However, increasing evidence suggests that changes in energy metabolism may occur prior to significant brain atrophy and the onset of clinical symptoms. This suggest that changes in glucose metabolism may increase the vulnerability of these cells and consequently contribute to neuronal loss. 5 The brain is comprised of several different cell types, all with distinctive metabolic phenotypes. Neurons are the highest energy demanding cell in the brain, consuming up to 80% of oxygen in the brain. 6 This alone suggests that neurons are highly sensitive to disruptions in glucose metabolism and mitochondrial dysfunction. Multiple neurodegenerative disorders including AD, HD and PD are commonly associated with a loss of glucose uptake and, perturbations in several glucose-related pathways including glycolysis, pentose phosphate pathway (PPP), and tricarboxylic acid (TCA) cycle and oxidative phosphorylation (OXPHOS). 6 As neurons have little regenerative capacity, it is of importance to reduce the local environment stressors to neurons that render them more vulnerable to death. 7 Understanding the mechanisms of metabolic disturbances during disease progression has become a major focus of recent research, and interventions in these processes may relieve the neurons from these metabolic stressors and reduced degeneration.
Alzheimer’s disease
Accumulation of Aβ and phosphorylated tau are believed to contribute to early cognitive decline, however the findings from one recent study in 551 AD patients suggests that the aggregation of these proteins alone may not be sufficient to develop clinical symptoms. 8 Interestingly, glucose hypometabolism appears to be crucial to further the cognitive decline between mild cognitive impaired and clinically diagnosed AD patients. 8 Using [18F]-fluoro-deoxyglucose (18FDG) positron emission tomography (PET) scans, various studies have revealed a correlation between cortical hypometabolism and cognitive decline in AD patients, even at early stages of disease.9–11 Glucose uptake is reduced by up to 45% in the hippocampus, parieto-temporal lobe, posterior cingulate cortex and frontal cortex of AD patients. 9 Young cognitively normal individuals with ApoE4 mutations showed similar patterns of reduced glucose uptake decades before the expected onset of clinical symptoms of early onset familial AD. 10 Furthermore, a 2-year longitudinal study indicated that ApoE4 carriers at risk for familial AD showed the greatest decline in glucose uptake in the parietal and temporal cortices, where there is extensive early neuropathological legions in patients with AD. 11 However, the results are more conflicting during the symptomatic stages of disease, with some studies indicating that ApoE4 is associated with decreased glucose metabolism, whilst others found no association. 12 As disturbed glucose metabolism can impair the functional capacity of neurons and subsequently contribute to neuronal death, this loss of glucose uptake in these memory-related cortical regions has been associated with cognitive impairments. 13 In a more recent longitudinal study, a reduction in glucose metabolism measured by FDG-PET imaging was strongly associated with cognitive decline in patients presenting with both Aβ and phosphorylated tau pathology. 14 Although these findings show that glucose hypometabolism in the central nervous system (CNS) may contribute to the functional and structural changes observed in AD, there are also conflicting results. A recent study of 232 Aβ-positive patients with mild cognitive impairment or AD, concluded that decreased FDG-PET correlated with local atrophy, suggesting it is a consequence rather than a contributing factor to neurodegeneration. 15 Furthermore, it must be noted that these studies were not sufficiently sensitive to detect glucose uptake at a cellular level. Thus, it cannot be determined which cells are contributing to the change in glucose utilisation.
Along with a decrease in glucose uptake, the expression of glucose transporters was also altered in autopsied AD brains. The expression of both GLUT1 and GLUT3 were decreased in the cerebral cortex of AD patients and animal models, with a loss of GLUT3 also found in the dentate gyrus of the hippocampus of patients.16,17 As GLUT3 is primarily expressed on neurons it is important to note that this loss in expression was still apparent after correcting for neuronal loss. 17 In contrast, GLUT2, which is exclusively expressed in astrocytes, was elevated over 2-fold in the frontal cortex, thus these changes may be more representative of the altered brain cellular composition at the late stages of disease. 16 The activities of several enzymes involved in glucose metabolism, downstream of uptake have been noted to be impaired in individuals with AD. Microarray analysis of post-mortem brains identified that 15 of the 51 genes associated with glycolytic, TCA cycle and oxidative phosphorylation were reduced in AD. 18 Specifically two regulatory enzymes in the glycolytic pathway, hexokinase and phosphofructokinase are reduced in the post-mortem brain.19,20 Thus, together with the finding of increased glucose accumulation in the brains of AD patients, 21 this supports the hypothesis that deficits in glycolytic activity may contribute to a loss of glucose usage and energy production in AD. However, as these studies were conducted in post-mortem brain samples, it is possible that these decreases in both the glucose transporters and enzyme activity could be due to either neuronal loss or a change in cellular composition within the brain, rather than an early pathological feature.
Additionally, mitochondrial dysfunction has been shown to contribute to metabolic deficits in AD.22–24 Both bioinformatic and protein expression studies have identified the oxidative phosphorylation pathway as one of the most affected processes in AD.22,24 Several studies have investigated the activities of the enzymes involved in oxidative phosphorylation. Collectively these studies have reported between a 10–50% loss in the activities of Complexes I–IV in different cortical regions, with most significant reductions observed in complex II, III and IV activity.23,24 However at the mRNA level, the results are more controversial, with some studies reporting a reduction in complex I, IV and V expression, whilst others have shown increases in complex III and IV.25,26 These changes have been found in both early and late-stage AD patients. 26 The activities of complexes I and III were also reduced by 45% in the brains of TgCRND8 mice, a model of AD which overexpresses the human mutant amyloid precursor protein (APP). 27 Moreover, the mitochondrial membrane potential was 40–60% lower in two cell lines (7WD4 and 7PA2) that are used as cellular models of AD, compared to Chinese hamster ovary cell controls. 28 In addition to the loss in ETC activity, a 41% and 57% reduction in the activities of key enzymes in the TCA cycle, pyruvate dehydrogenase complex (PDHc) and oxoglutarate dehydrogenase complex (OGDHc) were found in the post-mortem brain from AD patients. 29 Interestingly, the activity of transketolase, an enzyme involved in non-oxidative branch of the pentose phosphate pathway, was also lower in AD brains. 30 All three enzymes are multi-enzyme complexes, that include thiamine diphosphate as one of the co-enzymes. Thiamine diphosphate levels were reduced by up to 21% in autopsied brain samples from AD patients compared to controls, suggesting that this loss in thiamine diphosphate contributes to the loss of these enzyme complexes. 31 PDH E1α expression was also reduced by 19% in mitochondria isolated from the brains of 9-month-old 3xTg mice. 32 Overall, this suggests that glucose metabolism is reduced in AD due to dysfunction in glucose uptake, as well as reductions in glycolytic and TCA cycle activity (Figure 1). This impairment of glucose metabolism, and thus ATP production, increases the vulnerability of cells to functional and structural changes, and may contribute to neuronal apoptosis.
Disturbances in cellular glucose metabolism in neurodegenerative disorders. In
Huntington’s disease
Similar to AD, cellular glucose metabolism is also perturbed in the CNS in HD. Striatal glucose uptake, quantified by functional imaging studies using 18F-FDG PET scans, was reduced in symptomatic HD patients. 33 Reduced glucose uptake in caudate, putamen and thalamic regions were positively correlated with disease severity. 34 In newly diagnosed patients, this loss in glucose uptake was restricted to the frontal and inferior parietal cortex; however, in patients who had displayed symptoms for longer than 5 years, a global loss of glucose uptake was observed. 33 Both early symptomatic patients and asymptomatic mutant huntingtin (mHTT) protein carriers displayed reduced glucose uptake in the caudate nucleus and putamen, with the loss of glucose uptake correlating with the number of CAG repeats. 35 In addition, reduced striatal 18F-FDG uptake was found in 30-50% of mHTT carriers and chorea-free high-risk individuals, suggesting that the abnormalities in CNS glucose metabolism precedes the clinical or structural alterations in HD patients. 36 Although these studies are unable to determine what cells are responsible for this loss in glucose uptake, a study utilising primary cortical neuronal cultures from the HD140Q/140Q mouse model showed a reduction in glucose uptake. 37 Consistent with the results using in vivo imaging, glucose transporters were also reduced in post-mortem samples of HD patients. 38 The expression of GLUT1 and GLUT3 were reduced by three- and four-fold, respectively, in the post-mortem samples of caudate and cortex from patients with HD. 38 However, the expression of these transporters were unchanged in post-mortem samples from patients in the earlier stages of disease. Thus, this loss, in particularly of GLUT3 may be more indicative of the significant neuronal degeneration in these brain regions. 38 Furthermore, overexpression of human GLUT3 (hGLUT3) in drosophila HD models rescued the neurological dysfunction and increased life expectancy. 39 Overall this demonstrates that glucose uptake is reduced in the HD cortex, in part due to the loss of glucose transporter expression.
Alongside a loss in glucose uptake, glycolytic activity is impaired in HD. 40 Lactate accumulation, which suggests an increase in glycolytic flux, was also demonstrated in the striatum and occipital cortex of HD patients using localized 1H spectroscopy. Interestingly, lactate concentrations positively correlated with CAG repeat numbers. 41 This increased flux through glycolysis is not necessarily due to changes in the striatal neurons themselves. In fact, activated microglia and reactive astrocytes are both present in the HD striatum, and their levels correlate with disease severity in HD patients. 42 Both these cells are known to increase their glycolytic rates in these activated states, which suggests that these cell populations may be contributing to the increased glycolytic flux observed in HD.
Similarly to AD, mitochondrial metabolism has also been reported to be impaired in both patients and animal models of HD. Complex II activity and expression levels were shown to be reduced in HD mouse models and patients.43,44 These reductions were limited to the striatum, suggesting complex II activity could directly contribute to the neuronal vulnerability in HD. Further, inhibition of complex II via chronic administration of 3-NP in rats lead to the development of striatal lesions, along with HD-associated motor deficits such as bradykinesia and wobbling gait. 45 However, iPSC-derived striatal neurons and astrocytes from unaffected individuals showed no differences in mitochondrial mass, mitochondrial membrane potential or the activities of mitochondrial enzyme activities, suggesting that mitochondria may not be affected in the early stages of the disease process. 46 Along with deficits in the complex activity, a loss of the TCA cycle enzymes PDHc and OGDHc was also reported in post-mortem samples from HD patients, indicating that mitochondrial dysfunction is evident in HD similar to AD, at least in the later stages of disease (Figure 1).47,48
Parkinson’s disease
An early study found that a global loss in cerebral glucose uptake was observed in PD patients using 18F-FDG PET scans in comparison to healthy controls, with no selective metabolic changes in the striatum. 49 This loss in uptake was even more severe in PD patients with dementia. Both blood flow and glucose uptake were reduced on the contralateral side of the frontal cortex to the side of the body with muscle deficits in patients with unilateral disease. 49 Likewise, a more recent study has revealed widespread glucose hypometabolism in PD patients with mild to severe cognitive impairments in the parietal, occipital, temporal and frontal lobes. 50 Furthermore, parieto-occipital and temporal glucose hypometabolism was found to be correlated with poor performance in cognitive assessment. 51 However, this pattern of reduced glucose uptake could be detected even in patients without significant cognitive impairment. This suggests that altered cortical glucose metabolism can occur independently of cognitive decline and is further exacerbated as cognitive states worsen. Thus, like AD and HD, PD is associated with reduced brain glucose uptake in regions that are vulnerable to neuronal death. Interestingly, although a decline in glucose uptake is observed in PD and considered to be detrimental, pre-treatment with 2-deoxyglucose protected human dopaminergic neurons against rotenone-induced death. Moreover, 2-deoxyglucose treatment decreased the vulnerability of SNpc dopaminergic neurons to 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-induced damage. 52 This response only seemed to occur when 2-deoxyglucose was administered prior to the toxic insult, therefore it is believed that the protection may be due to the conditioning of cells, which induces upregulation of alternative metabolic pathways, such as the pentose phosphate pathway. 52
To the best of our knowledge no data has been collected on the expression of glucose transporters or the activities of enzymes involved in glycolysis in PD. However, several studies have shown that enhancing glycolysis is beneficial in animal models of PD.53–55 Exogenous glucose 6-phosphate isomerase, a glycolytic enzyme, protects drosophila and primary mouse neurons from proteotoxicity induced by α-synuclein. 53 Furthermore, the anti-histamine drug meclizine, which increases glycolytic flux, protected against the neurotoxicity of 6-hydroxydopamine (6-OHDA) in SH-SY5Y cells and rat primary cortical neurons. 54 Most recently, the α1-adrenergic receptor blocker terazosin, which binds to and enhances the activity of phosphoglycerate kinase 1, and subsequently glycolysis, attenuated neurodegeneration and improved behavioural phenotypes in both the MPTP mouse, and 6-OHDA rat models of PD. 55 Terazosin or similar drugs were also shown to reduce the risk of developing PD by 12 to 37% in men, compared to those who received an α1-adrenergic receptor blockers that did not enhance glycolysis. 56 Overall, this shows that modifying and enhancing glycolytic flux may be neuroprotective in both in vitro and in vivo models of PD.
Mitochondrial dysfunction is also recognised as a disease modifier of PD. Complex I activity is reduced in the post-mortem brain of parkinsonian patients. The maximal respiration was also reduced by up to 25% in mitochondria isolated from the ventral midbrain and striatum of mice overexpressing human wild-type α-synuclein. 57 In the same study, Complex I activity was also reduced by 40% in mitochondria isolated from the ventral midbrain. 57 This loss of Complex I activity was specific to regions containing nigrostriatal dopaminergic neurons. 57 A significant decline in complex I activity was also observed when wild-type α-synuclein accumulated in mitochondria of cultured human dopaminergic neurons. This decline in activity was further exacerbated in cells that were incubated with α-synuclein carrying the A53T mutation, which is known to be present in familial PD patients, suggesting that complex I function is particularly vulnerable in PD. 58 The key role of complex I activity in the progression of PD was further supported by a study investigating MPTP-induced PD symptoms in patients that took illicit drugs. A metabolite of MPTP, 1-methyl-4-phenylpyridinium (MPP+) accumulates in the mitochondria of dopaminergic neurons and inhibits complex I activity, leading to the loss of ATP production and increased reactive oxygen species, resulting in dopaminergic neuron loss. 59 Genetically deleting the core subunit of Complex I, NADH – ubiquinone oxidoreductase core subunit S2 in the dopaminergic neurons of mice was sufficient to drive neuronal loss and the development of parkinsonian-like symptoms. 60 Despite this, patients with genetic complex I deficiencies rarely develop a parkinsonian phenotype, suggesting that dysfunction in oxidative phosphorylation alone is not sufficient to develop PD. 61 Like what is observed in AD, the activities of both PDHc and OGDHc are affected in PD. In the lateral third of the SNpc, OGDHc immunostaining was reduced, and correlated with the level of degeneration in post-mortem brain samples. 62 OGDHc activity was also reduced by approximately 50% in the cerebellum of PD patients, suggesting this loss in activity is not limited to the affected brain regions in PD. 63 Furthermore, pyruvate dehydrogenase alpha 1 (PDHA1) protein, which forms the core of the PDHc, is relocated from the mitochondrial matrix to Lewy bodies in the PD brain, which is likely to result in the loss of activity. 64 Thus, mitochondrial dysfunction is a common feature of neurodegenerative disorders, potentially reducing mitochondrial ATP production and consequently contributing to neuronal vulnerability in these diseases (Summarised in Figure 1).
Are the changes in glucose metabolism a cause or consequence of disease
AD, HD, and PD all show similar profiles in metabolic dysfunction, including reduced glucose uptake and mitochondrial dysfunction. If these disturbances are contributing to neuronal degeneration, it begs the question as to why such similar metabolic phenotypes results in different patterns of neurodegeneration, and consequently different clinical symptoms. Several reports suggest that protein aggregation, which is another common feature of these neurodegenerative diseases occurs prior to, and contributes to, the disturbances in energy metabolism, in particular mitochondrial dysfunction. In AD, Aβ deposition has been reported as the triggering event that induces impairments in glucose metabolism, brain atrophy and cognitive impairment. Several studies have shown that glucose uptake is reduced in neurons exposed to Aβ. In primary hippocampal cultures, Aβ25–35 reduced glucose uptake by inhibiting the docking and fusion of GLUT3 to the plasma membrane. 65 Another study also reported that this could be mediated through Aβ-induced oxidative stress. 66 The HTT protein which has been implicated in HD, has also been shown to regulate energy metabolism. Knocking out HTT in mouse embryonic stem cells severely limited ATP production via oxidative phosphorylation whilst increasing glycolytic activity, potentially as a compensatory mechanism, suggesting that wild-type HTT protein plays a role in regulating energy metabolism pathways. 67 The mHTT was also shown to negatively corelate with the ATP/ADP ratio in both mouse embryonic stem cells and human lymphoblasts, indicating that mHTT in HD can promote loss of energy metabolism. 68 However, another study found no change in ATP levels in knock-in embryonic stem cells containing one copy of humanised exon 1 with an extended polyglutamine tract (HTT-Q140/Q7). 67 HTT has also been shown to be localised to the mitochondrial outer membrane, with mHTT reportedly lowering mitochondrial membrane potential via defects in calcium handling. 69 Similarly, evidence indicates that α-synuclein also preferentially accumulates in mitochondria, and this accumulation leads to mitochondrial defects in PD. 70 Aggregated α-synuclein inhibited activity of both Complex I, which leads to reduced respiration and ATP production. 71 Furthermore, α-synuclein accumulation reduced oxygen consumption and reduced cell viability in neurons, astrocytes and skin-derived fibroblasts. 72 Overall, this suggests that protein aggregation is driving at least some of the metabolic disturbances observed in these neurodegenerative disorders.
The degenerating neurons in all three disorders are all uniquely vulnerable to metabolic stress as they typically require more energy. In AD, hippocampal neurons located in the CA1 region are the most vulnerable to degeneration as they have higher firing rates, which means they have a higher energy demand. 47 In HD, GABAergic striatal projection neurons are the initial cell type affected. These neurons are termed medium spiny neurons due to their long axons and spiny dendrites. These long axons may increase their vulnerability. Furthermore, these neurons usually maintain a hyperpolarised state, which requires high amounts of energy. 73 Finally, dopaminergic neurons in the SNpc characteristically have long unmyelinated axons with a large amount of synapses. This neuronal architecture would require large amounts of energy to maintain membrane potential and propagate action potentials. 74 Therefore, regardless of whether impaired glucose metabolism is causative or a consequence of other features of these neurodegenerative disorders, there is specific evidence that upregulating glucose metabolism can delay clinical symptoms. Therefore, it would be beneficial to further understand the complexities in the regulation of glucose metabolism and oxidative phosphorylation during neurodegeneration.
Lipids can replace glucose as a fuel in neurodegenerative disorders
It is evident that glucose metabolism is impaired in neurodegenerative disorders, via a multitude of mechanisms, including reduced glucose uptake, along with a loss in glycolytic flux and mitochondrial dysfunction. As the brain requires constant energy supply to maintain its functions, evidence suggests that the brain can adapt to using different fuel sources such as fatty acids and ketone bodies to support ATP production. 75 Fatty acids and ketones usually contribute to less than 5% of the brain’s metabolism. But during a prolonged fast, ketone bodies are elevated in the plasma and can contribute up to 60% of the brain’s energy requirements. 75 Therefore, approaches which support brain energetics, such as ketone treatments, could potentially slow disease progression that is due to energy deficiency.
Alzheimer’s disease
Many studies have investigated whether elevating ketone levels in the blood could improve symptoms in both AD patients and rodent models. A single dose of MCFAs, which are readily metabolised to ketone bodies in the liver, improved scores for the Alzheimer’s Disease Assessment Scale–Cognitive Subscale (ADAS-Cog) and paragraph recall in ApoE4-negative AD patients. 76 These acute effects are assumed to be caused by the use of ketone bodies as an energy source. Another acute study found no cognitive improvement in 20 Japanese patients with mild-to-moderate AD following the ingestion of MCFAs. 77 However, in these same patients, treatment with 1 g/kg/day of MCFAs for 8 weeks improved some cognitive functions compared to their baseline scores, despite no increases in plasma βHB levels. 77 It must be noted that this study did not identify the participants according to ApoE4 genotype. Several other long term trials that did stratify patients based on their ApoE4 genotype determined that MCFA or βHB treatment was only beneficial for patients who were ApoE4-negative.78,79 A 90-day placebo-controlled trial of caprylic acid, a MCFA, in mild to moderate AD patients improved ADAS-Cog scores in ApoE4-negative patients. 78 Another cross-over study showed that 30-day MCFA supplementation in ApoE4 negative AD patients led to a greater improvement in ADAS-Cog scores than the placebo treatment. 79 It is suggested this genotype-dependent effect of MCFA treatment is due to the differences in pathophysiology between ApoE4 positive and negative patients. Mitochondrial function has been shown to be lower in ApoE4 positive individuals which may limit use of ketones. However, further research is required to elucidate the mechanisms that underlie this phenotype.
Huntington’s disease
Ketogenic diets and βHB supplementation have been trialled in rodent models of HD, with several studies showing promising responses. R6/2 mice fed a ketogenic diet from 6 weeks of life showed a 13-fold increase in plasma βHB concentrations, delay in weight loss and complete reversal of impairments in working memory in female R6/2 mice at 12 weeks of age. 80 Interestingly, the ketogenic diet had no effect on working memory in R6/2 males, however it must be noted that unlike in females, working memory was already impaired in males prior to the implementation of the diet. 80 So, it seems that the diet can only delay the onset of cognitive deficits in this model. A ketogenic diet rescued some motor deficits in R6/2 mice when fed from birth. 81 The mice fed a ketogenic diet travelled further and faster in the open field test, but the ketogenic diet had no effect on the latency to fall on the accelerating rotarod test. 81 Despite an increase in brain concentration of βHB, the ketogenic diet did not prevent brain atrophy in the R6/2 model. 81 In comparison, βHB infusion significantly delayed the onset of motor deficits in the R6/2 mouse model and extended lifespan by approximately 30% on average. 82 In the 3-NP model of HD, βHB infusion completely ameliorated striatal neuronal loss and partially improved locomotor activity. 82 Together this suggests that fatty acid supplementation is neuroprotective and able to partially slow disease progression in HD.
Parkinson’s disease
MCFA and βHB treatment improved symptoms in both patients and rodent models of PD. βHB increased survival by up to 60% in cultured neurons that were incubated with both MPP+ and rotenone.83,84 βHB also improved mitochondrial membrane potential and reduced cytochrome C release in a rotenone induced in vitro model of PD. 84 Twenty-four hour infusion of βHB using mini osmotic pumps also protected SNpc dopaminergic neurons against MPTP toxicity, and reduced motor abnormalities in mice. 85 OXPHOS is enhanced in these neurons via a mechanism independent of complex I activity, which circumvents the known deficits in complex I activity in PD. 85 This study also reported βHB treatment improved neuronal cellular respiration and ATP production, improved motor skills and elevated dopamine concentrations in the mesencephalon of mice following MPTP injections. 85 Octanoic acid, a MCFA, also reduced the MPTP-induced depletion of dopamine in the striatum. 86 Similar to βHB, octanoic acid is reported to be neuroprotective by increasing mitochondrial metabolic activity in the striatum. 86 In a recent study, PD patients underwent a 8-week dietary intervention, where they were administered either a low fat or ketogenic diet. 87 Interestingly, both diets led to improvements in non-motor and motor scores, assessed using the Movement Disorder Society – Sponsored Revision of the Unified Parkinson’s Disease Rating Scale (MDS-UPDRS). 87 However, the ketogenic diet improved non-motor scores significantly more than the low-fat diet, with the biggest differences found in fatigue, pain, daytime sleepiness and cognitive impairments. 87 This could possibly result from supplying neurons with an alternative energy source and thereby increasing ATP production, but further studies would be required to clarify the underlying causes of this improvement.
Harmful side effects of fatty acid metabolism in the brain
Even though increasing fatty acid metabolism has shown some positive outcomes in treating neurodegenerative disorders, there are two main reasons why relying on β-oxidation to fuel neurons long term may not be beneficial. Firstly, to generate the same amount of ATP, fatty acids require approximately 15% more oxygen than when using glucose as a fuel. 88 Thus, primarily using fatty acids as a fuel source increases the risk of hypoxia. Secondly, β-oxidation increases mitochondrial electron leakage, consequently increasing superoxide generation. 89 Fatty acids can also bind to the ETC complexes, which impairs their activity. This not only reduces ATP generation, but also increases the production of ROS. 88 Elevations in oxidative stress are known to induce cellular damage, impairment in the DNA repair system and mitochondrial dysfunction, all of which have been known as key factors in the development of neurodegenerative disorders. 90 Therefore, long-term reliance on fatty acid β-oxidation to produce ATP could be harmful, by increasing oxidative stress. Instead of elevating fatty acid and ketone body metabolism as a therapy for neurodegenerative disorders, it may be more beneficial to understand the mechanisms that are driving the loss in glucose uptake and metabolism and try to develop therapies that can directly restore glucose metabolism.
Does a loss of insulin signalling reduce glucose uptake in neurodegenerative disorders?
One potential mechanism that contributes to a loss of glucose uptake in neurodegenerative disorders is insulin resistance. Insulin is produced in the β-cells of pancreatic islets and is secreted in response to elevated blood glucose. 91 By binding to the insulin receptor, insulin facilitates glucose disposal in tissues such as adipose tissue and skeletal muscle. Insulin resistance occurs when there is a breakdown along this signalling pathway. 91 This loss in insulin signalling is not limited to the periphery. Although the brain was once considered to be an insulin-insensitive organ, more recent reports have found both insulin and its receptors localised to neurons in the brain. 92 This suggests that insulin signalling could play an important role in regulating glucose uptake and metabolism in these neurodegenerative disorders.
Recently, the role of phospholipids in insulin resistance has been receiving increased attention. It has been proposed that insulin resistance may be a consequence of lipid disorders such as abnormalities in cell membrane phospholipid composition. 93 Modifications of membrane composition, in particular phosphatidylcholine (PC), phosphatidylethanolamine (PE) and ceramides, alter membrane fluidity and as a consequence perturb the insulin signalling and the translocation of glucose transporters to the cell membrane. A change in the PC/PE ratios in cell membranes of various tissues can influence energy metabolism. Although it is yet to be determined the role that phospholipid composition plays in managing insulin resistance and glucose uptake in neurodegenerative disorder, studies have shown an accumulation of certain phospholipids can inhibit glucose uptake.
Alzheimer’s disease
Diabetes mellitus (specifically type II diabetes) has been proposed as a risk factor for AD and decline in cognitive function. 94 A longitudinal cohort study of 824 individuals identified a 65% increase in the risk of developing AD in individuals with diabetes mellitus when adjusted for age, sex, and educational level. Using a random effects model they found that diabetes mellitus was associated with lower levels of cognition, memory, and visuospatial ability. 95 However, others failed to show a correlation between type II diabetes and AD pathology.96,97 Furthermore, in several diabetic mouse models of AD, increased tau phosphorylation is observed. 98 Indeed, defects in insulin receptor function, induced in rats by intracerebroventricular injection of streptozotocin, led to impaired learning, memory and cognition, along with decreased glucose uptake and oxidative glucose metabolism. 99 Similarly, 3xTg-AD mice with a diabetic phenotype induced by streptozotocin had exacerbated memory impairments with greater accumulation of toxic Aβ and phosphorylated tau in the cortices and hippocampus. 100 Thus, it is evident that there is a connection between underlying mechanisms that regulate the diabetic and AD phenotypes.
An expanding body of evidence also suggests that insulin abnormalities and insulin resistance contribute to AD pathophysiology. 101 Higher plasma insulin concentrations and Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) scores in healthy middle-aged people were shown to predict cognitive decline and increase the risk of developing AD later in life. 102 Furthermore, AD patients were shown to be glucose intolerant, despite increased plasma insulin concentrations in both fasting states and following an oral glucose tolerance test (OGTT). 103 Insulin resistance and glucose intolerance also promoted AD-like neuropathology in mouse models of AD. 104 Diet-induced insulin resistant Tg2576 mice were found to have a 2-fold increase in Aβ plaques in the hippocampus. This increase in Aβ plaques was correlated with an earlier decline in cognitive ability. 104 Peripheral insulin resistance is also reported to correlate with Aβ deposition in the frontal and temporal cortex. 105 However, it must be noted that some studies contradict these results, thus it is clear that the role that peripheral insulin resistance plays in the development and progression of AD is not completely understood.
There also appears to be a disruption in the relationship between central and peripheral insulin levels in AD patients. The ratio of CSF/serum insulin was reduced in AD patients compared to that of healthy controls, and this reduction negatively correlated with the severity of disease. 106 The expression of the insulin receptor (IR), insulin-like growth factor 1 (IGF-1), IGF-1 receptor (IGF-1R) were also reduced in post-mortem brain samples, suggesting that CNS insulin signalling is compromised in AD. 107 Along with this decline in the levels of expression of these receptors, their ability to bind to cell membranes may be impaired in AD. An increase in saturated fatty acid content in phospholipid species decreases membrane fluidity and reduces membrane bound insulin receptors. 108 In contrast, the presence of polyunsaturated fatty acids increases fluidity and has been associated with improved insulin sensitivity. 108 This shift from unsaturated towards saturated fatty acids, is also proposed to affect glucose transporters translocation to the cell membrane. 108 Polyunsaturated fatty acids were significantly reduced in both PC and PE species in the CSF of patients with AD, suggesting that cell membranes in AD have less fluidity. 109
Downstream signalling of the insulin receptor is also impaired in AD due to inactivation of insulin receptor substrate 1 (IRS-1), by serine phosphorylation. IRS-1 acts as an intermediary between IR/IGF-1R and the downstream phosphatidylinositol 3-kinase (PI3K)-protein kinase B (AKT) signalling cascade. 91 Thus, the loss of IR and IGF-1R expression, along with inactivation of IRS-1 suggests that neuronal insulin signalling is impaired in AD. Indeed, PI3K and AKT phosphorylation was reduced in post-mortem brain samples from AD patients. 110 Furthermore, the levels and activation of the insulin-PI3K-Akt signalling components correlated negatively with the level of tau phosphorylation and positively with tau O-GlcNAcylation. 110 This suggests that impaired insulin-PI3K-Akt signalling might contribute to neurodegeneration in AD through decreased O-GlcNAcylation and consequent tau hyperphosphorylation. Changes in cell membrane composition may also contribute to this loss in the insulin-PI3K-Akt signalling pathway. Ceramides, members of the sphingolipid family, have also been proposed to stabilise the cell membrane structure and modulate the distribution of receptors. Increased proportions of ceramides in the cell membrane antagonise insulin signalling by disrupting PI3K/Akt signalling, thus impairing glucose uptake. 111 Ceramide levels were shown to be elevated in autopsied brain tissue from AD patients, along with several enzymes involved in ceramide formation. Higher serum ceramides levels were associated with an increased risk of AD later in life. 112 Furthermore, inhibiting ceramide production protected neurons from Aβ-induced cell death. 112 Collectively, these results indicate that insulin resistance, due to a loss of insulin signalling, could play a role in the pathogenesis of AD (Figure 2).
Mechanisms of insulin resistance in neurodegenerative disorders. Diabetes mellitus and insulin resistance are common features of the neurodegenerative disorders, AD, HD, and PD, however the underlying mechanisms that drive a loss in both insulin and glucose tolerance are unique for each disease. In
Huntington’s disease
Diabetic-like symptoms have also been identified in HD patients as well as in rodent models. 113 The correlation between diabetes and HD has been documented in two early studies that found a higher prevalence of diabetes mellitus in HD patients than in control population.114,115 It was also suggested that affected relatives of a HD proband with diabetes were 7 times more likely to develop diabetes than the non-HD affected relatives. 115 However, there are also studies that show contradictory conclusions associated with glucose homeostasis in HD patients. Two separate studies involving 21 HD and 21 age-matched control subjects demonstrated no difference in glucose and insulin responses to an OGTT.116,117
Unlike in AD, insulin secretion rather than insulin signalling seemed to be the primary contributor to the development of the diabetes-related phenotype in HD. Several studies have shown that at least some patients with HD display abnormal glucose tolerance during an OGTT.118,119 In one study, although patients showed similar glucose tolerance, they were shown to be less insulin sensitive using two different indexes, the HOMA-IR and the insulin sensitivity index. 120 Early-phase insulin secretion was also reduced in these HD patients, and it suggested that this loss of insulin secretion underlies the loss in insulin sensitivity. 120 Boesgaard et al. (2009) demonstrated that the capacity of β-cells to release insulin in response to glucose was reduced in HD patients, and this was negatively correlated with the length of the polyQ region in mHTT. 118 End-stage R6/2 mice were hyperglycaemic and hypo-insulinemic, and both basal and glucose-stimulated insulin secretion were reduced in isolated islets. 121 Both β-cell mass and insulin content within the pancreatic islet cells were significantly reduced in R6/2 mice at end-stage compared to wild-type mice. 121 Insulin mRNA levels in the pancreas were also shown to be approximately 50% and 25% of that in wild-types at 8 weeks and 12 weeks of age, respectively. 122 This loss in insulin production and secretion is further supported by the outcomes of various diabetic drug treatments in mouse models of HD. R6/2 mice responded to glibenclamide, which stimulates insulin release. However, no response was observed when mice were treated with rosiglitazone, which increased peripheral insulin sensitivity. 123 Exendin-4 treatment, another anti-diabetic drug that indirectly stimulates insulin release was also shown to be beneficial in the N171-82Q mouse model of HD. Exendin-4 alleviated the disturbed peripheral glucose metabolism and impaired pancreatic function, as well as delayed the onset of mortality in these mice, most likely due to a combination of peripheral and central effects of exendin-4. 124 Together, these findings suggest that glucose uptake might be impaired due to the loss of pancreatic β-cell mass and insulin secretion in HD (Figure 2).
Although the impairments in insulin signalling in HD seem to be due to a loss of insulin synthesis and release rather than insulin resistance, a change in phospholipid membrane composition could still impact the translocation of glucose transporters to the cell membrane, and ultimately reduce the glucose transporters in the membrane. In plasma from HD patients, both PC and PE levels were decreased in correlation to clinical progression. 125 Whereas, in the CSF of these same patients PC levels were elevated and negatively correlated with some functional assessments. 125 PE levels were unchanged in the CSF, suggesting that in the CNS the PC/PE ratio is increased and may contribute to a loss of function in HD. 126 Interestingly, Htt14A2.6 PC12 cells, which express the HTT exon 1 containing either 25Q or 97Q repeats have increased in membrane fluidity. 127 However, similar to AD, ceramide accumulation has been shown to occur early in disease progression in the R6/2 mouse model of HD. 128 Thus, it appears further investigations are needed to fully understand how membrane fluidity is affected in HD, and what role phospholipid composition plays in regulating that fluidity, and in turn glucose uptake.
Parkinson’s disease
A large body of evidence has shown that type II diabetes increases the risk of developing PD by up to 40-50%. 129 Indeed, the presence of diabetic phenotype in PD has been shown to be associated with more severe impairments in parkinsonian features, specifically cognitive decline, postural instability and gait difficulty. 130 However, a meta-analysis investigating the relationship between diabetes and PD, found that the association was not significant. 131 Between 50–80% of patients with PD display an abnormal response to OGTT, and increased insulin resistance. 132 Furthermore, in response to an OGTT, glucose clearance was reduced in PD patients, with no change in plasma insulin concentrations compared to healthy controls. 133 In this study reduced glucose clearance was associated with a longer disease duration and an increase in the severity of symptoms. 133 Thus, glucose intolerance in PD patients is associated with a more severe phenotype, accelerated disease progression and an increased risk of developing dementia.
Several lines of evidence suggest that the insulin response in the brain may play a role in the development of PD. For example, insulin receptor immunoreactivity was shown to be reduced in the SNpc in post-mortem brain samples of PD patients. 134 Furthermore, in the rat 6-OHDA model of PD, serine phosphorylation of IRS-2 was elevated, consequently leading to a loss of downstream AKT phosphorylation. 135 Moreover, intranasal insulin treatment in rats ameliorated decline of locomotor activity and motor performance induced by 6-OHDA in comparison to non-treated 6-OHDA-lesion rats. 136 In the same study, insulin treatment was shown to alleviate 6-OHDA-induced injury in the SNpc, further supporting neuroprotective role of insulin. 136 Moreover, post-treatment with exendin-4 reduced methamphetamine-induced rotational behaviour, and increased dopamine production in the SN of mice injected with both 6-OHDA and MPTP.137,138 In mesencephalic and SH-SY5Y neuronal cell cultures, known to be rich in dopaminergic neurons, exendin-4 protected dopaminergic cells from 6-OHDA damage. 138 Thus, as in AD, it seems that insulin resistance contributes to disease progression and neuronal loss in PD (Figure 2).
Unlike in AD and HD, little is known about the mechanisms that drive insulin resistance in PD. However, similar changes in membrane lipid composition have been observed in PD, when compared to these other neurodegenerative disorders, suggesting that a change in membrane fluidity could contribute to a loss of insulin signalling in PD. The incorporation of polyunsaturated and saturated fatty acids into cell membranes is altered in PD, with the unsaturation index of lipid rafts from the frontal cortex reduced by 38% and 52% in PD and incidental PD patients, respectively. 139 This is caused by both a dramatic decrease in both n-3 and n-6 long chain poly unsaturated fatty acids, along with an increased incorporation of saturated fatty acids in phospholipids. 139 This results in a more viscous membrane, and potentially reduced brain insulin signalling in PD. Furthermore, ceramide levels are elevated in the plasma and post-mortem brain samples from PD patients. 140 Higher plasma ceramide levels were also associated with a disease severity. 140 Overall, these results show that cell membrane fluidity is reduced in PD due to changes in phospholipid composition. To date it is unknown how these changes alter insulin signalling in PD or other neurodegenerative disorders, but it would be on interest to investigate these links in the future.
Is insulin resistance a consequence of protein aggregation in neurodegenerative disorders?
Similarly, to the deficits in energy metabolism, insulin resistance appears to be a common feature of these neurodegenerative disorders. There is evidence that protein aggregation is also driving this insulin resistance both in the periphery and CNS. Tau hyperphosphorylation drives insulin accumulation in SH-SY5Y neuroblastoma cells, which consequently decreases insulin receptor and GLUT4 expression. 141 Aβ oligomers also reduced phosphorylation of insulin receptor in response to insulin in mature hippocampal neurons. 142 Alternatively, ApoE4 expression in mice impaired insulin signalling in the brain by binding to the insulin receptor and inhibiting translocation to the cell membrane. 143 This may explain why some studies found altered glucose metabolism is associated with ApoE4 expression. In relation to HD and PD, both mHTT and α-synuclein aggregation have been shown to alter the activity of pancreatic islets. Glucose-stimulated insulin secretion was reduced in NIT-1 cells, a pancreatic β-cell line, that was transfected with HTT protein with 150 CAG repeats. In R6/2 mice, pancreatic β-cell dysfunction occurs after mHTT aggregates are observed in the nuclei of β-cells. Pathological phosphorylated α-synuclein deposits have also been found in pancreatic β-cells in over 90% of PD patients. 144 One study found that α-synuclein inhibits insulin secretion in mice pancreatic islets, through direct or indirect interactions between α-synuclein and the insulin granule KATP channel. 145 Ablation of α-synuclein in mice also induced glucose intolerance and impaired glucose-stimulated insulin secretion, where insulin secretion was reduced by 25% in the isolated islets. 146 In contrast, mice overexpressing human α-synuclein and acute peripheral administration of α-synuclein had a positive effect on glucose homeostasis and insulin secretion. 146 Thus, it appears that α-synuclein is important in regulating insulin secretion on pancreatic β-cells. More importantly, there are also studies that suggest insulin and insulin resistance can influence protein aggregation, suggesting these interactions are complex in these neurodegenerative disorders. 147 Indeed these processes become even more complicated if we consider, glycation, another mechanism that may link several pathways that are associated with neurodegeneration, including insulin resistance, mitochondrial dysfunction and protein aggregation. Glycation is the non-enzymatic reaction of reducing sugars or carbonyl compounds with proteins or lipids, wherein reactive carbonyl groups link to free amino groups. 148 This process can lead to the formation of advanced glycation end products (AGEs) which are known to disrupt the structure and function of proteins. 148 Indeed, glycation and AGE formation are reported to promote protein aggregation and disease progression in AD, HD and PD. 149 Glycation is also influenced by glucose metabolism, as methylglyoxal, an active carbonyl compound can be produced as a by-product of glycolysis, via the degradation of the triose phosphates, glyceraldehyde-3-phosphate and dihydroxyacetone-phosphate. 148 This would indicate that higher glycolytic rates would increase methylglyoxal levels. However, it is suggested that impairments in downstream pathways, including the glyceraldehyde-3-phosphate dehydrogenase activity and TCA cycle may contribute to methylglyoxal synthesis, due to the accumulation of glyceraldehyde-3-phosphate. 148 It must be noted that although there is some evidence to suggest that methylglyoxal levels are elevated in both diabetic patients and patients with these neurodegenerative disorders, the mechanisms underlying this elevation are not fully understood. 150 Furthermore, methylglyoxal is also reported to interfere with insulin secretion and signalling, however this is beyond the scope of this review, and is more thoroughly examined by Schalkwijk and Stehouwer. 150 Ultimately, it appears that it is not a single insult that underlie the selective vulnerability of specific neurons in each of these diseases. Rather it is a complex array of molecular interactions, whereby, disturbances in energy metabolism may be one aspect that contributes to this neuronal vulnerability that is observed in these neurodegenerative disorders. Clearly, more research into the complex interactions between protein aggregation, impaired glucose metabolism, mitochondrial dysfunction, oxidative stress, and inflammation is required to understand how they all contribute to neuronal degeneration.
Conclusion
This review has summarized the evidence that demonstrates impairments in glucose metabolism, in particular glucose uptake and mitochondrial dysfunction are a common feature of multiple neurodegenerative disorders. Although it is not yet understood whether these changes are a cause or consequence, glucose metabolism appears to play an important role in the progression of neurodegenerative disorders, likely due to the reduction in ATP synthesis which increases neuronal vulnerability to toxic and damaging stimuli. We have highlighted that although increasing fatty acid metabolism appears to be an effective strategy in overcoming this loss of glucose metabolism, long-term treatments have their drawbacks. Increased fatty acid metabolism is likely to increase ROS production, which is known to contribute to neurodegeneration. Therefore, we identified other common features of these neurodegenerative disorders that could be contributing to these deficits in glucose metabolism, particularly the glucose uptake. Future studies should address whether modifying the actions of insulin can increase brain glucose uptake and metabolism. In AD and PD, this target would need to address elevating tissue insulin signalling. Whereas, in HD, insulin secretion would need to be addressed. Moreover, future investigations need to identify if changes in cell membrane composition or protein aggregation is affecting insulin signalling and glucose uptake in neurodegenerative disorders and contribute to mitochondrial dysfunction. In conjunction, future research needs to further investigate the complex relationship between several molecular disturbances that occur in neurons which all interact and contribute to the vulnerability of neurons in these diseases. Although a significant amount of research has been conducted on this topic, due to the complexities of metabolism and many cellular processes that these pathways interact with, we are far from a complete understanding of the role that metabolism plays in these neurodegenerative disorders. Further investigation into these complex interactions may present new therapeutic approaches for treating neurodegenerative diseases.
Footnotes
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
