Glucose metabolism impairment in Parkinson’s disease

Impairments in systematic and regional glucose metabolism exist in patients with Parkinson's disease (PD) at every stage of the disease course, and such impairments are associated with the incidence, progression, and special phenotypes of PD, which affect each physiological process of glucose metabolism including glucose uptake, glycolysis, tricarboxylic acid cycle, oxidative phosphorylation, and pentose phosphate shunt pathway. These impairments may be attributed to various mechanisms, such as insulin resistance, oxidative stress, abnormal glycated modification, blood-brain-barrier dysfunction, and hyperglycemia-induced damages. These mechanisms could subsequently cause excessive methylglyoxal and reactive oxygen species production, neuroinflammation, abnormal aggregation of protein, mitochondrial dysfunction, and decreased dopamine, and finally result in energy supply insufficiency, neurotransmitter dysregulation, aggregation and phosphorylation of α-synuclein, and dopaminergic neuron loss. This review discusses the glucose metabolism impairment in PD and its pathophysiological mechanisms, and briefly summarized the currently-available therapies targeting glucose metabolism impairment in PD, including glucagon-likepeptide-1 (GLP-1) receptor agonists and dual GLP-1/gastric inhibitory peptide receptor agonists, metformin, and thiazoledinediones.


Introduction
Parkinson's disease (PD) is the second most common neurodegenerative disease with unclear etiology, characterized by dopaminergic neuron death and Lewy bodies in the surviving substantia nigra (SN) neurons (Chase et al., 1998;Shastry, 2001;Sulzer, 2007;Wakabayashi et al., 2013). Impaired glucose metabolism has recently been associated with PD based on: 1) poor control of glucose promotes the progression of motor symptoms in PD (Ou et al., 2021); 2) although no conclusive evidence supported that diabetes elevates risk of PD, several studies have suggested such an association (Cereda et al., 2011;Liu and Tang, 2021), and diabetes and high HbA1c are also related to severe neuroaxonal damage and cognitive impairment in PD (Uyar et al., 2022); 3) drugs causing energy metabolism dysfunction, such as rotenone and 1-Methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) induce parkinsonian symptoms (Langston, 1987); while glucose is the primary energy source of neurons (Dienel, 2019;Mergenthaler et al., 2013); These findings suggest roles of glucose metabolism impairment in the pathophysiology of PD. However, the specific mechanism by which glucose metabolism impairment participates in the pathophysiology of PD is unclear. This review discussed how glucose metabolism impairment, specifically hyperglycemia, participates in PD, and its mechanism.

Glucose metabolism in brain and neurons
Glucose is essential energy substance for brain (Mergenthaler et al., 2013). It participates in adenosine triphosphate (ATP) production, oxidative stress, neurotransmitter synthesis, neural function regulation, and the formation and maintenance of brain structure (Dienel, 2019). Under physiological conditions, glucose is continuously supplied to the brain, and provides almost all brain-required ATP (Dienel, 2019).
In the brain, the energy consumption of neuron signal transductionrelated activities accounts for 70% of the total energy consumption (Yu et al., 2018), among which Na-K-ATPase consumes more than 50% of brain energy consumption (Dienel, 2019), while the energy consumption of other non-signal transduction-related activities accounts for 30% (Yu et al., 2018). Excitatory neurons consume 80-85% of brain energy consumption, while that of inhibitory neurons consumes 15-20% (Yu et al., 2018). In the brain, the energy consumption of gray matter has obvious regional differences, and is significantly higher than that of white matter, while the energy consumption related to non-signal transduction activities in white matter is higher than that of gray matter (Dienel, 2019;Sokoloff et al., 1977). Notably, the energy metabolism of different regions in the brain changes with age, and the metabolism of bilateral frontal lobe, orbital gyrus, cingulate gyrus, hippocampus, basal ganglia, thalamus, temporal lobe and other parts gradually decreases with age (Mielke et al., 1998). The synthesis rate of local ATP in the brain is highly matched with the local energy demand (Dienel, 2019). Consistently, there is a highly regional and temporal correlation between brain functional activities and glucose metabolism, that is, glucose metabolism is elevated in active areas and active periods of brain (Dienel, 2019).
Adapting the violent fluctuations of local energy demand, brain presents high oxidative activity and has sufficient reserves of oxidative capacity (Dienel, 2019). Under certain conditions, the oxygen consumption of healthy brain can rapidly increase by two to three times and maintain for at least two hours (Siesjö, 1978), synaptic terminal can quickly increase its respiratory rate by six to ten times (Kauppinen and Nicholls, 1986). However, brain rarely uses its oxidative capacity reserve for energy supply, but increases its glucose metabolism rate instead (Dienel, 2019). Brain glucose utility is extremely efficient. Awake adult brain can complete the whole process of glucose metabolism from glucose uptake to complete oxidation within about 0.4-1 millisecond (the duration of once to twice synaptic transmission, or the duration of calcium channel to open and close once) (Dienel, 2019). Notably, substantia nigra compacta (SNc) dopaminergic neurons, which are known to play vital roles in the pathophysiology of PD, are especially susceptible to oxidative stress, plausible factor including: 1) SNc dopaminergic neurons present high energy demand due to their large axonal arborization and high number of axon terminal, which require high energy supply for maintaining of action potential conduction and axon terminal function, resulting in less respiratory reserve (Pacelli et al., 2015). Additionally, as autonomous pacemakers, dopaminergic neurons depend on L-type voltage-dependent Ca 2+ channels to maintain their auto-rhythmicity (Chan et al., 2007;Trist et al., 2019). Correspondingly, the density and activity of mitochondria in axons of SNc dopaminergic neurons are also high, providing higher activity of oxidative phosphorylation (Pacelli et al., 2015); 2) in the process of dopamine releasing from synaptic vesicles into cytosol, oxidants including H 2 O 2 , •O 2 -, •OH, and dopamine semiquinone radicals are produced (Trist et al., 2019).
Neurons mainly rely on oxidative metabolism to meet their energy requirements, while astrocytes mainly rely on glycolysis, the mechanisms including:1) neurons lack 6-phosphofructose-2-kinase/fructose-2,6-bisphosphatase-3 (Pkfkb3) due to its continuous proteasome degradation by anaphase-promoting complex/cyclosome (APC/C-Cdh1), the enzyme is involved in production of fructose-2,6bisphosphate which is responsible for activating glycolytic enzyme phosphofrukinase-1, while the activity of APC/C-Cdh1 is low in astrocytes, so high level of Pkfkb3 promote glycolysis process (Almeida et al., 2004;Herrero-Mendez et al., 2009); 2) in neurons, there is a balance between glycolysis and PPP, increase in glycolysis flux along with decrease in PPP flux which is responsible for antioxidative stress, it is reported that overexpression of Pkfkb3 activates neuronal glycolysis, but result in oxidative stress and apoptosis (Herrero-Mendez et al., 2009).
Both oxidative and nonoxidative process are involved in neuron activity to meet increased metabolic demands, correspondingly, cerebral blood flow (CBF) and cerebral metabolic rate of glucose consumption also increase, but more than oxygen consumption (Bélanger et al., 2011). Astrocytes play a role in regulating CBF according to statement of neurons, in other words, astrocytes induce vasoconstriction when neuronal is under resting state and oxygen is sufficient, and induce vasodilation when neuronal is active and oxygen is rapidly consumed (Ances et al., 2001;Bélanger et al., 2011;Offenhauser et al., 2005). In addition to glucose, various metabolic intermediates of glucose in brain can be energy substrates in brain, and lactate is an essential energy sources (Bélanger et al., 2011;Bouzier-Sore et al., 2006;Zielke et al., 2009). As mentioned above, astrocytes present a high glycolysis rate. A significant numbers glucose which metabolize via glycolysis is transferred to lactate and released to extracellular space (Bouzier-Sore et al., 2006;Itoh et al., 2003), neurons rapidly oxidize lactate to meet their energy demands when they are active, following with high glycolysis flux in astrocytes, which enables supplements of extracellular lactate pool (Bélanger et al., 2011). But there is still uncertain the main fuel (glucose or lactate) for neuronal activity (Dienel, 2012).
Glucose uptake and transport in brain mainly depend on glucose transporter 1 (GLUT1) in endothelial cells and astrocytes, and GLUT3 and GLUT4 in neurons (Ashrafi et al., 2017;Pearson-Leary and McNay, 2016;Simpson et al., 2007). Notably, glucose storage in brain can only sustain for three to four minutes at resting state (Dienel, 2019), brain glucose uptake can increase to two to three times than the rate at resting state if necessary (Gruetter et al., 1998;Holden et al., 1991).
Glucose is metabolized through four pathways (Dienel, 2019), including: 1) Glycolysis, one glucose molecule is turned into two pyruvic acid molecules, and generate two ATP molecules, together with substrates of the biosynthesis of some neurotransmitters, modulators and oxidative substrate; 2) Glycogenesis, glycose could be stored in astrocytes as glycogen, and acts as active and important participant in brain energetics, the turnover of which produces 50% more ATP than glycolysis (Dienel, 2019); 3) Tricarboxylic acid cycle (TCA), pyruvic acid produced by glycolysis is turned into acetyl-CoA, which enters TCA and produces carbon dioxide and water, together with 30 or 32 molecules of ATP (Dienel, 2019); 4) Pentose phosphate shunt pathway (PPP), which metabolizes glucose and generates reduced nicotinamide adenine dinucleotide phosphate (NADPH) and ribulose-5-phosphate (Dienel, 2019). Notably, in brain, both glucose and glycogen can be the precursor of glucose-6-phosphate in astrocytes, while neurons could only metabolize glucose (Dienel, 2019). Additionally, malate-aspartate shuttle and other pathways also provide a little energy (Dienel, 2019).
Generally, glucose metabolism in brain involves many rapid, complex and highly integrated processes that are highly adapted to brain function.

Glucose metabolism impairment in PD
Glucose metabolism abnormality has been identified in PD Marques et al., 2018), and is considered as one of its non-motor symptoms (Batisse-Lignier et al., 2013;De Pablo-Fernández et al., 2017).
Systematically, abnormal glucose tolerance occurs in 50-80% PD patients (Sandyk, 1993). The prevalence of diabetes in PD patients was reported as 10.02% (95% confidential interval (CI) 7.88 − 12.16) (Komici et al., 2021). Interestingly, hyperglycemia and diabetes could increase the risk of PD. A retrospective cohort study found that prediabetes (HbA1C between 5.7% and 6.4%) was independently associated with elevated risk of developing PD (hazard ratio (HR): 1.07, 95% CI:1.00-1.14) (Sánchez-Gómez et al., 2021). A large cohort study found that elevated fasting glucose increases the risk of PD by 1.185 folds (Rhee and Lee, 2021). Excessive fasting glucose variability was also associated with increased risk of PD, with each one standard deviation-unit increase in fasting glucose variability associated with 9% increase in the risk of PD in midlife population without diabetes (Chung et al., 2021). A nationwide data research from Austria (including 1.8 million patients) related both type 1 diabetes (HR: 2.3, 95%Cl: 1.9-2.7) and type 2 diabetes (HR: 1.5, 95%Cl 1.4-1.6) to increased incidence of PD (Klimek et al., 2015). Another large retrospective cohort study from United Kingdom, including more than 2 million type 2 diabetes patients and 6 million reference cohort patients, found increased incidence of PD in type 2 diabetes patients compared with reference cohort patients (HR 1.32 (95%CI: 1.29-1.35)) (De Pablo-Fernandez et al., 2018). Recently, a meta-analysis reported that diabetes can increase the risk of PD by 1.21 times (Chohan et al., 2021). Although there are several researches reported no correlation between the two diseases (Palacios et al., 2011;Savica et al., 2012), or even reported converse results (D'Amelio et al., 2009;Lu et al., 2014), most of researches support a close link between PD and diabetes. Notably, a study found a relationship between duration of diabetes and risk of PD, that is, only those with long diabetes duration (usually more than 5 years) may present increased risk of PD incidence (Jeong et al., 2020). Additionally, diabetes is related to increased occurrence of specific phenotypes in PD patients, such as postural instability, abnormal gait (Giuntini et al., 2014;Kotagal et al., 2013). Moreover, diabetes may also play a role in progression of PD (Ou et al., 2021), PD patients with diabetes usually present more severe motor symptoms, worse dopaminergic medicine responsiveness (Mollenhauer et al., 2019;Sandyk, 1993), and worse cognitive impairment (Pagano et al., 2018). Therefore, PD may present systematic glucose metabolism impairment, while hyperglycemia and diabetes, which causes systematic glucose metabolism impairment may also increase the risk of PD.
Regionally, glucose metabolism abnormality has been reported widespread in brain in PD patients (Dunn et al., 2014;Eggers et al., 2009;Meyer and Hellwig, 2014). PD patients present characteristic glucose hypometabolism in posterior temporoparietal, occipital, and sometimes frontal areas, accompanied with glucose hypermetabolism in putamen, sensorimotor cortex and cerebellum (Meyer and Hellwig, 2014).
Brain neurons exhibit impaired glucose metabolism, even at early stages of PD (Dunn et al., 2014). Patients with early-stage PD have glucose hypermetabolism in bilateral pallidum, SN, unilateral caudate, shell nuclei (Eggers et al., 2009) and hypometabolism in cortex (Borghammer et al., 2010). In advanced PD patients, reduction in glucose uptake were found in the parietal, frontal, and temporal cortex, as well as in the caudate nucleus regardless of discontinuation of drugs or treatment with levodopa (Berding et al., 2001). Glucose hypermetabolism in prefrontal regions is only identified in patients with less progressed PD, but not in patients with more progressed PD, implying frontal glucose metabolism dysfunction exacerbates with PD progression (Szturm et al., 2021).
PD patients with different subtypes present glucose metabolism impairment in different brain regions. Greater glucose hypometabolism accompanied by worse dopamine uptake was found in ventral striatum in akinetic-rigid PD patients compared with tremor-dominant PD patients (Eggers et al., 2014).
Brain glucose metabolism impairment in PD patients was also related to treatment. Overall brain glucose consumption decreases one hour after levodopa administration, while such reduction disappears after withdrawal of levodopa (Berding et al., 2001). Moreover, glucose uptake in ventral/orbital frontal cortex and thalamus is significantly reduced after levodopa treatment (Berding et al., 2001). Deep brain stimulation on bilateral subthalamus nuclei (STN) could increase glucose metabolism in pallidum, superior brainstem, dorsolateral prefrontal cortex, and posterior parieto-occipital cortex, while decrease glucose metabolism in orbitofrontal cortex and parahippocampus gyrus (Li et al., 2006).
Glucose metabolism abnormality may also reflect the cognition impairment in PD patients. In early-stage of PD, glucose hypometabolism is more severe in temporoparietal region in patients with mild cognitive impairment compared with patients without cognitive decline (Firbank et al., 2017). In patients with PD dementia (PDD), the pattern of glucose hypometabolism is similar to that in patients with Alzheimer's disease (AD), which presents in the posterior cingulate cortex, lateral parietal, lateral temporal and lateral frontal binding areas, while patients with PDD presents more involvement of visual cortex (Vander Borght et al., 1997), and non-demented PD patients exhibit widespread cortical glucose hypometabolism without selective temporoparietal defects (Peppard et al., 1992). Interestingly, patients with PDD and Lewy body dementia (DLBD) also present similar patterns of glucose hypometabolism in bilateral inferior frontal, medial frontal and right parietal lobes (Yong et al., 2007). However, the pattern of glucose metabolism impairment in PD patients has certain characteristics, such as hypermetabolism in anterior cingulate and lateral temporal cortex, by which PD and DLBD can be distinguished (Grassetto et al., 2014;Yong et al., 2007).
However, whether these regional glucose uptake impairment in brain of PD patients reflects disease-related brain activity changes, or just reflect a phenomenon of PD pathophysiology is unclear, and needs further investigation.

Neuronal energy supply impairment in PD
Neuronal energy deficit, namely decreased ATP level, has been observed in PD (Saxena, 2012). Damage in any process of energy supply, including glycolysis dysfunction, TCA dysfunction, and oxidative phosphorylation, decreases energy production and results in neuronal dysfunction and degeneration, and thus, participates in the pathogenesis of PD (Saxena, 2012).
PINK1 mutation inhibits phosphorylating serine-250 of complex I NadufA10, which is essential for CoQ reduction, and impairs complex III, eventually contributes to mitochondrial depolarization and ROS production (Amo et al., 2014;Morais et al., 2014;Morais et al., 2009). Such mutation could induce mitochondrial ROS-mediated hypoxia-inducible factor 1 (HIF-1) stabilization, and subsequently, increases the level of pyruvate dehydrogenase kinase 1, which inhibits pyruvate dehydrogenase and reduces flux of TCA, eventually impairs energy supply (Kim et al., 2006;Requejo-Aguilar et al., 2014). However, HIF-1 could also increase the level of hexokinase-2 (HK-2), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and glucose transporter 1, which stimulates glycolysis and inhibits PPP in neurons, possibly compensating shortage of ATP and attenuating oxidative stress (Requejo-Aguilar et al., 2014) (Fig. 1). Therefore, energy insufficient is present in PD, but there are also some other mechanisms to compensate this kind of shortage, more studies are needed to explore these compensate mechanisms, and provide more directions for treatment.
GAPDH is an enzyme in glycolysis which metabolite glyceraldehyde triphosphate to 1,3-diphosphoglycerate for glucose metabolism and energy production (El Kadmiri et al., 2014). In PD patients, GAPDH and α-synuclein colocalize in Lewy Bodies (Tatton, 2000), suggesting that GAPDH could contribute to pathological mechanism of PD. Both monomeric and oligomeric forms of α-synuclein could bind with GPADH when GAPDH is partially oxidized, and result in subsequent inactivation and aggregation of GAPDH, eventually inhibit glycolysis (Barinova et al., 2018;Melnikova et al., 2020). Moreover, glycated α-synuclein could enhance its binding on GAPDH (Semenyuk et al., 2019). Interesting, aggregated GAPDH promote aggregation of α-synuclein, but prevent its amyloid transformation and prevent its toxicity on dopaminergic neurons (Ávila et al., 2014; Barinova et al., 2018;Torres-Bugeau et al., 2012), and glycated GAPDH are uncapable of binding with α-synuclein (Sofronova et al., 2021). Physiologically, this negative feedback may be protective for dopaminergic neurons. However, this feedback could be disrupted by high glucose and ultimately leads to dopaminergic neuron damage and understanding the mechanisms may provide new insights into the role of glucose metabolism in PD.
These findings suggest a role of glycolysis impairment and related energy supply deficiency in PD, but further studies are needed.

Oxidative phosphorylation
The fact that MPTP and rotenone selectively blocking respiratory chain complex I and inducing dopaminergic neuron degeneration and parkinsonian symptoms indicates a role of respiratory chain defect in the pathogenesis of PD (Gerlach et al., 1991).
In idiopathic PD patients, respiratory chain complex I is damaged in SN (Keeney et al., 2006;Mann et al., 1994), the activity of complex I and IV are decreased in peripheral blood (Müftüoglu et al., 2004), while accumulated α-synuclein inhibits complex I activity, increases free radical production and mitochondrial autophagy in dopaminergic neurons (Chinta et al., 2010;Devi et al., 2008).
shortage and oxidative stress in early-stage of PD, and provide a treatment method for PD. These evidences suggest damaged oxidative phosphorylation in both sporadic and genetic PD patients, but further investigation is needed.

PPP
Decreased level of G6PD, a rate-limiting enzyme of PPP, has been identified in putamen of patients with early-stage PD, suggesting suppressed PPP and lowered anti-oxidative stress capacity (Dunn et al., 2014). Moreover, PPP is associated with lipid metabolism, which was reported to play an important role in Parkinson disease (Alecu and Bennett, 2019). Metabolic flow analysis found that, under high glucose stimulation, 13 C-labeled glucose was more likely to enter PPP pathway (Haythorne et al., 2019), leading to excessive NADPH production (Tu et al., 2019). NADPH is a key mediator for the synthesis of lipids such as unsaturated fatty acid catalyzed by SCD1 (Stanton, 2012). Moreover, in PPP pathway, ribulose 5-phosphate is catalyzed by G6PD and isomerized to xylose 5-phosphate (X-5-P), and X-5-P could activate the transcription factor CHREBP, which in turn elevates SCD1 expression (Diaz-Moralli et al., 2012;Ran et al., 2018) that further promote the synthesis of unsaturated fat acid. Interestingly, excessive unsaturated fatty acid in dopaminergic neurons can promote α-synuclein inclusion body formation, which is the pathological characteristic of PD (Fecchio et al., 2018).

Insulin resistance
Insulin resistance is considered as the core link between diabetes and PD (Cheong et al., 2020). Up to 58.4% PD patients have insulin resistance (Hogg et al., 2018), which is related to more severe phenotype (Cereda et al., 2012;Ou et al., 2021), more rapid progression, and higher risk of PDD (Bosco et al., 2012).
In patients with type 2 diabetes who presenting insulin resistance, the level of insulin increases in brain according to its increase in blood, while chronic hyperinsulinemia downregulates expression of insulin receptor (There are lots of insulin receptors in dopaminergic neurons of ventral tegmental area of brainstem (Unger et al., 1991), SN (Athauda and Foltynie, 2016;Unger et al., 1991), and basal ganglia (Athauda and Foltynie, 2016)) in blood-brain barrier (BBB), suppressing insulin transport to brain, eventually inducing dopaminergic neurons apoptosis (Cardoso et al., 2009;Meara et al., 1999;Santiago and Potashkin, 2013) (Fig. 2). Similarly, deletion of insulin receptor mRNA in the SNc and increased insulin resistance have been identified in PD patients (Moroo et al., 1994;Takahashi et al., 1996).
Additionally, elevated insulin-like growth factors-1 (IGF-1) level in serum and cerebrospinal fluid induced by long-term microglia activation has been identified in newly-diagnosed PD patients (Godau et al., 2011;Mashayekhi et al., 2010), reflecting degenerative changes in PD.

Oxidative stress
SNc dopaminergic neurons are particularly vulnerable to oxidative stress, placing these neurons at risk for degeneration, especially when glucose metabolism is impaired.
However, more direct evidences are necessary to further illustrate the role of glycated modification in the mechanism of PD.

Dysfunction of BBB
Previous study has proven compromised integrity of BBB in PD patients by identification of erythrocytes and serum protein leakage from vessel, hemosiderin deposit around capillaries in striatum, capillary endothelial degeneration, dysfunction of tight junction proteins, and basement membrane changes in postmortem examination (Gray and Woulfe, 2015;Pienaar et al., 2015), and gadolinium leak and cerebral microbleeds by magnetic resonance imaging (Al-Bachari, 2016;Sweeney et al., 2018;Takemoto et al., 2021). The cerebrospinal fluid/plasma albumin ratio, maker of BBB dysfunction, is increased in PD patients compared with healthy population, and the ratio is related to coexisting diabetes (Janelidze et al., 2017). Therefore, glucose metabolism impairment, especially in diabetes, may contribute to the pathogenesis of PD.
Dual GLP-1/GIP receptor agonists also improve clinical symptoms of PD potentially via mechanisms similar to GLP-1 receptor agonists (Cao et al., 2016;Feng et al., 2018;Jalewa et al., 2017;Yuan et al., 2017). However, clinic efficacy of dual GLP-1/GIP receptor agonists is better than that of single GLP-1 receptor agonists, possibly because of its stronger effect in protecting dopaminergic neurons from degeneration, reducing inflammation and increasing neuroprotective growth factor Yuan et al., 2017).

Others
It is reported that terazosin could reduce incidence of PD, delay progression of PD, and decrease related complications by increasing phosphoglycerate kinase 1 activity, stimulating glycolysis, promoting ATP production, and preventing dopaminergic neuron loss (Cai et al., 2019;Tang, 2020). Antihistamine drug meclizine could also increase glycolysis to offset neurotoxicity of 6-OHDA in dopaminergic neurons (Hong et al., 2016).

Conclusion
Impairments in systematic and regional glucose metabolism exist in patients with PD at every stage of the disease course, which affect each physiological process of glucose metabolism including glucose uptake, glycolysis, TCA, oxidative phosphorylation, and PPP. These impairments may be attributed to various mechanisms, such as insulin resistance, hyperglycemia-induced damage, oxidative stress, abnormal glycated modification, and BBB dysfunction, resulting in energy supply insufficiency, neurotransmitter dysregulation, aggregation and phosphorylation of α-synuclein, and dopaminergic neuron loss. Though the role of glucose metabolism in PD has been extensively studied, the abnormities of glucose metabolism in PD are a cause or a result is still unclear, which needs further study in future. The mechanisms linking PD and abnormal glucose metabolism are complex, some molecular disturbance including hyperglycemia, oxidative stress and glycation interact with each other, and the exact regulatory mechanism is poorly understood. Figuring out the mechanism by which feedback between α-synuclein and GAPDH was disrupted in high glucose condition may provide new insights into the role of glucose metabolism impairment in Parkinson's disease. Furthermore, whether and how PPP pathway modulate glucose flux and lipid synthesis in PD are also worthy future investigation, and may provide new target for treatment of PD. There are few treatments for glucose metabolism impairment in PD patients. Further investigation may provide more insights into glucose metabolism impairment of PD, and may assist the research and clinical practice in relieving PD.

Declaration of Competing Interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Xi Liu reports financial support was provided by National Natural Science Foundation of China. Xi Liu reports financial support was provided by Natural Science Foundation Project of Chongqing. Xi Liu reports financial support was provided by Kuanren Talent Program of The Second Affiliated Hospital of Chongqing Medical University.

Data availability
No data was used for the research described in the article.