Metabolic Dysregulation Contributes to the Progression of Alzheimer’s Disease

1Alzheimer’s disease (AD) is an age-related neurodegenerative disease characterized by a progressive loss of neuronal structure and function. It is the most common type of dementia worldwide. The condition involves a progressive deterioration in memory, cognition, and mobility. Its pathophysiology is extremely complex and heterogeneous, entailing accumulation of senile plaques caused by abnormal amyloid β (Aβ) metabolism, and neurofibrillary tangles caused by tau hyperphosphorylation (i.e., the formation of p-tau). Furthermore, the cerebrovascular system is seriously damaged, including the disturbance of the blood-brain barrier (BBB) and cerebral amyloid angiopathy. Increased levels of reactive oxygen species (ROS) induce the transcription of pro-inflammatory genes and the release of cytokines (e.g., interleukin-1β [IL-1β], IL-6, and tumor necrosis factor-alpha [TNF-α]) and chemokines that cause neuroinflammation. In addition, reactive microglia and astrocytes and other pathological events also contribute to the dysfunction and deprivation of synapses and, ultimately, neuronal death. Aging is a major risk factor for AD. Some similarities and differences occur in glucose metabolism-related proteins have been observed in the brain during normal aging and AD. The apolipoprotein E (ApoE) gene is responsible for synaptic repair and neuronal structure maintenance and is a major risk factor for the sporadic form of AD. Of the three major isoforms (i.e., ApoE2, ApoE3, and ApoE4), people with the ApoE4 allele are at higher risk of developing AD than the others.

2The brain consumes the greatest amount of energy of all the organs in the body. Because neurons require large amounts of energy to maintain their normal activities, a metabolic decline in the aging brain contributes to cognitive impairment. There is an age-related decrease in glucose utilization in most human brains. In addition, reduced O2 uptake has been observed in the aging rodent brain. The pathological metabolic alterations in aging (e.g., cerebral glucose hypometabolism) are early and consistent events in the progression of AD. Glucose, the main transportation form of carbohydrate in our blood, is also the crucial and primary energy substrate for the brain under physiological conditions. However, alternative substrates, such as glycogen, ketone bodies, and amino acids, are also critical. Energy hypometabolism, particularly a decline in glucose metabolism, is one of the earliest and most common anomalies observed in patients with AD. Indeed, insulin and insulin growth factor-1 (IGF-1) signaling help to maintain and control metabolism and cognition in the central nervous system (CNS), and insulin resistance is one of the main risk factors for AD. The main intracellular energy metabolism pathways occurring in our brains are complicated and include anaerobic glycolysis and the pentose phosphate pathway (PPP) in the cytoplasm, as well as oxidative phosphorylation (OXPHOS) in mitochondria and the tricarboxylic acid (TCA) cycle (also known as the citric acid cycle and Krebs cycle).

3Metabolic processes are regulated by a series of key enzymes. Indeed, a growing body of evidence suggests the presence of organic impairment of mitochondria and damage to related metabolic enzymes. In addition, oxygen and glucose metabolic rates are drastically changed in many neurodegenerative diseases, including AD due to marked alterations in the glycolytic pathway and TCA cycle. To add insult to injury, metabolism dysregulation is related to inflammatory responses, particularly in microglia. reported that Aβ could directly activate microglia to produce inflammatory factors by shifting their metabolism from OXPHOS to aerobic glycolysis. Thus, understanding the relationship between dysfunctional metabolism and AD could provide new insights into the pathogenesis of AD to support the development of new therapies. In this review, we discuss normal metabolic processes, aging, AD-related dysregulation, and relevant treatment strategies.

4As the major energy source for the brain, glucose is metabolized to ATP, an unstable high-energy compound. Glucose metabolism in the brain involves several stages. First, a signal is received by the brain to trigger glucose uptake (e.g., insulin signaling). Next, the physiological process of glucose uptake occurs. This process is dependent on glucose transporters (GLUTs) spread throughout the brain that allow glucose to cross the BBB and reach the neurocytes (e.g., astrocytes). There are several different GLUT subtypes in the brain. GLUT1 (55 kDa) transfers glucose from the blood into the extracellular space of the brain through the BBB endothelium, while GLUT1 (45 kDa) and GLUT3 take up glucose into astrocytes and neurons, respectively. The insulin-sensitive GLUT4 is found in discrete subsets of neurons. GLUT8 is located in the intracellular compartment of hippocampal and cerebellar neurons regulated by hormone, while its exact location and function are still undefined. After uptake, glucose is metabolized through the glycolysis pathway to pyruvate, generating ATP. Finally, pyruvate is converted to acetyl coenzyme A (acetyl-CoA) via the TCA cycle, which eventually forms an electro-gradient to drive the rotation of the V-ATPase machinery, and energy is transferred from the electro-potential to the bound ATP. The hydrogen generated from this oxidation is transformed into water and ATP through complexes I, II, III, and IV of the electron transport chain (ETC). This OXPHOS reaction occurs on the inner mitochondrial membrane (IMM). Concurrently, the carbon dioxide produced by decarboxylation is transported through the blood to the respiratory system and expelled.

5Blood glucose accesses the brain via the GLUTs with the help of insulin. Decreased expression of insulin-sensitive GLUTs is strongly associated with a decline in glucose uptake. Insulin and the insulin receptor (IR) are vital factors in regulating glucose utilization and energy homeostasis between the CNS and peripheral circulatory system. Insulin signaling is reciprocally linked to the mammalian target of rapamycin (mTOR) pathway via the phosphoinositide-3-kinase (PI3K)/Akt axis. As an intracellular energy sensor, mTOR is activated by growth factors, including amino acids and high cellular energy status. Thus, the mTOR signaling pathway plays a major role in regulating cell growth and lipid and glucose metabolism. Additionally, both inhibited insulin signaling and altered protein homeostasis in early AD can lead to aberrant mTOR activation. The coupling of PI3K/Akt to IGF-1 and IR can be eliminated through serine phosphorylation of insulin receptor substrate-1 (IRS-1), mediated by mTOR, and IRS-1 inactivation and degradation, which is a prominent trigger of brain insulin resistance (BIR). Insulin also downregulates mTOR due to continuous activation of IRS-1 by mTOR. The insulin signaling cascade is also regulated by the unique Ser/Thr/Tyr kinase biliverdin reductase-A (BVR-A). Oxidative stress-induced impairment of BVR-A kinase activity is an early event. Moreover, glucose starvation (hypoglycemia) reduces the intracellular ATP/AMP ratio, activating AMP-activated protein kinase (AMPK). As the main sensor of intracellular fuel status activated by energy stress, AMPK participates in the induction of several genes responsible for the growth, maintenance, and repair of neuronal cells and synapses. In addition, AMPK can regulate the plasticity of the hippocampal synapse, the cornerstone of learning and memory. To sum up, GLUTs are regulated by multiple pathways, and a balance among these pathways is crucial to maintaining the stability of energy metabolism.

6The brain is mainly composed of terminally differentiated neurons. During aging, neurons with relatively low regenerative ability are unable to adapt to alterations in the basal metabolic rate, and the energy-driven state is decreased or degenerated, which may contribute to various neurodegenerative diseases. 18F-fluorodeoxyglucose (FDG) and Pittsburgh Compound B (PIB) positron emission tomography (PET) (PIB-PET and FDG-PET, respectively) are suitable for detecting brain glucose uptake. Micro-FDG-PET scans show that cerebral glucose uptake in normal-aging humans and aged rats is declined compared to the youth. Moreover, the role of reduced cerebral glucose uptake in age-related cognitive impairment has been verified in clinical studies. The expression of neuronal GLUTs, such as GLUT3 and GLUT4, significantly decreases with aging, while GLUT1 (55 KDa) expression in vascular endothelial cells decreases only slightly. BIR, resulting in the partial inactivation of insulin signaling and the impairment of PI3K/Akt and several downstream pathways, has been linked to aging. Ultimately, these changes contribute to a variety of characteristics of normal aging in the body.

7Compared to normal aging, a strong reduction in glucose consumption can be observed in AD. Alterations in glucose metabolism can affect the maintenance of neurotransmission and neuronal function and impact the ability to learn and memorize. Specific damage to cerebral glucose metabolism has been detected in the posterior cingulum cortex and temporoparietal cortices using FDG-PET. In contrast, glucose metabolism is relatively constant in areas of primary sensorimotor and visual cortices, basal ganglia, and the cerebellum in AD patients. Decreased glucose utilization may occur even before the clinical symptoms of AD based on a study involving mild cognitive impairment (MCI) subjects. Glucose uptake in the precuneus, an area of early Aβ deposition, is significantly decreased in individuals with the disease 10 years before the appearance of symptoms. Tau mutants regulate mitochondrial trafficking by altering the fragmentation of mitochondria in neuronal cells. Two different sites in human islet amyloid polypeptide sequence are sensitive to BACE1-mediated APP cleavage. Aβ may be relevant to the interaction of IR and the GLUTs. The disturbance of neurogenic glucose metabolism caused by impaired insulin signaling results in AD characteristics that parallel the pathophysiology of non-nervous tissues in type 2 diabetes mellitus. The density of the neuronal GLUT3 transporter is associated with local cerebral glucose utilization. A reduction in GLUT1 and GLUT3 in AD patient brains is correlated with a decline in brain glucose consumption and cognition impairment. The decline may result from the abnormal hyperphosphorylation of tau and decreased hypoxia-inducible factor-1 α (HIF-1α) levels, which contribute to the transcriptional activation of GLUT. GLUT4 is reduced in male 3xTG-AD mice; however, cerebral glucose uptake does not coincide with GLUT4 expression in female mice. In addition, astrocyte activation may contribute to the GLUT2 upregulation observed in postmortem brain tissue from AD patients.

8The insulin and IGF-1 signaling (IIS) pathway has considerable effects on metabolism regulation and cognitive function. IIS binds to tyrosine kinase receptors and IGF-1 receptor (IGF-1R), which are widely distributed in the hippocampus and cerebral cortex in AD. The insulin signal is inhibited in the AD brain, which is closely connected to inefficiency in glucose metabolism. The impairment of insulin signaling also contributes to abnormalities in mitochondrial structure and function. Moreover, significant gene expression alterations observed in the AD brain are related to the generation and transmission of insulin signals. For example, the post-synaptic β-aminobutyric acid (GABA) - a receptor accumulates on the cell surface rapidly when the insulin pathway is activated; however, the genes encoding GABA receptors are markedly reduced in AD brain tissue. Insulin competitively inhibits the insulin-degrading enzyme (IDE) to degrade Aβ and elevates extracellular Aβ levels by promoting its secretion. IDE is involved in insulin function/resistance and metabolism-related processes and plays an important role in the degradation of Aβ monomers. However, its expression is reduced in AD patients. Energy-transducing pathways always occur in the mitochondria, and the initial activation of IIS requires mitochondria to produce low-level H2O2, which reflects the energy state of mitochondria and is involved in the regulation of redox. The insulin receptor (IR) and insulin receptor substrate (IRS) renders them susceptible to oxidation (and activation) by H2O2. Excess H2O2 can give rise to BIR, which is a risk factor for diabetes mellitus and AD. Thus, the regulation of redox-sensitive signals in the mitochondria should also be considered. BIR compromises the intracellular translocation of low-density lipoprotein receptor-related protein 1 (LRP1, a receptor for ApoE) to the plasma membrane of hepatocytes, potentially hindering the hepatic elimination of circulating Aβ. In a 3xTg-AD mouse model, oxidative stress first causes consistent activation of IRS-1 and then activates negative feedback mechanisms (e.g., mTOR) to disable IRS-1 hyperactivity and cause BIR. Overall, BIR is regulated by aberrant insulin signaling and may contribute to disordered glucose metabolism, oxidative stress, BBB dysfunction, and energy supply insufficiency, which are pathological features of diabetes mellitus that can further affect Aβ generation and clearance. Thus, insulin resistance that induces glucose hypometabolism might be the main cause of energy deficiency in AD brains, which resemble, but is distinct from, the manifestation of diabetes. Thus, it is hypothesized that AD is a neuroendocrine disorder that may be identified as “Type 3 diabetes,” reflecting a new mechanism of neurodegeneration.

9As mentioned above, insulin mediates the phosphorylation of downstream mTOR by activating PI3K/Akt signal pathways in our brains. As one of the cellular degradation pathways of misfolded and unfolded proteins in neurodegeneration, autophagy is regulated by mTOR, which affects ATP levels and biosynthetic pathways. Because the maturation of autolysosomes and their retrograde transport are impeded in AD, a massive accumulation of defective autophagosomes can lead to autophagy intermediates (autophagic and lysosomal vesicles) within large swellings along dystrophic and degenerating neurites. Defective autophagy leads to the overproduction, aggregation, and diminished clearance of Aβ and p-tau, oxidative damage, and mitochondrial dysfunction, which, in turn, contribute to the impairment of the metabolic pathways controlled by insulin and mTOR. Moreover, mTOR signaling hyperactivity inhibits the induction of the autophagy clearance system and increases the accumulation of Aβ in AD. Thus, defective autophagy causes various effects in AD.

10Triggering receptor expressed on myeloid cells 2 (TREM2) is one of the immune receptors expressed in the plasma membrane of microglia in the brain that can recognize phospholipids, lipoproteins, and apoptotic cells. Studies have shown that TREM2 defects can lead to impaired mTOR activation and the enhancement of AMPK activation in microglia in AD patients and an AD mouse model. This effect can result in hyperactive autophagy and microglial energy impairment, which can be compensated by energy repletion. Thus, defective autophagy caused by abnormal mTOR regulation plays an important role in the development of AD. Thus, TREM2 represents a risk factor for AD; however, the mechanism is unknown.

11Moreover, decreased metabolic levels are more likely to be detected in the brains of ApoE4 carriers. ApoE4 is a crucial risk factor for late-onset AD. There are key structural differences between the three isoforms of ApoE. ApoE2 has two cysteine residues (Cys, residues 112 and 158), while ApoE3 has a positively charged arginine residue at position 158, and ApoE4 has positively charged arginine residues at both positions 112 and 158. The differences in the protein structures of the three ApoE isoforms affect their interactions with other proteins and peptides. Accumulating evidence has shown that ApoE, particularly ApoE4, binds to residues 12–28 of Aβ, and this binding promotes Aβ accumulation and strengthens Aβ pathology, especially oxidative stress. Oxidative stress-related modification analysis of plasma and cerebrospinal fluid (CSF) proteins demonstrated that ApoE oxidation could affect the antioxidant activity of thiol, which allows the formation of lipoprotein particles caused by excessive oxidative damage. Under high levels of oxidative stress, lipid peroxidation produces the highly reactive and neurotoxic molecule, 4-hydroxynonenal (HNE), which covalently binds to Cys residues. Covalently modified Cys residues significantly alter the structure and function of modified proteins. HNE binds to Cys residues in ApoE2 and ApoE3, protecting other proteins from HNE damage. However, ApoE4 lacks Cys residues. Therefore, it cannot scavenge HNE, permitting this neurotoxic molecule to modify neuronal proteins and induce cell death. ApoE2 appears to have neuroprotective effects in the AD patient brain and exhibits the most robust metabolic profile for glucose uptake and glycolysis. In contrast, ApoE4 causes the most detrimental effects on aging and AD brains. There is reduced cerebral glucose metabolism in cognitively normal individuals carrying ApoE4. The brains of ApoE4 mice have lower GLUT3 mRNA levels compared to ApoE3 mice, which contributes to inadequate energy supplies in the ApoE4 brain. Understanding the metabolic process and its effects on the early or progressive alterations in AD may provide new treatment strategies for this irreversible neurodegenerative disease.

12Several strategies are available for the development of therapeutics to prevent or slow down the progression of AD. Given that BIR is the main risk factor for AD, showed that a single dose of intranasal insulin can significantly improve the memory of patients with AD or mild cognitive impairment. Based on PET findings, intranasal insulin increases 18F-fluorodeoxyglucose uptake in the precuneus, frontal, cuneus, and parietotemporal regions of the brain. A Phase 2/3 multi-site clinical trial with intranasal insulin conducted between 2014 and 2018 showed no significant adverse reactions, and treatment improved cognitive function. Therefore, the specific regimen for intranasal insulin injection and its feasibility require further study. Moreover, one of the common drugs used in diabetes (peroxisome proliferator-activated receptors [PPAR]-γ agonists) increases metabolic efficiency by enhancing insulin sensitivity to reduce Aβ levels. Because the mTOR signaling pathway participates in multiple processes that regulate neuronal functions, rapamycin can be used to improve learning and memory and reduce Aβ and tau pathology. In addition, oral administration of rapamycin in a 3xTg-AD mouse model relieves memory symptoms. Moreover, rapamycin analogs (Rapalogs), which have been approved by the FDA when used concurrently with metformin, are recommended to pharmacologically address the impaired glucose metabolism.

13Evidence suggests that short- and long-acting intranasal insulin therapy can improve memory in AD patients in an ApoE4-dependent manner. In response to reduced glucose metabolism, mammalian cells promote the synthesis and utilization of ketone bodies, circulating energy sources for tissues in times of fasting or prolonged exercise. Targeted replacement of ApoE in mice with the human ApoE genes demonstrated that brains from ApoE2 and ApoE4 mice have stronger absorption and metabolism of ketone bodies than brains of ApoE3 mice. Moreover, ApoE–/– mice (an AD model) exhibit circadian rhythm disturbances and increased tau deposition. These phenomena are related to energy shortage and degeneration of the suprachiasmatic nucleus, which can be alleviated by supplementing with ketone bodies in the absence of glucose. Therefore, ApoE and ketone bodies may represent new therapeutic targets for improving brain energy metabolism in patients with AD. The specific mechanisms associated with ApoE and ketone bodies and their related treatment pathways need further investigation.

14Earlier, we defined the important proteins needed for glucose to flow from the blood into the brain and subsequently regulate metabolic changes under normal conditions. We then summarized a series of changes in these proteins and other risk factors (e.g., ApoE4) in aging and AD. The insulin signaling pathway is closely linked to metabolic change, and the manifestation of decreased brain energy metabolism caused by insulin resistance is similar to diabetes. Thus, the “Type 3 diabetes” hypothesis is well supported. Furthermore, PET is a method used clinically to detect glucose intake and a reference for AD diagnosis. Given these metabolic changes, we described several potential clinical treatments; however, an effective treatment still requires further study.

15Glycolysis, as the first step of glycometabolism and one of the main energy sources, plays an essential role in CNS metabolism. GLUT1 and GLUT3, mainly expressed on astrocytes and neurons, mediate the entry of glucose into these cells. Glucose can then be converted into glucose-6-phosphate (G6P) and then into fructose-6-phosphate (F6P) by hexokinase. F6P, which continues to participate in the subsequent steps of glycolysis, is converted into pyruvate. On the other hand, G6P not only participates in both glycogen synthesis and the PPP, but also non-competitively inhibits hexokinase. Phosphofructokinase (PFK), which takes part in the second step of glycolysis, catalyzes F6P into fructose 1,6-bisphosphate (F1,6BP). Fructose 2,6-bisphosphate (F2,6BP), which upregulates PFK-1, is generated via a reaction catalyzed by fructose 2,6-bisphosphatase isoform 3 (PFKFB). Astrocytes are reported to have higher glycolysis activity due to higher PFK1 activity. In contrast, neurons exhibit lower glycolytic capacity due to lower levels of PFK1 and 6-phosphofructo-2-kinase and PFKFB. Thus, astrocytes have a significant effect on glycolysis in the CNS. In addition, the astrocytic levels of the abovementioned enzymes are elevated when rat hippocampal astrocytes are co-cultured with neurons, which suggests that neurons can affect astrocyte glycolysis.

16Microglia maintain normal brain function by providing trophic support, respond to changes in CNS metabolism, and carry out classic immune cell functions that promote phagocyte clearance. An in vivo study reported that the immune response of microglia is based on glycolysis metabolism. As first reported in 2002, the inflammasome, a multi-protein complex assembled by intracytoplasmic pattern recognition receptors, plays an important role in the natural immune system. NLRP3 is the most studied component of the inflammasome. While the function of the NLRP3 inflammasome in astrocytes is controversial, it is known to be expressed and activated in microglia. Evidence shows that pentabromophenol (PBP) and tetrabromobisphenol A (TBBPA) improves the metabolic rate of glycolysis in mouse microglia and promotes the activation of NLRP3 inflammasome. Hexokinase is one of the key glycolytic enzymes. reported that N-acetylglucosamine activates the inflammasome, particularly the NLRP3 inflammasome, through hexokinase inhibition by promoting its disassociation from the outer mitochondrial membrane (OMM).

17Studies in mouse models of different ages found that the levels of glycolysis products (e.g., G6P and F1,6BP) decreased in adult mice compared to young mice. However, recent studies demonstrated that glycolysis increased with a decline in the resting cerebral blood flow of the aged brain. It also increased in naturally aging astrocytes. Furthermore, glycolysis dysfunction can lead to age-related neurodegeneration. Therefore, glycolysis has a complex relationship with growth and aging in the brain.

18Considerable research has examined the associations between AD and the enzymes intimately linked to glycolysis (i.e., hexokinase, glyceraldehyde 3-phosphate dehydrogenase [GAPDH], and pyruvate kinase [PK]). reported that Wnt signaling promotes glucose metabolism by increasing the expression of hexokinase and PFK, which causes neuroprotective effects. However, decreased hexokinase and PFK expression and dysregulated Wnt signaling are observed in AD. The accumulation of G6P can reduce hexokinase activity by competitively inhibiting ATP binding to the active site of the enzyme. In addition, an unbiased metabolomics approach demonstrated G6P accumulation in humans and mice with AD, which restrains glycolysis. Hexokinase binds to the OMM via the voltage-dependent anion channel (VDAC), which controls the mitochondrial permeability transition pore (MPTP). When hexokinase associates with VDAC, the MPTP tends to be closed. In the postmortem brain tissue of AD mice and patients, hexokinase levels were decreased while VDAC1 levels were elevated. Furthermore, the interaction of VDAC1 with hexokinase I can generate outer membrane potential in brain mitochondria. Outer membrane potential metabolic-dependent homeostasis can prevent the mitochondrial permeability transition, which leads to Ca2+ activation and neuronal cell death. In addition, this may be involved in resistance to neuronal death and neurodegenerative disorders such as AD. Glycogen synthase kinase 3 (GSK3), a serine/threonine protein kinase, disassociates VDAC1 from hexokinase. The activation of GSK3 promotes apoptosis in neuroblastoma cells by reducing the level of cyclin D1, which inhibits the extrinsic apoptotic signaling pathway mediated by death receptor. Tau phosphorylation, one of the characteristics of AD, is regulated by GSK3. Notably, GSK3 inhibition attenuates the symptoms of mild cognitive impairment and restrains oxidative stress in AD. Activation of the NLRP3 inflammasome resulting from mitochondrial DNA synthesis in macrophages results in damage to macrophage mitochondria. The NLRP3 inflammasome-driven proinflammatory cascade in microglia is augmented by impaired mitochondrial function. Furthermore, NLRP3–/– or caspase-1–/– mice with AD-related mutations largely avoid spatial memory loss and other sequelae associated with AD, suggesting a key role for the NLRP3 inflammasome in the pathogenesis of AD. Overall, the hexokinase dysfunction observed in AD brain samples is mediated by the accumulation of G6P that dissociates hexokinase from mitochondria. This process activates the NLRP3 inflammasome, indicating that G6P accumulation may contribute to neuroinflammation in AD. Therefore, the effect of hexokinase activity on glycolysis in the CNS and mitochondrial function is a starting point to study the metabolic mechanism of AD.

19GAPDH participates in the sixth step of glycolysis, catalyzing glyceraldehyde 3-phosphate to 1,3-bisphosphoglycerate and increases the NADH:NAD+ ratio. 2-Deoxyglucose (2DG), which inhibits glycolytic processes, prevents neurodegeneration by eliminating microglia (microglia can damage neurons in an inflammatory situation). In addition, it reduces the impact of Aβ on neuron cells. Moreover, an increased NADH concentration reverses the abovementioned effects of 2DG. Hou et al. found that a decreased NADH:NAD+ ratio is a possible way to reduce AD-associated pathology. Recent research also reported many other functions of GAPDH beyond glycolysis, including DNA repair, control of the activity of kinase and phosphotransferase, and an integral association with membrane ion pumps participating in Ca2+ release. Hyperphosphorylated tau protein leads to microtubule depolymerization, resulting in neuron dysfunction. showed that tau bound to denatured GAPDH but not the native molecule and that the aggregation of the non-native GAPDH in solution was ameliorated by tau. further reported that GAPDH aggregation induced by nitric oxide led to MPTP opening and cell death. An increase in GAPDH disulfide bonding in AD patients and transgenic AD mice compared with controls suggests a potential relationship between GAPDH disulfide bonding and protein aggregation. In addition, GAPDH promotes Aβ amyloidogenesis in vitro. A recent study found that the concentration of S-glutathionylated-GAPDH in the blood of patients with AD was significantly higher than in the control group, indicating that this indicator is related to the degree of neuronal apoptosis during the progression of AD. Thus, GAPDH and its substrates are associated with neuronal cell death in AD. Moreover, Aβ causes microglial inflammation and induces a shift in the metabolic pathway from OXPHOS to glycolysis in the 5XFAD mouse model. During this process, Aβ induces Akt phosphorylation to activate the mTOR-HIF-1α pathway. HIF-1α then increases GAPDH expression. Inhibition of this pathway decreases the levels of pro-inflammatory cytokines, including IL-1β and TNF-α. These data demonstrate the relationship between glycolysis and inflammation in microglia induced by Aβ.

20PK, a rate-limiting enzyme in glycolysis, has four isomers: M1, M2, L, and R. PKM2, which controls the levels of glycolytic intermediates as well as ATP, is linked to neurodegenerative diseases. As described by the Warburg effect, upregulation of the Wnt/β-catenin pathway can promote glycolysis, which is connected to PKM2. In AD, the Wnt/β-catenin pathway is downregulated (partially via inactivation of PKM2) and this results in oxidative stress and cell death. An increase in PK2 expression has been reported in AD transgenic mouse models. PKM2 promotes cell proliferation by binding to receptors of activated growth factors and then induces dimer formation through phosphorylation. In addition, a study that induced yeast pyruvate kinase Cdc19 to form amyloid-like aggregates in vitro found similar characteristics between Cdc19 and PKM2, its mammalian counterpart. Thus, the association between PK and AD might be worthy of investigation.

21The latest AD patient symptom management guidelines consider pimavanserin and citalopram to be the two most promising medications for AD. Pimavanserin selectively and inversely excites 5-hydroxytryptamine 2A (5-HT2A) receptor. Citalopram prevents the reuptake of 5-hydroxytryptamine (5-HT) by inhibiting 5-HT transporter, blocking reuptake on the presynaptic membrane. Thus, both pimavanserin and citalopram increase the concentration of 5-HT in the synaptic cleft. found that oxygen deficit and glucose deprivation inhibit the contractive response of umbilical arteries to 5-HT, which implies that aerobic glycolysis can increase sensitivity to 5-HT. Research into human pancreatic ductal adenocarcinoma also indicates that the levels of enzymes participating in glycolysis increase after 5-HT stimulation. Furthermore, conscious mice injected with 5-HT exhibit an increased brain glucose concentration. In addition, 5-HT facilitates glycolysis through PKM2 upregulation in breast cancer cells. Therefore, an increased concentration of 5-HT stimulates glycolysis, although the specific glycolysis-related mechanism in the CNS remains unclear.

22Besides the international guidelines for AD treatment, other recent guidelines, such as Chinese guidelines for the treatment of AD, propose cholinesterase inhibitors and excitatory amino acid receptor antagonist as well as certain traditional Chinese medicines for the treatment of AD. The relationship between cholinesterase and glycolysis in CNS has been studied in the last century. A study in 2013 indicated that carbachol (a cholinergic agonist) could increase fluxes in both glycolysis and OXPHOS in SH-SY5Y neuroblastoma cells. Glucose metabolism dysregulation can influence acetyl coenzyme A and indirectly retard the synthesis of acetylcholine. In addition, the activity of choline acetyltransferase (formed from choline and acetyl coenzyme A) is below normal in AD patients’ brains. In AD treatment, cholinesterase inhibitors, such as donepezil, may help patients to control the symptoms of the disease.

23Excitatory neurotransmission, particularly that of glutamate, as well as the receptors involved, play essential roles in synaptic plasticity via N-methyl-D-aspartic acid (NMDA) receptors (NMDARs). NMDAR stimulation facilitates glucose uptake and glycolysis in oligodendroglial cells. However, intense signal stimulation by glutamate would produce excitotoxicity, leading to CNS damage via excessive Ca2+ entry. A study of mouse astrocytes provided evidence that glutamate promotes glycolysis and damages mitochondrial aerobic capacity. Aβ has two forms, a soluble oligomeric form and an insoluble aggregate form. The former is primarily responsible for neurodegeneration and diminished synaptic function in AD, and it can interact with NMDARs, leading to mitochondrial Ca2+ overload and cell apoptosis. Memantine, as one of the uncompetitive NMDAR antagonists, has a strong voltage dependency. A study in a male Wistar rat model showed that memantine and lithium could reverse the decreased in IL-4 induced by oligomeric Aβ1–42, whereas lithium alone had no effect. Meta-analyses also indicate that memantine, particularly in conjunction with cholinesterase inhibitors, inhibits or slows the progress of AD symptoms. Therefore, NMDAR antagonism by memantine might prevent excessive glycolysis mediated by excitatory neurotransmission and the resultant excitotoxicity in AD.

24In short, aging and AD are closely related to abnormal glycolysis in the nervous system. Dysfunction of the key enzymes involved in glycolysis can affect the generation of Aβ and tau, the activation of the NLRP3 inflammasome, and even mitochondrial function. A variety of drugs for the treatment of AD, backed by the existing guidelines, can regulate glycolysis in the nervous system. Moreover, controlling the activity of enzymes and products that regulate glycolysis (e.g., hexokinase, GAPDH, and G6P) could significantly improve the nervous system and may represent a new direction for AD treatment.

25The TCA cycle is the central hub for energy metabolism, macromolecule synthesis, and redox balance. In normal aerobic organisms, most cellular glucose is converted into pyruvate through glycolysis. Pyruvate is subsequently oxidized by PDH to acetyl-CoA, which is fed into the TCA cycle. One acetyl-CoA molecule can generate six molecules of NADH and two molecules of FADH through this pathway. The coenzyme NADH contains many electrons, which are transferred to the ETC on the IMM. Finally, ATP and H2O are generated. This process is called OXPHOS and also generates ROS as a byproduct to maintain cellular homeostasis. However, excessive ROS production may contribute to oxidative stress, with mitochondria as the first target. The body utilizes antioxidants (e.g., superoxide dismutase [SOD] and glutathione [GSH] to offset the adverse effects of ROS, ). However, the brain is an organ with high oxygen consumption and low antioxidant defenses. Thus, it is vulnerable to oxidative stress. In addition, α-ketoglutarate (α-KG) is a TCA cycle intermediate that is transformed into glutamate and GABA by glutamate decarboxylase (GAD) and glutamate dehydrogenase. Reducing equivalents (i.e., NADH) generated from OXPHOS synthesize ATP in conjunction with glutamate and GABA, which helps maintain synaptic plasticity.

26In addition to antioxidants, the TCA cycle products themselves affect the mitochondrial redox balance. Nicotinamide nucleotide transhydrogenase (NNT), NADP+-dependent isocitrate dehydrogenase (ICDH) 2, and malic enzyme can generate NADPH using the electrons from NADH. All these proteins can affect redox activity in mitochondria. The redox state of mitochondria can regulate energy metabolism via the oxidization of several metabolic enzymes, including aconitase, α-ketoglutarate dehydrogenase (α-KGDH), malate dehydrogenase, and succinyl-CoA-3-oxoacid CoA transferase (SCOT), and complexes I, II, and V. It can also regulate S-glutathione glycosylation, a modification that reflects the redox state of mitochondria by reversibly forming mixed disulfide bonds between protein cysteine sulfhydryl and GSH. Therefore, NADPH plays a crucial role in a series of processes mediating glucose metabolism as part of the redox-energy metabolism axis.

27Proteomics analysis found that the dysregulation of TCA enzyme levels during aging includes the upregulation of malate dehydrogenase 1 (MDH1), fumarate dehydrogenase 1 (FH1), and subunits of NADH dehydrogenase, succinate dehydrogenase, and pyruvate dehydrogenase (PDH). Impaired TCA cycle metabolism is also associated with the downregulation of ICDH 1/2 and a subunit of succinyl-CoA ligase in the brains of aged mouse brains. Furthermore, a reduction in the metabolites from the TCA cycle has been measured in both aging and AD mouse models, while increases in the levels of the ceramides and sphingosine-1 phosphate (i.e., inflammatory metabolites) occurred in the aging group. The study also found that the levels of NADH and acetyl-CoA were positively correlated with age and the degree of AD, while glutamine and GABA concentrations were negatively correlated. A decline in glutamine and GABA levels is associated with impaired neurotransmitter circulation. Moreover, acetyl-CoA and NADH+ have allosteric inhibitory effects on PDH, which subsequently affects the TCA cycle. Many studies have demonstrated that in the aging brain, as mutations in mitochondrial DNA (mtDNA) increase, the expression of respiratory chain complexes I, III, and IV are suppressed. The oxidants (i.e., H2O2) are produced while NADPH is insufficient. In addition, the high H2O2 levels alter the redox environment in cells and can be partially released into the cytoplasm through VDAC. Redox homeostasis is not only one of the characteristics of aging but also of neurodegenerative diseases, such as AD. A clinical study on the quantitative detection of GSH in the occipital cortex showed that GSH levels in the elderly are significantly lower than those in the young. This age-dependent decline seems to be associated with cognitive impairment. Recent studies have shown that GSH depletion may cause metabolic stress in neurons by generating more ROS, which may eventually contribute to cognitive impairment. As an important factor in regulating the energy-redox axis, NNT may also be a potential regulator in aging. Indeed, NNT overexpression can restore the mitochondrial NAD+ levels, enhance the reprogramming efficiency of senescent cells, and prolong the life-span of mesenchymal stem cells by delaying senescence. Also, glucose metabolism is decreased and the life span is shortened in NNT–/– mice. Therefore, there is a relationship between the TCA cycle and aging-related changes in the nervous system.

28The majority of the ATP required by the brain is generated from the TCA cycle. Metabolomics analysis revealed that the utilization rate of glucose is altered in astrocytes extracted from 5xTg-AD mouse. Several proteins and TCA cycle enzymes involved in glucose metabolism are also altered in AD brain tissue. Pyruvate is taken up into mitochondria via the mitochondrial pyruvate carrier, as the final metabolite of glycolysis. Reduced mitochondrial pyruvate carrier activity might contribute to the inactivation of PDH caused by p-tau and alterative ketone body metabolism. Furthermore, pyruvate is the substrate for PDH and reduced in rodent models of AD. Acetyl-CoA is made by PDH complex-catalyzed oxidative decarboxylation and flows into the TCA cycle. Acetyl-CoA levels in the synaptosomes of Tg2576 AD mice model are reduced. Moreover, acetyl-CoA is a substrate for the acetylation of the lysine group in AD pathological marker proteins, including β-site amyloid precursor protein (APP) cleaving enzyme 1 (BACE1) and APP. Aβ causes a reduction in acetyl-CoA levels in neurons and glial cells. Both acetyl-CoA and succinyl-CoA are essential components for the formation of precursors of acetyl-choline, a neurotransmitter closely related to cognitive function. In addition, an examination of brains from deceased AD patients showed that the activities of the PDH complex, ICDH, and the α-KGDH complex are reduced, while the activities of SDH and MDH are increased. Citrate synthase (CS) activity appears to be negatively regulated by ApoE4 and also decreased in AD patients. As a downstream product of α-KG, succinyl-CoA may be reduced due to the downregulation of the pathway enzymes influencing the TCA cycle. However, there may be upregulation of succinate and aspartate. Other enzymes and metabolites related to the metabolic process (e.g., citric acid, cis-aconitic acid, and fumaric acid) are decreased. Thus, the changes in energy acquisition caused by these factors have various effects on many aspects of the body.

29Under aerobic conditions, the TCA cycle is mainly active in mitochondria. Hence, the integrality of mitochondrial morphology and function are particularly essential for this process. Impaired mitochondria are observed in AD and lead to glucose hypometabolism, OXPHOS damage, excessive ROS accumulation, elevated oxidative stress, and disruption of the main pathway of glucose metabolism. Mitochondria are not only the main source of ROS but also the target of its attack. 8-oxoguanine (8-oxoG) accumulates in mtDNA and the cytoplasm of hippocampal neurons, which is an obvious sign of oxidative stress in AD patients. A vicious cycle occurs because oxidative stress can further aggravate mitochondrial dysfunction. The activity of the ETC complex is significantly reduced in AD, suggesting impaired OXPHOS. This phenomenon has been confirmed in mitochondria isolated from 3-month-old AD mice and brain tissue from AD patients. The impairment of mitochondrial respiratory function in AD patients is also negatively correlated with Aβ levels and may be caused by the effect of Aβ on mitochondrial OXPHOS. Similarly, another study demonstrated that Aβ could inhibit mitochondrial complexes I and IV.

30Mitochondria-driven glucose metabolism abnormalities have been recorded by magnetic resonance spectroscopy (MRS) or nuclear magnetic resonance (NMR) in several studies with AD rodent models. Decreased phosphomonoester levels and increased levels of phosphocreatine and adenosine diphosphate reflect changes in the oxidative metabolic rate, suggesting that oxidative stress occurs in the AD brain. A reduction in glutamate, GSH, and GABA suggests a damaged glutamatergic and GABAergic glucose oxidation and neurotransmitter cycle, which is also present in these mouse models. Thus, glutamine synthase impairment and decreased glutamate flux through the GABA pathway may be caused by mitochondrial dysfunction. The depletion of mitochondrial GSH leads to increased H2O2 levels and decreased mitochondrial membrane potential in neurons and astrocytes. Aβ can induce the internalization of the glutamate A2 (GluA2) subunit, which is highly Ca2+ impermeable and contributes to the production of proinflammatory cytokines by microglia that accelerates neurotoxicity in AD patients. Therefore, altered levels of various enzymes and metabolites of the TCA cycle have an impact on mitochondrial function, resulting in abnormal redox homeostasis, reduced ATP production, and increased ROS production that injures surrounding tissues.

31The cognitive impairment observed in AD can be improved by upregulating acetyl-CoA levels to attain normal mitochondrial function. This view is supported by experiments using CMS121 and J147 as candidate drugs against AD in mice and long-term oral administration of acetyl-L-carnitine to AD patients, which provides additional acetyl-CoA. Specific therapeutic drugs for other metabolites require further research.

32Nutraceuticals appear to be a feasible approach for protecting mitochondria. Micronutrients are key co-factors that sustain mitochondrial metabolic hemostasis, such as the generation of ATP, construction of the electron transport complex, and oxygen detoxification. Thus, micronutrient deficiencies may cause drops in essential enzymatic activities and subsequently increase ROS production, decrease cellular energy metabolism, and aggravate mitochondrial abnormalities, promoting Aβ toxicity and AD progression. Coenzyme Q10 (CoQ10), an antioxidant forming part of the ETC, can enhance mitochondrial function and promote ATP synthesis. CoQ10 also inhibits nerve cell death resulting from oxidative stress and neurotoxins and has neuroprotective effects in double transgenic AD mouse models. Moreover, CoQ10 stabilizes mitochondria and reduces ROS production in fibroblasts from AD patients. Thus, mitochondria-specific nutraceuticals (e.g., vitamins and CoQ10) can be beneficial for AD patients.

33In summary, the TCA cycle and OXPHOS in the mitochondria can produce the greatest amount of energy for the body. The enzymes and metabolites involved in these reactions undergo various changes during aging and AD development. The imbalance in mitochondrial redox and resulting by-products (e.g., ROS) generated by these changes may cause inflammation in the surrounding tissues. Thus, to a certain extent, metabolic dysfunction could affect the occurrence and development of neurodegenerative disorders. However, most current treatments target symptom attenuation, and there is still no feasible research on specific treatments involving these altered intermediate products.

34The PPP is a significant component of intracellular oxidative catabolism, which is vitally important for oxidative stress resistance and the production of essential material for biological synthesis. Although the pathway does not generate ATP, it can produce NADPH to maintain the reduced form of GSH. When functioning as one of the most effective antioxidants, GSH is oxidized by ROS and is converted into the oxidized glutathione (GSSG), which subsequently enters a loop along with GSH peroxidase and GSH reductase.

35The PPP is regulated by enzymes and the NADPH/NADP+ ratio. Alterations in either of the two factors will critically affect the pathway. A crucial enzyme system supports the phosphopentose pathway, among which G6P dehydrogenase (G6PD) and transketolase are particularly important. G6PD and transketolase are both rate-limiting enzymes, responsible for the redox equilibrium and the non-oxidative branch in the PPP, respectively. In addition, transketolase plays a pivotal role in the material communication between glycolysis and the PPP. Moreover, an elevated NADPH/NADP+ ratio significantly inhibits G6PD, helping to disrupt G6P flux into the pathway. In conclusion, the PPP maintains redox hemostasis to prevent the initiation and development of oxidative stress in the brain.

36During aging, damaged mitochondria that produce less ATP and more ROS accumulate, leading to oxidative stress. G6PD plays a vital role in protecting neurons against endogenous ROS-mediated neurodegeneration in aging mice. In the cerebral cortex of aged mice, there is a decline in the levels and activities of G6PD and other GSH-regenerating enzymes. Moreover, G6PD appears to be neuroprotective against endogenous ROS in the aged human brain. However, alterations in G6PD levels in the human brain have not been identified. Furthermore, decreased NADPase levels and increased NAD kinases (NADK) levels may alleviate oxidative stress during human aging by promoting the synthesis of NADPH and inhibiting the production of NADH.

37Multiple risk factors, such as Aβ peptide, tau aggregation, and ApoE, play crucial roles in the PPP impairment contributing to AD. Oxidative stress and chronic inflammation are two critical factors demonstrated to trigger an elevation in the level of Aβ and subsequently the aggregation of toxic oligomers, particularly the Aβ42 forms. Later, aggregation of the Aβ peptides disturbs the redox balance, finally establishing a toxic cycle in AD patients. Tau phosphorylation, aggregation, and accumulation are closely linked to APP malfunction, which will inhibit cellular metabolism, including the PPP. Mutant ApoE not only is involved in Aβ clearance and aggregation, but also reduces the rate of glucose metabolism before AD symptom onset, along with the low flux of glucose into the PPP.

38NADPH, a reducing cofactor prominently recycled in the PPP, participates in the conversion of the oxidized form of GSH into the reduced form. Hence, the PPP establishes a crucial relationship between glucose metabolism and the redox equilibrium. Notably, Aβ not only triggers ROS generation but also decreases GSH levels in astrocytes. These effects lead to disequilibrium and aggravate oxidative stress. Interestingly, various groups have found that there is increased glucose flux through the PPP to counteract Aβ toxicity. Research has identified a decreased level of ribose-5-phosphate and an elevated level of lactic acid, which indicates upregulation of the PPP. Moreover, the dramatic degradation of phosphofructokinase B3 results in a higher level of glucose metabolism via the PPP than via glycolysis in neuronal cells.

39Accumulating evidence indicates that there is an alteration of the enzymes in the phosphate pentose shunt, which could significantly impact the antioxidant system in AD. G6PD is involved in generating NADPH for the reduction of the oxidized GSH, which helps to maintain the redox hemostasis. Several studies reported elevated levels of sulfhydryls and upregulation of G6PD in AD, which may be neuroprotective. However, the mechanism underlying this protective effect remains elusive. Notably, another report found that G6PD activity was decreased in the hippocampi of human AD brains. In addition to the enzymatic activity, antioxidants are generated in astrocytes to resist oxidative stress. In response to the aggregation of Aβ, ROS levels increase and there is a decrease in GSH in astrocytes in both AD and mild cognitive impairment.

40Furthermore, the activities of transketolase and their common coenzymes thiamine diphosphate, GSH peroxidase, glutathione-S-transferase, and δ-aminolevulinate dehydratase (δ-ALA-D) have all been demonstrated to be decreased in AD. Multiple metabolites altered by ApoE are identified within the PPP, including gluconolactone, gluconate, and G6P. These findings may provide a novel strategy to increase GSH production by regulating the activity of these enzymes in the PPP.

41AD therapies concentrate on oxidative stress, mainly targeting the PPP. For example, emphasized that advanced glycation end products and their methylglyoxal precursors are both biomarkers and pathogenic factors in AD, with direct neurotoxicity related to oxidative stress and apoptosis. Meanwhile, they proposed that the formation of advanced glycation end products occurred through a lower methylglyoxal concentration, which could be achieved by permitting a higher flux through the PPP. hypothesized that hollow manganese Prussian white nanocapsules (HMPWCs) participated in the resistance to the harmful effects of tau by relieving neuronal inflammation, eliminating ROS, and inhibiting tau hyperphosphorylation. Targeting of astrocytic NRF2, a regulator of GSH synthesis, could be a potent therapeutic strategy in AD.

42Because NADPH is essential for reducing GSSG to GSH, PPP is a pivotal part of the oxidative stress observed in AD pathology. Nevertheless, the upregulation of glucose flux into the PPP increases the synthesis of NADPH. However, it is controversial whether the activity of G6PD is activated or suppressed in AD. Hence, it is uncertain whether NADPH production is increased or decreased, and further investigation is needed. Despite the uncertain mechanism, there are still strategies to relieve PPP dysfunction and oxidative stress in AD.

43In this review, we summarized the metabolic deficits, including glucose metabolism dysregulation, glycolysis dysfunction, and PPP impairment, in AD (Figure 1). These deficits cause significant effects on AD pathogenesis. One direct consequence of these deficits is the inhibition of ATP generation, which leads to insufficient energy to support the normal neuronal functions and, ultimately, neurodegeneration. In addition, the metabolic deficits indirectly trigger neuronal death via mitochondrial dysfunction, oxidative stress (increased ROS, decreased NADPH), and inflammation.

44Compared to previous review articles, we not only described changes in the regulation of signal pathways but also discussed in detail, the changes of implicated enzymes and metabolites during energy metabolism in AD. Moreover, we highlighted the intimate relationship between AD pathogenesis and glucose metabolism dysregulation, implying that a focus on the preclinical and clinical manifestations of glucose metabolic dysregulation in AD might be a promising strategy to diagnose and prevent or slow the progression of this disease.

45Peripheral diabetes appears to aggravate these metabolic changes rather than cause them. Moreover, peripheral diabetes-related abnormalities do not directly influence the expression of diabetes-related genes in the brain, suggesting that AD pathology may play a key role in altering gene expression, which is correlative with diabetes in AD patients. The GLUT isoforms have variable effects on glucose utilization. Although altered cerebral glucose uptake is currently considered a predictive method for diagnosing AD, there is still a lack of adequate research to support its widespread use. Because of the constantly changing pathological processes in AD, the metabolite signal changes in the blood cannot reflect the changes in the brain in time, and there is still no accurate and easy-to-use index for preclinical and clinical diagnosis and treatment. Thus, it is difficult to assess whether peripheral signals related to the state of the disease are also reflected in the brain. The combination of multi-omics analysis and multi-type data will increase our understanding of the profound mechanisms of AD and identify potential biomarkers for diagnosis, prognosis and monitoring treatment of this disease. Therefore, further research is needed to develop more accurate and convenient diagnostic techniques for clinical use.

46Alzheimer’s disease treatment currently concentrates on symptom attenuation, targeting the specific enzymes or intermediate products. Different types of medications are used to recover the normal function of the proteins or ameliorate the intermediate outcomes, including some exogenous antioxidant interventions (vitamin E, polyphenol, and deuterated lipids). Although symptom treatment is the most widespread method for managing AD, it is not the best strategy to slow or reverse the neuronal degeneration because it only targets the manifestations of the disease and not its essence. Targeting these pathways may lead to the development of effective treatments (Table 1). We suggest that future AD treatments focus on early metabolic changes, clinical predictions, diagnosis, prevention, and the combined treatment of multiple pathogens at the pre-clinical and clinical stages using personalized drugs to prevent or delay the progression of AD. Further efforts are still needed to understand the metabolic basis of the etiology and pathogenesis of AD.

47XY and XZ contributed to the conceptualization and methodology. XY, YH, BW, SW, and XZ wrote the first draft of the manuscript. YH created the figure. BW prepared the table. All the authors approved the final manuscript.

48The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.