Homeostasis of carbohydrates and reactive oxygen species is critically changed in the brain of middle-aged mice: Molecular mechanisms and functional reasons

Highlights

• •Glucose is the main source of energy in the brain, whereas lactate and ketone bodies are minor sources.
• •Glycolytic and mitochondrial generation of ATP declines with the aging.
• •Aging is accompanied by intensified generation of reactive oxygen species leading to oxidative stress.
• •At middle-age, reorganization of energy metabolism and intensity of oxidative stress in the mice brain  takes place.

Abstract

The brain is an organ that consumes a lot of energy. In the brain, energy is required for synaptic transmission, numerous biosynthetic processes and axonal transport in neurons, and for many supportive functions of glial cells. The main source of energy in the brain is glucose and to a lesser extent lactate and ketone bodies. ATP is formed at glucose catabolism via glycolysis and oxidative phosphorylation in mitochondrial electron transport chain (ETC) within mitochondria being the main source of ATP. With age, brain's energy metabolism is disturbed, involving a decrease in glycolysis and mitochondrial dysfunction. The latter is accompanied by intensified generation of reactive oxygen species (ROS) in ETC leading to oxidative stress. Recently, we have found that crucial changes in energy metabolism and intensity of oxidative stress in the mouse brain occur in middle age with minor progression in old age. In this review, we analyze the metabolic changes and functional causes that lead to these changes in the aging brain.

Graphical abstract

1. Introduction

The brain plays a central role in the regulation and integration of physiological and biochemical processes in the whole body, as well as in maintaining of homeostasis and adaptive responses of animals to external and internal stimuli. In humans, the brain is also responsible for language production, thinking, decision-making, and behavior via sensory, motor, and cognitive processes. All these processes are the result of receiving, processing, and transmission of information in the brain by means of electro-chemical signaling, which requires significant energy supply. Therefore, it is not surprising that the brain belongs to high-energy consuming organs. It is estimated that being only 2% of the total body mass, the human brain consumes up to 20% of body's energy in a resting state - about 20% of the oxygen and 25% of the glucose consumed by the human body are spent on cerebral functions [1], [2], [3].

The high energy demand of the brain is due to the large number of nerve cells that need energy for proper functioning and to carry out specific metabolic functions [4]. In mammals, the brain contains from 10¹⁰ to 10¹¹ electrically excitable cells (neurons) and the same or up to 10-fold more glial cells [5]. The total brain energy consumption increases proportionally with the number of neurons among different species, including humans [5], and the total energy consumption of neurons in the state of signaling and rest is constant in different mammals [6]. In the cerebral cortex of mice, about 25–30% of the brain's energy is spent for functioning the neurons and glial cells themselves. The remaining 70–75% is used for signaling - sending and processing electro-chemical signals through the neural networks of the brain [6]. In the human brain, 15–20% of the energy is spent by glia and inhibitory neurons, the rest is required by excitatory neurons. Non-signaling needs predominate in white matter, while signaling needs dominate in gray matter [7]. Among the brain cells, neurons consume 70–80% of all energy produced in the brain, and the rest is utilized by glial cells [4,6]. In this way, neurons are obviously a highly specialized system that clearly controls their energy needs [8]. Energy needs are not the same throughout the brain but increase with increasing neuronal activity in certain local areas of the brain [9], [10], [11].

Glucose is a main energy source for the brain. In nerve cells, glucose is predominantly metabolized via glycolysis to form pyruvate which can enter either the tricarboxylic acid (TCA) cycle or be converted to lactate. Neurons also use lactate as a fuel, converting it back to pyruvate which is transferred into mitochondria for aerobic energy production. Oxidative phosphorylation in electron transport chain (ETC) of mitochondria is a main pathway of ATP production in nerve cells [2,12]. Brain aging is accompanied by compromised bioenergetics which include decline in glycolysis with simultaneous drop in supply of glucose and oxygen [11,13], and impaired mitochondrial function with lower ATP production [14]. Mitochondrial dysfunction is thought to be a main contributor to increased steady-state levels of reactive oxygen species (ROS) that are responsible for the intensification of oxidative stress during aging [3,14,15]. Mild oxidative stress, namely a low increase in a steady-state (stationary) ROS levels, induces adaptive stress response. At the same time, oxidative stress of strong intensity, when stationary ROS level becomes significantly higher, enhances oxidative damages of biomolecules because the capacity of defense mechanisms is overwhelmed [3,16]. In the brain, protection against ROS relies largely on the use of NAPDH as a cofactor in glutathione- and thioredoxin-related antioxidant mechanisms. NAPDH is mostly produced in the pentose phosphate pathway (PPP) at the reactions catalyzed by glucose-6-phosphate dehydrogenase (G6PDH) and 6-phosphogluconate dehydrogenase [17,18]. This pathway functionally connects ROS homeostasis with glucose catabolism [17,19]. Our recent studies have shown that during normal brain aging glucose metabolism shifts from glycolysis to the PPP to augment antioxidant defense and resist intensification of oxidative stress in old age [20,21], that is discussed below.

2. Energy demand and metabolic processes in neurons

In the brain, most of the energy is consumed at the synapses — the tiny gaps between nerve cells where signals are sent and received, i.e., transmitted from cell to the cell. It is calculated theoretically that no more than 2% of the total ATP consumption in the brain is used for protein synthesis, while 2–25% of the total ATP consumption is used for phospholipid synthesis [4,22,23]. The high level of energy consumption in phospholipid metabolism is probably associated with the active involvement of phospholipids in the dynamic signaling and membrane-recirculation processes in the brain [24]. In particular, it was found that the energy used by phospholipid metabolism in the rat brain is spent on fatty acid metabolism, phosphorylation of phosphatidylinositol and maintenance of the asymmetry of the phospholipid bilayer, as well as on the synthesis of phosphatidylinositol and ester phospholipids [25]. Energy consumption for actin dynamics and remodeling is evaluated in the range from 1% to 50% of neuronal energy consumption [26,27].

Human brain is estimated to consist of ∼86 billion of neurons from which each can have hundreds or thousands of synapses; in some cases, like in the cerebellum, Purkinje cells form up to 10⁵ synapses, which can be active with tens or hundreds of signals per second at a given time [5,12]. Electrical potential changes at synapses are based on ion gradients, particularly on a gradient of sodium ions (Na⁺), which is maintained by the activity of the Na⁺/K⁺-ATPase. The maintenance of Na⁺ gradient across the cell membrane ensures also the regulation of other ions, such as Ca²⁺, H⁺, Mg²⁺, Cl^– and HCO₃^–, which are transported together with Na⁺ via symporter or antiporter ion exchangers to maintain their physiological levels in the cells [12]. Synaptic vesicles were shown to be a major place of presynaptic basal energy consumption [28]. The presynaptic ATP is mainly consumed by the Na⁺/K⁺-ATPase in the plasma membrane, Ca²⁺-ATPases in the plasma membrane and endoplasmic reticulum, vacuolar H⁺-ATPase, motor proteins and protein disassembly machineries [4,29]. Much of the brain energy is spent on restoring potentials on the postsynaptic membrane of neurons after depolarization [30,31]. In addition, activation of postsynaptic receptors triggers processes that use ATP intensively [4,29]. It is believed that gradual synaptic dysfunction caused by disturbances in calcium homeostasis, changes in ionotropic/metabotropic receptors and mitochondrial dysfunction contributes to the cognitive decline associated with aging [14,[32], [33], [34]]. In addition, neurodegenerative disorders are frequently accompanied by deficits in synaptic energy connected with mitochondrial dysfunction [14,30].

In a state of rest, the metabolic activity of the brain also remains quite high, as a serious limitation of electrical activity in the brain reduces energy consumption by only two to three times. [28,35]. The lack of energy significantly affects the brain metabolism leading to rapid impairment of cognitive function. Nerve terminals are the areas that are sensitive to the lack of energy because ATP pools there are very limited. The ability of ATP synthesis on demand to satisfy activity-driven ATP hydrolysis depends to a big extent on the magnitude of local metabolic processes at rest [28].

Neuronal functions such as synthesis de novo, release and uptake of neurotransmitters, vesicle recycling, axonal growth and axonal transport also contribute to synaptic energy depletion [4,9,36]. Because of their high energy requirements and polarized structure, neurons require specialized mechanisms such as axonal transport to maintain energy homeostasis throughout the cell, especially at distal synapses and in long axons [37]. The machinery of axonal transport consists of tubulin microtubules, which extend along the entire length of the axon and provide the main cytoskeletal “racks” for transport, and motor proteins – kinesins and dyneins, which move cargoes from the cell body to the axon terminal (anterograde transport) and towards the cell body (retrograde transport), respectively. Motor proteins bind and transport different cargoes including mitochondria, cytoskeletal polymers, autophagosomes, and synaptic vesicles with neurotransmitters [38,39]. The polymerization and remodeling of tubulin microtubules requires energy in the form of GTP [40,41].

Axonal transport can be “fast” and “slow”. Membrane-associated vesicles and organelles are transported by the “fast” transport at relatively high speeds (>0.5 μm/s), whereas the transport of soluble cargoes (e.g., synapsin) or filamentous cargoes such as neurofilaments take much longer (<0.1 μm/s) [38]. Numerous mechanisms which mediate axonal transport operate to ensure the maintenance of presynaptic homeostasis by replenishing presynapses with newly synthetized or additional proteins or organelles, such as synaptic vesicles and mitochondria, and removing old and damaged synaptic components [38].

Synthesis and transport of peptide neurotransmitters differ from those used for low molecular mass neurotransmitters. Peptide-secreting neurons generally synthesize large polypeptide precursors in the rough endoplasmic reticulum of the cell body and these precursors undergo then maturation. Polypeptide precursors are transported to the Golgi apparatus and then are packaged into vesicles. After packing into vesicles, polypeptides undergo post-translational modification, including proteolytic cutting, disulfide bond formation, glycosylation, and phosphorylation. The peptide-filled vesicles are then transported along the axon to the synaptic terminal by a mechanism of fast axonal transport [42,43].

The synthesis of low molecular mass neurotransmitters usually occurs within presynaptic terminals. The synthesis of neurotransmitters involves enzymes that are synthesized in the body of the neuron and then transported to the axon terminals by slow axonal transport. Newly synthetized neurotransmitters are then loaded into synaptic vesicles [44,45]. Monoamines are stored in medium-sized, densely packed vesicles, whereas neurotransmitters such as glutamate, gamma-aminobutyric acid, glycine, and acetylcholine are accumulated in small transparent vesicles at nerve terminals. Vesicular storage of neurotransmitters depends on specific vesicular transporters and is controlled by an electrochemical H⁺ gradient across the vesicle membrane generated by a vacuolar ATP-dependent proton pump [46].

Mitochondria play a key role in meeting the energy needs of axons by generating ATP in the electron transport chain through oxidative phosphorylation. It is estimated that in neurons mitochondria give ∼93% ATP, while glycolysis provides only ∼7% ATP [30]. Recent data demonstrate how axonal mitochondrial traffic and anchoring are coordinated to respond to altered energy demands, but precisely how motor proteins are recruited to and released from transport vesicles remains unknown. Probably, a local decrease in energy availability acts as a signal to neighboring mitochondria, which can then be recruited to the right place to meet the local energy demand [39].

Impaired axonal transport in many neurodegenerative disorders is thought to contribute to the pathogenesis [37]. Axonal transport also deteriorates during aging [47]. In mouse model, the decline in axonal transport was found to occur in two distinct phases – first phase in young animals between 3 and 6 months of age, and the other during old age after 18 months, separated by a period of relative stability. In addition, neurons in older mice were still able to maintain higher rates of fast axonal transport. This indicates the existence of signals that can reverse the age-related decline in axonal transport [48]. Some data indicate that axonal energy demand may be supplemented by local glial cells, including astrocytes and oligodendrocytes [39].

3. Energy demand and metabolic processes in glia

Glia include astrocytes, oligodendrocytes, and microglia with specific and quite different functions. Glia was estimated to cover from ∼17.5% [4] to 40% of aerobic energy metabolism in the brain [49,50]. Some processes with high energy expenditure, such as phospholipid turnover and maintenance of phospholipid polarization in membranes, are likely to be higher in glia than in neurons because of the large surface area of the thin outgrowths of astrocytes and myelin of oligodendrocytes [23].

Astrocytes are the most abundant glial cells which display many metabolic functions. Particularly they regulate ion and water homeostasis [51,52], blood flow to meet neuronal energy demand [53,54], supplying the building blocks of neurotransmitters [55] and energy substrates to neurons [1,56]. Additionally, astrocytes participate in synapse formation, elimination, functional maturation, and plasticity during brain development and in adulthood [51,57] as well as brain repair after injury [58].

Increased neuronal activity leads to increased blood flow in the brain, and astrocytes play a main role in the mediation of this response. Increase in intracellular Ca²⁺ triggers synthesis of vasoactive metabolites of arachidonic acid in astrocytes. Astrocyte are in direct contact with blow vessels through their end-feet and release vasoactive metabolites onto blood vessels. Astrocytic prostaglandin E2 and epoxyeicosatrienoic acids act as dilators of blood vessels, whereas 20-hydroxyeicosatetraenoic acid has a vasoconstrictor effect [54]. Regulation of blood flow by astrocytes depends on oxygen availability. Under physiological conditions, astrocytic Ca²⁺-signaling leads to vasodilation, whereas under hyperoxic conditions vasoconstriction is stimulated [12,54,59]. Astrocytes are also involved in the formation of vascular tone, as the release of both 20-hydroxyeicosatetraenoic acid contracts vascular smooth muscle cells, forming vascular tone [54].

Axons are covered by multiple layers of myelin membrane formed by oligodendrocytes in the central nervous system (CNS) and by Schwann cells in the peripheral nervous system [60]. Myelination of axons makes them capable of fast and efficient nerve conduction. Besides saltatory nerve conduction, myelin has other important functions in nervous tissue. A growing body of evidence supports the active role of oligodendrocytes in association with neuronal processes, for example, by providing metabolic support to neurons and regulating ionic and water homeostasis [61,62]. In the CNS, myelination is also stimulated by axonal activity and astrocytes, while myelin clearance requires involvement of microglia, phagocyting cells that reside in brain. After myelination, long-term axonal integrity depends on glial supply of metabolites and neurotrophic factors [63]. White matter in the CNS consists mostly of myelinated axons, whereas dendrites and cell bodies form gray matter. It is estimated that the white matter of the CNS consumes approximately one third of the energy consumed by gray matter. On the other hand, myelination is energetically costly, and the metabolic cost of producing enough lipids and proteins for myelin synthesis may be higher than the energy saved by accelerating axonal conduction [64]. It is assumed that the main function of myelination is to provide fast nerve conduction, thereby increasing information processing and improving cognitive functions [30,64]. The composition of myelin differs from other biological membranes in terms of protein and lipid composition. Myelin consists of 60−80% lipids and 15−25% proteins in dry mass. The lipid part of myelin includes glycerolipids, cholesterol, and relatively high levels of sphingolipids, which include ceramides, sphingomyelins and glycosphingolipids [65]. Glycolipid called galactosylcerebroside is the primary lipid of myelin. Sphingomyelin makes the myelin sheath more durable. Myelin is not formed without cholesterol. The main source of cholesterol present in myelin is its de novo synthesis in oligodendrocytes or neighboring astrocytes [66]. In addition to structural function, sphingolipids regulate development of the nervous system, myelination, and maintenance of myelin stability. Plasma membrane organization is disturbed by changes in sphingolipid composition. Changes in the sphingolipid composition of myelin may have a crucial contribution to the phenotype of diseases characterized by demyelination [67]. The accumulation or dysregulation of sphingolipids is proposed to contribute to the pathogenesis of age-related neurodegenerative diseases such as Alzheimer's disease, Parkinson's disease and amyotrophic lateral sclerosis [68].

Microglia are resident macrophage-like cells of the CNS that make up to 5–12% of the total number of nerve cells in the mouse brain and 0.5–16.6% in the human brain [69]. Microglia within the CNS have a high energetic demand to function as they monitor for abnormalities and connections between neurons [70,71]. In the response to injury or pathogenic agent, microglia rapidly develop a classic pro-inflammatory program, releasing pro-inflammatory mediators such as tumor necrosis factor-alpha (TNF-$\alpha$), interleukins (IL-1$\beta$, IL-6, IL-12), ROS, nitric oxide (^•NO), prostaglandins, and chemokines that help recruit other immune cells in the affected area and enhance the inflammatory response. Once the noxious stimulus has been eliminated, it is crucial that the inflammatory response is muted and resolved; this is achieved by active redirecting the microglial phenotype towards an alternative immunomodulatory profile, characterized by the release of anti-inflammatory cytokines such as transforming growth factor beta (TGF-$\beta$) and IL-10 [72,73]. These immunomodulatory mediators inhibit the release of pro-inflammatory factors and promote tissue regeneration, thereby facilitating the return to homeostasis. It has been suggested that microglial dysfunction plays a crucial role in neuroinflammation, and age-related changes in microglial Ca²⁺-signaling have been reported to underlie this phenotype of impairment [3,69,73]. In fact, microglial pro-inflammatory activation has been implicated in the pathology of many neurodegenerative disorders, especially Parkinson's and Alzheimer's diseases, where neuronal damage occurs as a consequence of a prolonged pro-inflammatory response [73,74]. Microglia, as well as other immune cells, themselves undergo age-related morphological and functional changes called activation [3,75]. In particular, a known sign of aged glia is the accumulation of lipofuscin granules – an undegradable product of lipid oxidation due to membrane damage or damage to mitochondria and lysosomes [75,76]. Another signs are decrease in the volume of regions covered by microglial processes, disruption of the regularity of microglial domains and the appearance of microglial aggregates [77,78]. These morphological changes are accompanied by a decrease in the initial mobility of microglial processes and a slowdown in the directional movement of processes towards the lesion [79].

Epidemiological data on neurodegenerative diseases indicate that middle age is a critical period for the onset and progression of the age-related neurological diseases and identification of promising treatment may help to develop strategies to stop or reverse the pathologies [75]. The accumulated to date data indicate that in mice the first age-related changes in microglia functioning occur at the age of 9–12 months, when animals are not even considered old. Microglia in middle-aged mice showed small, not yet significant, changes in morphology compared to their counterparts in young mice [75,79]. However, a significant decrease in basal microglial motility was observed [79], indicating that the surveillance function may be impaired already at this early stage. In addition, one study found two distinct phenotypes of microglia aging: a reactive phenotype, which is predominantly found in middle-aged animals, and a dysfunctional/sensitive phenotype, which is ubiquitous in old mice [69].

4. Energy sources in neuronal and glial cells

Most tissues are able to use various organic substances, including carbohydrates, lipids, and proteins as a source of energy. However, the brain uses predominantly glucose and to a smaller extent ketone bodies and lactate. Glucose is a main brain's energy source and is mostly used for ATP production via oxidative phosphorylation which requires oxygen. It is estimated that brain consumed about 20% of oxygen utilized by the body. Oxidative metabolism in the brain is on average about 10 times higher than in the rest of the body [3,80]. A part of glucose is used to generate reducing equivalents (in the form of NADPH) and ribose for antioxidant defense and biosynthetic purposes [2,17]. Another features of the brain is that it is not able to accumulate sufficient energy reserves in the form of glycogen stores and possesses low gluconeogenic potential [81]. Due to this, the brain needs a continuous supply of glucose and oxygen. In human brain, when the blood glucose level decreases by only about twofold, severe neurological disorders may occur [28].

Ketone bodies can be utilized by the brain when glucose availability is reduced, in particular under fasting, exercise, low carbohydrates/high-fat diet. In those cases, the liver produces ketone bodies such as acetoacetate, β-hydroxybutyrate, and acetone from fatty acids and ketogenic amino acids and release them into bloodstream [2,82]. Ketones are important energy substrates for the brain during development, providing up to 30–70% of its energy needs [83]. Recently it was shown that astrocytes may convert short-chain fatty acids to ketone bodies to supply neurons (cited after [3]).

Substrates used by the brain are taken from the bloodstream by transporters in endothelial cells, whose tight junctions form the blood-brain barrier. Ketone bodies cross the blood-brain barrier via specific monocarboxylate transporters. Inside the cells, ketone bodies can be directed into oxidative or anabolic pathways depending on the needs of the cell [2,82]. Glucose is transported by several types of hexose transporters. GLUT1 is the main glucose transporter that delivers glucose to neurons and glia when sufficient glucose is supplied by blood [84]. It should be noted that GLUT1 is mostly expressed on endothelial cells and astrocytes but not on neurons. Neurons express mostly GLUT3 transporters, which have high affinity to glucose and are mainly located in axons and dendrites and provide constant supply of neurons with glucose when its level in blood is low [85,86].

After entering the nerve cell, glucose is phosphorylated by hexokinase to glucose-6-phosphate (G6P), which can be metabolized in three ways. First, G6P can be used in the glycolytic pathway, where its metabolic conversion produces two molecules of pyruvate, two molecules of ATP and two molecules of NADH (Fig. 1). Pyruvate can enter mitochondria and undergo oxidative decarboxylation to form acetyl-CoA, which is further metabolized through the TCA cycle and oxidative phosphorylation (OXPHOS). As a result, consumption of one glucose molecule results in formation in TCA three NADH molecules and one FADH₂ molecule that further are oxidized in the ETC of inner mitochondrial membrane with oxygen being a final acceptor of electrons. The energy of reducing equivalents is used for ATP synthesis in the process of oxidative phosphorylation [23,80]. Pyruvate can also be converted to lactate by lactate dehydrogenase (LDH), which reduces pyruvate to lactate and oxidizes NADH to NAD⁺ [12]. The second metabolic way of G6P utilization is its conversion by G6PDH in pentose phosphate pathway. The PPP is a combination of two sets of reactions, an oxidative part, which includes G6PDH and provides oxidation of glucose-6-phosphate to ribulose-5-phosphate, and a non-oxidative part, which provides interconversion of phosphorylated carbohydrate derivatives (Fig. 1). Further, ribulose-5-phosphate can be converted into ribose-5-phosphate, a precursor of nucleotides. In the brain, depending on the metabolic status of cells, the need for NADPH and ribose-5-phosphate can vary significantly, so the ratio between the oxidizing and non-oxidizing parts of the PPP may shift in the appropriate direction [18]. G6PDH and its partner in the PPP - phosphogluconate dehydrogenase - are thought to be the main cellular NADPH producers [3,17]. Since NADPH is a cofactor for glutathione (GSH) and thioredoxin-dependent antioxidant enzymes, glucose utilization through the PPP is essential to maintain the antioxidant potential of the brain [3,17,21]. Third way of G6P metabolism is its using for synthesis of small amounts of glycogen, the only energy reserves in the brain. Glycogen synthesis is important for glucose metabolism in astrocytes. Astrocyte glycogen acts as a form of protection against hypoglycemia, ensuring the preservation neuronal function [81,87].

Fig. 1.

Redistribution of carbohydrate metabolism from glycolysis to pentose phosphate pathway in mouse brain starting from middle age [21]. Blue arrows and labels denote glycolytic conversions, orange arrows and labels denote conversions of pentose phosphate pathway, green arrows and the label denote glycogen synthesis pathway, a characteristic of astrocytes. Abbreviations (unless not deciphered in the text): HK – hexokinase, GPI – glucose phosphate isomerase, PFK – phosphofructokinase, Aldo – aldolase, TPI – triosephosphate isomerase, GAPDH – glyceraldehyde 3-phosphate dehydrogenase, PGK – phosphoglycerate kinase, PGAM – phosphoglycerate mutase, Eno – enolase, PK – pyruvate kinase, LDH – lactate dehydrogenase, G6PDH – glucose-6-phosphate dehydrogenase, 6PGDH – 6-phosphogluconate dehydrogenase, PGM – phosphoglucomutase, G1P – glucose-1-phosphate, G6P – glucose-6-phosphate, F6P – fructose-6-phosphate, F1,6P – fructose-1,6-biphosphate, DHAP – dihydroxyacetone phosphate, GA3P – glyceraldehyde 3-phosphate, 1,3BPG – 1,3-bisphosphoglycerate, 3PG – 3-phosphoglycerate, 2PG – 2-phosphoglycerate, PEP – phosphoenolpyruvate, Pi – inorganic phosphate, 6PGL – 6-phosphogluconolactone, 6-PG – 6-phosphogluconate, R5P – ribulose-5-phosphate.

The metabolic fate of glucose seems to depend on the type of nerve cells, as the latter show different expression patterns of genes involved in glucose metabolism [2,12,88]. Astrocytes are thought to be more glycolytic, while neurons use predominantly oxidative metabolism [2,89]. Moreover, neurons and astrocytes show tight metabolic coupling. According to the “astrocyte-neuron lactate shuttle” hypothesis [90], lactate is mainly produced by astrocytes. From astrocytes, lactate is transported by monocarboxylate transporters (MCTs) into neurons: it is exported via MCT4 in astrocytes and taken up by neurons via MCT2 transporter [91]. In neurons, lactate is oxidized back to pyruvate, which is metabolized via TCA and OXPHOS in mitochondria [12]. It is shown that the main provision of energy metabolism of axons is carried out by the astrocyte-neuron lactate shuttle [92]. The metabolic role of lactate is not limited to meeting the energy needs of neurons. It also acts as a signaling molecule that modulates neuronal functions, including excitability, plasticity, and memory consolidation [90].

Astrocytes predominantly express LDH5 which possesses a high affinity for pyruvate ensuring its conversion to lactate [2,93]. In contrast, neurons contain mainly LDH1 isoform, which has a high affinity for lactate and therefore ensures lactate conversion to pyruvate [2,94]. Pyruvate becomes available for mitochondrial metabolism only through the malate-aspartate shuttle linking glycolysis and the TCA cycle [23].

It is assumed that high glycolytic activity in astrocytes is associated with high activity of pyruvate dehydrogenase kinase isoform 4 and 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase [2,88]. Pyruvate dehydrogenase kinase 4 (PDK4) phosphorylates E1α subunit of pyruvate dehydrogenase (PDH), the main enzyme of PDH complex. Phosphorylation inactivates PDH, thereby reducing the supply of acetyl-CoA to the TCA cycle [2,95].

The flux of metabolites through glycolysis largely depends on the activity of key enzymes, among which the main one is phosphofructokinase 1 (PFK1) [20,21]. The activity of the latter is modulated by a number of low molecular mass metabolites, in particular by fructose-2, 6-bisphosphate (F2,6P), which is produced by 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase (isoform PFKFB3) [88]. Astrocytes express high levels of PFKFB, while neurons lack PFKFB3 activity due to its fast proteasome-dependent degradation. Thus, glycolysis in neurons is of low intensity [9,88]. In line with this, resting cultured neurons preferentially consumed lactate over glucose if both substrates were available in the medium [96].

Neurons are unable to store glucose in the form of glycogen because they contain high levels of glycogen synthase kinase 3 (GSK3), which phosphorylates glycogen synthase, making it a target for ubiquitin-dependent proteasomal degradation [2]. Therefore, enzymes of glycogen metabolism and glycolysis together ensure the production of lactate from glycogen in astrocytes to meet the energy needs of neurons [92,97].

Brain aging is shown to be accompanied by compromised bioenergetics, which can be attributed to a decreased supply of glucose and oxygen [11,13,98], decline in glycolysis [11,20,21,99], and impaired functioning of mitochondria with diminished ATP production [14,21,100,101]. In addition, we found the age-dependent increase in the activity of G6PDH, a key PPP enzyme, in the aged brain [20,21]. Using mutant mice, defective in G6PDH, it was found that brain G6PDH protects against endogenously-induced oxidative DNA damage and neurodegeneration in aged animals [19]. The information regarding age changes in metabolism of other brain energy sources like ketones is scarce [102], [103], [104]. Recently we have shown that middle-aged (12 months) mice had a significant decrease in β-hydroxybutyrate levels in the cortex and cerebellum as compared with the values in young ones (6 months) [20,21]. The activity of β-hydroxybutyrate dehydrogenase (HBDH), the main enzyme of utilization of ketone bodies, remained unchanged in the cerebral cortex of males but decreased in the female brain with increasing age [20,21]. Such neurological disorders as Alzheimer's, amyotrophic lateral sclerosis, Parkinson's, and Huntington's diseases are characterized by deterioration of energy-providing pathways which take place long before clinic manifestations of the diseases [102,105,106].

5. Sources of reactive oxygen species in the brain cortex

Mitochondria provide most of the ATP needed in the brain to perform various neuronal functions, particularly synaptic transmission [107]. During normal cellular respiration, oxygen is consumed by mitochondria, and at the final stage of the oxidative phosphorylation it is reduced to H₂O in the reaction catalyzed by cytochrome c oxidase. Under normal conditions, less than 2% of oxygen leaks from the mitochondrial ETC with the formation of products of incomplete reduction of oxygen, so-called reactive oxygen species [15,108]. This leakage of protons arising from the oxidation of NADH and FADH₂ takes place mainly at complexes I (NADH dehydrogenase) and III (ubiquinone-cytochrome c oxidoreductase) (Fig. 2) [109,110] of the ETC and leads to the formation of reduced oxygen ion, known as superoxide (O₂^•−) [111]. Complex II also may generate some ROS amounts when complex I/III is inhibited or succinate concentrations are low [112,113]. Superoxide anion generated in mitochondria has a very short half-life, as it is scavenged by manganese superoxide dismutase (Mn-SOD) in the mitochondrial matrix or copper/zinc superoxide dismutase (Cu,Zn-SOD) in the intermembrane space and cytosol, leading to the formation of hydrogen peroxide (H₂O₂) [114]. Hydrogen peroxide is a non-radical ROS which is able to penetrate membranes, leave the mitochondrial matrix and oxidize different biomolecules in cytosol, among which molecules that contain SH-groups are primary targets. In the presence of electron donors such as transition metals (Fe²⁺ or Cu⁺), H₂O₂ further can be reduced to hydroxyl radical (HO^•), the most reactive among listed above ROS which can rapidly oxidize lipids, carbohydrates, proteins, and DNA (Fig. 2) [15,114,115]. Therefore, the strategy of cell protection against hydrogen peroxide includes prevention of its conversion to hydroxyl radical. In view of this, cells possess a number of enzymatic and non-enzymatic antioxidants that can successfully scavenge hydrogen peroxide. In particular, H₂O₂ can be converted to H₂O in the presence of catalase, peroxiredoxin, or glutathione peroxidase (Fig. 2) [115]. Full elimination of ROS does not occur, since ROS play important signaling roles; in particular, ROS regulate neuronal development, synaptic transition, and plasticity via controlled activation of redox-sensitive signal transduction pathways [116,117].

Fig. 2.

Formation of ROS by respiratory chain complexes. SOD - manganese-containing superoxide dismutase, GPx - glutathione peroxidase, GSH - reduced glutathione, GSSG - oxidized glutathione, QH₂/Q - reduced/oxidized ubiquinone, cI - complex I of mitochondrial respiratory chain, cIII - complex III of mitochondrial respiratory chain, FMN - flavin mononucleotide, Fe-S - iron-sulfur clusters of the complex I.

Nerve cells are particularly prone to the formation of ROS in large amounts, the distribution of which can differ significantly between different parts of the brain [118]. Moreover, it is well known that more ROS can be formed from damaged mitochondria in various pathological conditions and ROS production in mitochondria increases with age [3,14]. It is assumed that mitochondria are not only the main site of ROS formation, but also the main target for ROS; in particular, mitochondrial DNA is actively exposed to ROS attack due to close proximity to the sites of their production. Another reason for more extensive modification of mitochondrial DNA than nuclear one by ROS is a weaker mitochondrial DNA repair system. As a result, more oxidative damage are accumulated in mitochondrial than in nuclear DNA. As a consequence, oxidative damage leads to an increase in the frequency of mutations in mitochondrial DNA, which further disrupts the functioning of the mitochondrial ETC, leading to increased ROS production and increased damage to the mitochondria themselves [9,119]. In normal (e.g. non-ischemic) cells, most ROS (about 95%) are formed in the ETC, while the rest can be formed by a group of enzymes, e.g., NADPH-dependent oxidases, monoamine oxidases, lipoxygenases, cyclooxygenases, nitric oxide synthase, cytochrome P450, and dihydroorotate dehydrogenase [80,111,115,120].

NADPH oxidases (NOX) are electron transporting proteins that are localized in membranes and catalyze the transfer of one electron from NADPH onto molecular oxygen with the formation of superoxide anion radical, although certain amounts of H₂O₂ are supposed to be formed by this enzyme. NOXs were originally found in immune cells such as macrophages and neutrophils. To date, they were shown to be expressed in many non-phagocytic cells, including microvascular endothelium and brain microglia [121,122]. Normal function of NOXs is essential for such processes as defense against pathogens and neuronal signaling; in particular, due to production of low levels of H₂O₂ which can act as signaling molecule in many regulatory pathways including those regulating ion channels and Ca²⁺ signaling in neurons [120,123]. Excessive ROS production by NOX is thought to contribute to brain aging, neurotoxicity, and neurodegeneration [121,124]. NOX2, one of the main isoforms of NADPH oxidase in the CNS [125], was found to produce ROS involved in hippocampal microglial activation during development of postoperative cognitive dysfunction in aged mice [126].

А significant amount of ROS can also be formed by the monoamine oxidases from biogenic amines. In particular, oxidation of dopamine by monoamine oxidase in dopaminergic neurons leads to formation of H₂O₂ [127]. Nitric oxide synthase (NOS) which catalyzes oxidation of l-arginine to citrulline and ^•NO is another significant ROS source. There are three isoforms of this enzyme found in animal cells, in particular, two constitutive forms of NOS - neuronal (nNOS) in neurons and endothelial NOS (eNOS) in endothelial tissue, while the inducible isoform (iNOS) was found in glial cells. Nitric oxide acts also as a neurotransmitter, associated with synaptic plasticity and synaptic activity regulation through protein S-nitrosylation [128]. With a significant increase in the level of ^•NO, there is a transition from S-nitrosylation to irreversible oxidation of cysteine thiols to sulfuric (SO₃⁻) acid [129]. Nitric oxide at high levels can inhibit mitochondrial respiration, in particular through direct inhibition of pyruvate dehydrogenase, aconitase, and cytochrome c oxidase in mitochondrial ETC leading to glycolytic phenotype [130]. It should be noted that ^•NO stimulates glycolytic activity in astrocytes but not in neurons [111,131], promoting lipid peroxidation, protein nitration, and DNA damage [111,129].

6. The major targets of reactive oxygen species in the brain

Senescence affects all critical systems that provide living of cells. In particular, a gradual loss of function may encompass DNA repair, protein translation and folding, autophagy, antioxidant defense, energy production, and several other branches of metabolism [132,133]. Thus, cells gradually lose important functions with age until the moment when the loss is not compatible with life. All listed above life-maintaining processes and systems work flawlessly in young organism. However, genetic background and environmental impacts may lead to impairments in folding of critical proteins due to their oxidative modifications [134,135].

For example, it was shown that a set of enzymes involved in energy production is vulnerable to oxidative modification in aging brain. Such glycolytic enzymes as aldolase, triosephosphate isomerase (TPI), glyceraldehyde 3-phosphate dehydrogenase (GAPDH), and pyruvate kinase (PK), were oxidized in rat cerebral cortex during aging [135]. Phosphoglycerate kinase and PK were shown to be oxidatively modified in cerebellum, whereas α-enolase - in hippocampus [135]. The listed enzymes provide ATP production in glycolysis, therefore, their oxidation and loss of function would affect energy production in neurons and glia (Fig. 3). Nervous tissue requires energy to oppose age-related functional decline, especially for the synthesis of new proteins (in order to replace oxidized ones), proteolysis, and micro-, and chaperone-mediated autophagy [136,137]. Senescence affects other proteins directly or indirectly involved in ATP production, such as mitochondrial enzymes ATP synthase, isocitrate and malate dehydrogenases, and aconitase, as well as cytosolic creatine and adenylate kinases [135]. Similar targets of oxidative modifications were found in the brains of mice prone to accelerated senescence, also known as SAMP8 (senescence-accelerated mouse prone 8). In addition to several above listed proteins, redox proteomics revealed advanced oxidative carbonylation of lactate dehydrogenase 2, α-spectrin, and dihydropyrimidinase-like protein 2 [138]. Our study shows age-related loss for activities of NADP-dependent malate dehydrogenase, phosphofructokinase, pyruvate kinase, lactate dehydrogenase, glyoxalases I and II, aconitase, and glutathione peroxidase in the cerebral cortex of mice [21]. Similar trends for age-related loss of enzyme activity were characteristic of cerebellum [20]. It is interesting, that some of the enzymes demonstrated different dependence on age in males and females. For instance, aconitase, NADP-dependent malate dehydrogenase, and β-hydroxybutyrate dehydrogenase showed gradual decrease in the activity with age only in females, unlike a bell-shaped pattern in males [21]. On the contrary, activities of catalase, glyoxalase II, pyruvate kinase, and lactate dehydrogenase decreased only in the cerebral cortex of males. In addition to the cytosolic proteins involved in glycolysis, several mitochondrial proteins were found to be prone to oxidation. Except for already mentioned aconitase and ATP synthase, voltage-dependent anion channel (VDAC), pyruvate carboxylase, E1 component of pyruvate dehydrogenase, mitofilin (a component of mitochondrial contact site and cristae organizing system, also known as Mic60), adenine nucleotide translocase, aspartate amino transferase, subunits NDUFS1 and NDFUS2 of mitochondrial respiratory chain complex I, subunit UQCRC1 of complex III, and subunits of cytochrome c oxidase are among such proteins (Table 1) [15,139].

Fig. 3.

Glycolytic enzymes that are reportedly modified by ROS in mouse brain during aging. Red labels with white font denote enzymes subjected to age-dependent oxidative modification revealed by redox proteomics. Blue labels with white font denote enzymes that were not reported as particularly susceptible to oxidative modification. Purple labels with white font denote enzymes whose activity was remarkably decreased with age [20,21]. PPP - pentose phosphate pathway, HMP - hexose monophosphate pathway. All other abbreviations are as in Fig. 1. Names shown in bold font with big size denote metabolites whose concentration would supposedly grow with age in case of oxidative modification of specific glycolytic enzymes.

Table 1.

List of proteins that are undergone to oxidative modifications in different regions of rodent (mouse and rat) brain.

Protein name	Organism	Brain region	Source
UQCRC1	Rat	Hippocampus	[139]
ATP synthase, alpha subunit	Rat	Hippocampus, cortex	[139]
ATP synthase, subunit b	Rat	Whole brain	[140]
Glutamate dehydrogenase 1	Rat	Cortex	[139]
Aspartate aminotransferase 2	Rat	Whole brain	[140]
mtHsp70 (Hspa9, mortalin)	Rat	Cortex, striatum	[139]
Aldehyde dehydrogenase	Rat	Cortex	[139]
Elongation factor Tu	Rat	Striatum	[139]
Aconitase	Rat	Striatum	[139]
Isocitrate dehydrogenase, subunit α	Rat	Cerebellum	[139]
Isocitrate dehydrogenase	Mouse	Whole brain	[141]
Succinyl-CoA-ligase	Mouse	Whole brain	[141]
SOD2, Mn-containing	Mouse	Whole brain	[141]
VDAC	Rat	Cerebellum	[139]
VDAC	Rat	Whole brain	[140]
Adenine nucleotide translocase	Rat	Whole brain	[140]

Oxidative lesions to enzymes involved in glycolysis and the TCA, and consequent suspension in intensity of energy production, may lead to repurposing of glucose metabolism to the PPP. First two enzymatic reactions of PPP yield NADPH, which, in turn, is used to provide antioxidant defense by means of glutathione- and thioredoxin-dependent peroxidases [142]. Nevertheless, we saw that brain glutathione peroxidase was also subjected to age-related functional decline [21].

Neurons are critically dependent on ATP levels, since ATP is used for transmembrane potential maintaining and restoration after nerve impulses as we described above [143], [144], [145]. The lower steady-state levels of ATP in neurons, the slower the listed processes are expected to be [144,145]. In addition, age-related decline in antioxidant capacity may lead to increased lipid peroxidation. Lipids that are attributable to the cells of nervous tissue contain high amounts of polyunsaturated fatty acids, which are prone to peroxidation [146,147]. Lipid peroxides and products of their metabolism, such as 4‑hydroxy-2-nonenal, can be additional damaging factors for surrounding proteins. As we mentioned above, such glycolytic enzymes as aldolase, TPI, GAPDH, enolase, and PK are sensitive to oxidation. Among them, GAPDH is a well-known target, which contains sensitive to oxidation thiol groups [148]. Thiols of GAPDH can be modified via interaction with ROS or with the products of lipid peroxidation. In turn, blockade of GAPDH reaction may lead to a shift to formation of dihydroxyacetone phosphate. The latter metabolite is a precursor for formation of methylglyoxal, a glycating agent (Fig. 4). Advanced glycation leads to modification of essential proteins that eventually lose their functionality [149]. It was indeed demonstrated that the level of advanced glycation end-products (AGEs) increases in mouse brain with age [150]. Interestingly, this increase in AGE level was exacerbated by a diet with high glycemic index [150], whereas intermittent fasting caused a decrease in intensity of glycation processes in the brain [151].

Fig. 4.

Age-dependent changes in energy metabolism in aging mouse brain. A decrease in function is observed for glycolysis, glyoxalase system, ketone body metabolism, and oxidative phosphorylation [20,21]. Metabolites and metabolic pathways upregulated during aging are shown in red font, processes that downregulated during aging are shown in light-green font, metabolites and pathways with ambiguous or not quite clear changes during aging are shown in light gray font, metabolites whose concentration may supposedly grow due to inhibition of the lower part of glycolysis are shown in orange font. Abbreviations: PPP - pentose phosphate pathway, ODC - oxidative decarboxylation, TCA - tricarboxylic acid cycle, OXPHOS - oxidative phosphorylation, DHAP - dihydroxyacetone phosphate, G6P - glucose-6-phosphate, MG - methylglyoxal, GSH - reduced glutathione.

7. Links of neuronal metabolism that are the most vulnerable to age and age-related oxidative modification

There are several lines of evidence that cells try to compensate the lack of proteins they lose due to oxidative modification occurring in the aged brain. It was shown that expression of several listed above glycolytic enzymes increases in mouse brain during aging [152]. Expression of proteins prone to carbonylation was found to increase 1.5- to 2.5-fold in the brains of aged mice [153]. Our study also shows that middle-aged male mice had the highest activities of hexokinase, G6PDH, and aconitase in comparison with young and old counterparts [21]. In addition, middle-aged males showed the highest activities of NADH-linked respiration and cytochrome c oxidase. Aconitase and complex I of the mitochondrial respiratory chain, which is responsible for NADH-linked respiration contain iron-sulfur clusters which are very sensitive to oxidation [154]. These requirements of aging mouse brain may explain increase in the expression of G6PDH, which catalyzes NADP⁺ reduction, yielding NADPH.

In our studies, calorie restriction in the form of every-other-day-feeding (EODF) helped to improve several parameters of brain metabolic network. For instance, EODF provided activity of aconitase in the cortexes of middle-aged and old male mice as high as in young ones [21]. Similar effect was observed for glutathione-S-transferase activity, NADH- and succinate-linked respiration. Also, increased G6PDH activity was observed in the cerebella of male mice on EODF as compared to the ad libitum fed counterparts [20]. Observed changes in the mouse brain are confirmed by proteomic studies. For instance, it was found that the levels of glycolytic enzyme TPI and transketolase of PPP in the mouse brain grew with age (Fig. 4) [155]. Interestingly, several mitochondrial proteins such as glutamate dehydrogenase, malate dehydrogenase, mitochondrial ribosomal protein L37, ubiquinol cytochrome c reductase core protein 2, and voltage-dependent anion channel 1, showed definite peak in expression in the brains of middle-aged mice [155]. Similar pattern for the mitochondrial proteins was confirmed by Stauch et al. [152]. Namely, abundance of subunits of the oxidative phosphorylation complexes – NDUFB8 (complex I), UQCRC2 (complex III), MTCO1 (complex IV), and ATP5A1 (complex V) - was shown to reach peak at about 12 months of age, and drops on the 24th month [152]. An age-dependent functional decline in efficiency of the lower part of glycolysis that starts from 1,3-bisphosphoglyceric acid, in the mouse brain is confirmed by the metabolomic studies. Phosphorylated carbohydrates and pyruvate were found to prevail in the brains of adult mice [156]. Phosphorylated aldoses and ketoses are intermediates formed in the upper part of glycolysis and PPP. Pyruvate, despite formed in glycolysis, can also be formed via oxidation of lactate (Fig. 4). As we mentioned above, lactate can be produced by astrocytes and to a small extent arrive from other tissues [90]. On the other hand, it was shown that several links of brain mitochondrial metabolism, in particular, multi-subunit enzyme complexes such as PDH and alpha-ketoglutarate dehydrogenase, have lower capacities in older adults than in young ones [156]. A decrease in the activities of PDH and cytochrome c oxidase with age was observed in both, normal and triple-transgenic model of Alzheimer's disease female mice [104]. A drastic decrease in the activities of the mentioned enzymes was seen in the last trimester of life. There are several pieces of evidence that neurons of old female mice may repurpose their energy metabolism to using ketone bodies (formed from myelin of lipids) for production of acetyl-CoA instead of using pyruvate for this purpose [157]. In particular, it was shown that expression of long-chain hydroxyacyl-CoA-dehydrogenase, 3-oxoacid-CoA transferase 1, carnitine palmitoyltransferase 1, and sphingomyelinase was substantially increased in the white matter of aged females [104,157]. Nevertheless, this observation raises question on the oxidative metabolism of acetyl-CoA and ketone bodies obtained from myelin lipids at a functionally declined respiratory chain. Above, we mentioned age-related declines for the majority of mitochondrial respiratory chain complexes and TCA cycle enzymes in murine brain. Therefore, neither acetyl-CoA obtained from ketone bodies, nor that formed from pyruvate, could be effectively metabolized in aged brain. This may suggest reorganization of brain lipids rather than using them as alternative energy sources.

8. Redistribution of carbohydrate catabolism fluxes in senescing mouse brain

Earlier it was found that with brain aging redistribution of intermediates of carbohydrate catabolism between glycolysis and PPP in favor of the latter takes place (cited after [3]). However, neither mechanisms, nor reasons of that redistribution remained unclear for a long time. Later, such redistribution was supposed could be related to enhanced needs in phosphoribose for the biosynthesis of nucleotides and NADPH levels for lipid biosynthesis. Both these assumptions were explained by necessity of reparation of damaged nucleic acids that is associated with intensified DNA damage during aging [17,96]. In our recent works, we have found that in middle-aged mice the activity of key glycolytic enzymes phosphofructokinase and PK in cortex and cerebellum critically decreased compared to young animals [20,21]. At the same time, the activities of main NADPH producing enzyme glucose-6-phosphate dehydrogenase was substantially increased. Therefore, we proposed that the abovementioned redistribution of glycolytic intermediates between glycolysis and the PPP resulted from slowing glycolysis and accelerating of the PPP due to corresponding changes in the activities of key enzymes of these pathways. Interestingly, the changes visible in middle-aged mouse brain were generally kept in old mice. It is worth noting, that levels of lipid peroxides, products of ROS attack on polyunsaturated lipids, also critically increased in the middle age and virtually were not enhanced in old mice [20,21]. Therefore, we proposed that redistribution of fluxes of glucose catabolism intermediates taking place in the brain of middle-aged mice was related to corresponding changes in the activities of key glycolytic and PPP enzymes and could be responsible for prevention of intensification of oxidative stress in late phases of lifespan.

9. Possible evolutionary meaning of metabolic repurposing in senescing brain

Metabolic repurposing in aged brain, caused by a decrease in activities of certain glycolytic enzymes, as well as enzymes of the TCA cycle and oxidative phosphorylation, is not characteristic of all species. It was found that hexokinase 2 (HK2), which is localized and tightly bound to the mitochondrial outer membrane (MOM), consumes ATP synthesized in mitochondria, conferring a mild depolarization [158] (Fig. 5). This process is supposedly beneficial for longevity since decreases mitochondrial ROS production [158]. In house mouse, HK2 was shown to lose its binding to the MOM with age, whereas brain mitochondria of naked mole rats, long-living rodents, keep HK2 tightly bound to the MOM [158].

Fig. 5.

Mechanism of mild depolarization of mitochondrial inner membrane, observed in the brains and several other tissues of naked mole rats, long-living rodents [158]. Excess ATP synthesized in oxidative phosphorylation blocks mitochondrial respiratory chain and promotes hyperpolarization state on the mitochondrial inner membrane that, in turn, favor generation of ROS. Quick consumption of synthesized ATP by bound hexokinase prevents an ATP build-up and intensification of ROS generation. Abbreviations: G6P - glucose-6-phosphate, HK2 - hexokinase 2, VDAC - voltage-dependent anion channel, also known as porin, ANT - adenine nucleotide translocase.

Above-mentioned age-dependent metabolic changes in neurons observed during natural aging, may appear a compensatory response. We observe the metabolic repurposing from glycolysis to the PPP, that starts in middle age. As we already mentioned, the latter pathway is a source of important metabolites such as NADPH and ribose-5-phosphate. In turn, NADPH is used for the reduction of numerous antioxidants and related molecules, in particular, glutathione, thioredoxins, and glutaredoxins [16,142,159]. Additionally, NADPH is used for the biosynthesis of fatty acids in lipids which are oxidized by ROS and likely need to be replaced. Ribose-5-phosphate serves a precursor for the synthesis of nucleotides required for DNA replication and repair [160]. A note in proof for the increased nucleotide synthesis in brains of aged mice is provided by metabolomic studies [161]. As above mentioned, mitochondrial respiratory chain is one of the main generators of ROS. Therefore, pausing ATP production by the respiratory chain may provide a transient benefit for cell survival due to supposedly decreased ROS generation [162]. Again, metabolomic studies show an increase in the levels of adenosine monophosphate (AMP) accompanied by simultaneous decrease in the ATP levels [161]. The increase in AMP levels may also play a compensatory role since AMP activates AMP-activated protein kinase (AMPK) which, in turn, promotes catabolic processes such as glycolysis, TCA, as well as autophagy giving a pro-survival effect. However, the level of AMP is also controlled by conversion of AMP to IMP by AMP-deaminase [163]. Activation of expression of some proteins associated with lysosomes and lysosomal protein degradation, such as cathepsins and glial fibrillary acidic protein, was confirmed earlier by microarray studies [132].

10. Conclusion and perspectives

Structural and functional heterogeneity of the brain cells create a number of difficulties in the investigation of this organ. Cell culture and modern microscopy techniques helped to shed light on specific metabolic interactions between neurons and glial cells such as astrocytes and microglia. In most cases studies are focused on the operation of neurons as central, whereas a role of microglia is mainly seen as a supporting one. Recently, the active involvement of microglia and astrocytes in neuron functioning has been recognized, particularly in formation, maintaining and removing of neuronal synaptic contacts [38,51,57,164]. However, role of neurons in functioning of microglia is mainly out of focus and is waiting for studies.

In most cases, aging is accepted as a slow accumulation of diverse dangerous products of living processes. These products cannot be eliminated from the organism. Usually, such accumulated components result from non-enzymatic reactions between different cellular components, particularly proteins, lipids, carbohydrates, and nucleic acids, with reactive species of oxygen, nitrogen, carbon, etc. One group of such end products of living processes (EPLP) is called lipofuscins and was found as cellular fluorescent inclusions. Formation of such EPLP is associated frequently with the imperfect operation of antioxidant system that leads to escaping of some amount of ROS from elimination and interaction with cellular constituents. Other source of EPLP is related to slow interaction between carbohydrates and intermediates of their catabolism as well as products of lipid catabolism called reactive carbonyls species [151,165,166]. One more group of such reactive species called reactive nitrogen species, which was above mentioned as nitric oxide and peroxynitrite, is also involved in formation of EPLP. The latter are supposed to be responsible to some extent with intensification of oxidative stress during aging of organisms [3,111,131]. In our recent works with mouse brain, we have confirmed that older organisms have higher levels of markers of oxidative stress that reflects its intensification during lifespan [20,21]. It is worth noting, that such changes were found in middle age, whereas old mice had virtually the same levels of the markers. Therefore, the question was: what happens up to middle age that partially prevents further activation of oxidative stress? We turned our attention on energy-supplying processes and found that in comparison with young animals, activities of key glycolytic enzymes namely PFK and PK in the brains of middle-aged mice were substantially lower and at the same time the activity of key enzyme of the PPP – G6PDH – was substantially higher in middle-aged than in young mice. In old animals all studied characteristics virtually corresponded to ones in the brains of middle-aged mice. Such data were interpreted as potential mechanism resulting in redistribution of fluxes of intermediates of glucose catabolism from glycolysis to the PPP in the favor of the latter. That corresponds well to earlier information of such redistribution of fluxes in aged brains [3]. There is some reason in such redistribution because NADPH produced in the PPP can be used by antioxidant systems to combat ROS and prevent further activation of oxidative stress [129].

The presented above picture of reorganization of the brain homeostasis in the middle age is mainly descriptive and only slightly shed light on the molecular mechanisms responsible for redistribution of intermediates of glucose catabolism between glycolysis and the PPP. Further works should disclose real mechanisms underlying such changes. Among potential mechanisms of downregulation and upregulation of the activities of key glycolytic and the PPP enzymes is regulation of their expression at transcriptional and/or translational levels. Posttranslational changes, particularly inactivation of the enzymes due to their modification by ROS, also cannot be excluded. Our studies and those of other authors suggest that ROS can be the main inducers of changes in brain energy metabolism. Constant production of ROS in the brain leads to cumulative oxidative damages in biomolecules, especially glycolytic enzymes may be damaged. As a result, glucose will be used more in other pathways – in particular, in the pentose phosphate pathway for the synthesis of NADPH, which also gives the brain an advantage as it will enhance antioxidant protection. Our studies in mice indicate that dramatic switches in brain glucose metabolism and redox processes occur in middle age, and then changes occur very slowly. This may have important implications for initiating preventive strategies to slow down brain aging in humans. These and other suggestions should be carefully inspected that may give a key not only to understanding of these processes but also open avenue for affecting them and finally influencing aging process in a desired manner.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data Availability

• No data was used for the research described in the article.