Acetyl-CoA is a direct precursor substrate for the tricarboxylic acid (TCA) cycle coupled with energy production in the respiratory chain. In the brain, the largest fraction of acetyl-CoA is synthesised in mitochondria by the pyruvate dehydrogenase complex (PDHC) from pyruvate originating from glycolytic metabolism of glucose or from lactate oxidation by lactate dehydrogenase 1 (EC 1.1.1.1, LDH 1). The acetoacetate/β-hydroxybutyrate (AcAc/BHB), acetate, fatty acids and branched chain amino acids may provide smaller amounts of this metabolite in a concentration-dependent manner through β-ketothiolase (EC 2.3.1.16), acetyl-CoA synthase (ACCS, EC 6.2.1.1), beta-oxidation and branched ketoacids dehydrogenase pathways, respectively ^[1][2][1,2]. However, unlike peripheral tissues, the brain lacks significant metabolic flexibility, utilising plasma glucose as an obligatory, principal energy substrate. Non-glucose substrates cannot compensate for significant limitations in glucose supply apart from AcAc/BHB in starvation ^[1][2][3][1,2,3] (Figure 1). Nevertheless, irrespective of the precursor, acetyl-CoA is a merging endpoint metabolite, which enters the TCA cycle through a citrate synthase reaction. In resting conditions, the brain utilises 10 times more glucose and oxygen per weight than peripheral tissues; this is due to the continuous generation of action potentials in neurons in the range of 5–50 Hz. Moreover, restoration of neuronal membrane resting potential during each depolarisation/repolarisation cycle requires large amounts of energy. Hence, human brain neurons that make up 10% of all brain cells utilise 60–80% of glucose and oxygen. On the other hand, glial cells, which constitute 80–90% of human brain cells, produce 10% of the energy pool but export significant amounts of lactate to support neuronal energy metabolism. This is due to the prevalence of glycolysis over oxidative metabolism in astroglia and oligodendroglia ^[4][5][6][4,5,6]. Recent immunolabelling studies revealed that SOX9, an astrocyte-specific nuclear marker, is co-expressed with GLUT1 in astrocytes, constituting 10–20% of brain cells. However, they may be upregulated in several brain pathologies, including strokes, mini strokes and amyotrophic lateral sclerosis [7]. Moreover, physiological stimuli, such as feeding young mice with a high-fat diet for one month, increased astroglia size, branching and metabolism along with improving behaviour in animal behavioural tests [8].

Acetyl-CoA is also the substrate for hundreds of acetylation reactions catalysed by diverse acetyltransferases of different specificity, which can be found in each cellular sub-compartment. Products of acetylation include N-acetyl-aspartate and other acetylated amino acids and amines, such as acetylcholine, diverse acetylated proteins, carbohydrates and lipids, located in different cellular and subcellular compartments of the brain in concentrations varying from 10⁻² to 10⁻⁹ mol/L. Acetyl-CoA is also the key substrate for the synthesis of structural and metabolic pools of fatty acids, cholesterol, phospholipids and other lipids. They are indispensable for neuronal myelinisation, cell growth, remodelling and maintenance of cellular membrane integrity ^[9][10][9,10]. Acetyl-CoA itself may also directly regulate rates of diverse acetylating pathways in a concentration-dependent manner. It also reciprocally affects its availability in various subcellular and cellular compartments of the brain ^[11][12][11,12] (Table 1). Thereby, acetyl-CoA should be considered as a branching point between the catabolic energy-providing TCA cycle and multiple acetyltransferases synthesising acetylated derivatives. It is not only the metabolic substrate, but it also plays the second messenger role and takes part in phenotypic and genotypic modifications connected with basic functions of the brain, as well as with its maturation and aging ^{[13][14][15][16]}[13,14,15,16] (Figure 1).

However, in the past century, reports on direct assessments of acetyl-CoA in the brain have been scarce due to complicated assay methods ^[17][18][19][17,18,19]. Recently, several publications described and discussed regulatory mechanisms of intercellular metabolic fluxes of glucose and other acetyl-CoA precursors, as well as their role in prevention against neurotoxic insults and exerting neuroprotective effects in humans and animal models of diverse brain pathologies ^[5][6][14][5,6,14]. The apparent diversity of acetyl-CoA metabolic pathways in the brain results from the existence of several distinct neuronal neurotransmitter systems, as well as astroglial, oligodendroglial and microglial cells and their specific functional status in a given region. The intracellular acetyl-CoA is unevenly distributed among mitochondrial, cytoplasmic, endoplasmic reticulum, nuclear and axonal sub-compartments ^[6][14][6,14]. This is due to the relative impermeability of intracellular membranes including specific carriers for this metabolite ^[14][15][14,15]. Alterations of the acetyl-CoA level in mitochondrial and cytoplasmic compartments have been shown to regulate viability and cholinergic neurotransmission of cholinergic septal SN56 cells. Cholinergic neurons of the basal forebrain are responsible for multiple cognitive functions, including working, episodic and spatial memory, learning, attention, behavioural phenomena and sensory information assuring selectivity and precision ^{[6][13][14][20]}[6,13,14,20] (Figure 2).

Glucose is a principal energy source that fuels 95% of ATP synthesis in the brain. Its transport from blood plasma through the blood–brain barrier into brain extracellular space is carried out by the specific, high-capacity, average-affinity 55 kD GLUT 1 transporter located on the outer side of vascular endothelium. Its constant affinity for glucose is equal to about 8 mmol/L. It achieves an inward transport rate dependent on systemic glucose concentration, diurnal physiologic variations of which may be in the range of 3.5–10 mmol/L. The density of 55 kD GLUT1 on the blood–brain barrier (BBB) capillaries is inversely regulated by glucose concentration in the plasma. Chronic hyperglycaemia results in an adaptative decrease in 55 kD GLUT1 density on the BBB, which reduces glucose transport into the brain ^[21][22][23][21,22,23]. In turn, in chronic hypoglycaemia, hypoxia and several acquired and congenital metabolic diseases, the density of 55 kD GLUT1 increases, adjusting in part the rate of inward transport of glucose to these conditions ^[24][25][24,25]. Upregulation of GLUT1 is mediated by the phosphatidylinositol-3-kinase, AMP-activated protein kinase, cAMP response element–binding protein and hypoxia inducible factor pathways. On the other hand, the mitogen-activated protein kinase pathway downregulates GLUT 1 and 3 expression [25]. Such mechanisms tend to reduce concentration-dependent fluctuations in the rate of glucose transport into the brain. In fact, some of those patients display good tolerance to hypoglycaemia (Figure 1). Neurons take up glucose from extracellular brain space mainly by the high-affinity GLUT3 transporter with Km about 2.8 mM, which may secure its appropriate supply in moderate hypoglycaemia ^[21][23][21,23]. Pyruvate derived from glycolysis or from oxidation of exogenous or endogenous lactate may serve as a direct precursor of neuronal acetyl-CoA. Lactate is released from astroglial or oligodendroglial cells by MCT4 transporters with low affinity (Km 22–28 mM) for lactate and transported into neuronal cell body axons and presynaptic terminals, respectively. Neurons express high-affinity MCT2 transporter with Km for lactate and pyruvate equal to 0.5–0.75 and 0.08 mmol/L, respectively; this promotes the direction of lactate/pyruvate transport toward neuronal cells ^[23][26][27][23,26,27]. Such direction of metabolic flow is also facilitated by the fact that LDH-1 isoform, which is primarily expressed in neurons, oxidises lactate to pyruvate, whereas the LDH-5 isoform expressed in astrocytes promotes the reduction of pyruvate to lactate. Despite their distinct distribution, LDH-1 and LDH-5 do not directly modulate the lactate flow between neurons and astrocytes [6]. Nevertheless, LDH5 would facilitate lactate accumulation and its subsequent release from astroglia, whereas LDH1 would support the metabolism of lactate directly towards pyruvate and its oxidative decarboxylation (Figure 1). However, in vivo lactate cannot replace glucose entirely as an energy precursor either in physiological or in pathological conditions. Lactate at high-physiologic 1 mmol/L concentration can substitute only about 10% of glucose. However, during lactic acidosis evoked by intensive exercises, hypoxia, diabetes, chronic obstructive lung disease, disseminated cancer or inherited metabolic diseases, blood lactate may reach 10 mmol/L and higher concentrations and replace up to 25% glucose in acetyl-CoA and energy production [28]. On the other hand, cultured primary neurons and neuronal stem cells survived in glucose-free medium in the presence of lactate as the only energy substrate [29]. Moreover, SN56 cholinergic neuronal cells utilising pyruvate with malate as sole energy substrates retained stable levels of acetyl-CoA and structural integrity [14]. Various types of neurons and glial cells may respond differentially to the same neurotoxic conditions. For instance, in chronically hypoxic BV2 microglial cells, the acetyl-CoA level was three times higher than in normoxic ones [30]. This apparent inconsistency was explained by a hypoxia-induced increase in mRNA and protein levels of hexokinase 2 (HK2, EC 2.7.1.1), yielding activation of the glycolytic pathway. This led to the rise of pyruvate synthesis, which through the PDHC provided more acetyl-CoA for acetylations of nuclear histones 3 and 4 ^[31][32][33][31,32,33]. Acetylated histones induced the inflammatory phenotype in BV2 by elevating proinflammatory gene expression followed by augmentation of CD11 and IL-1β levels. Inhibition of HK2 or PDHC activities by their specific inhibitors, lonidamine or 3-bromopyruvate, decreased acetyl-CoA levels and alleviated proinflammatory responses of cultured BV-2 microglial cells. Similar effects were observed after knockdown of HK2 with specific siRNA, but not after suppression of HK1 and HK3. This proves that only the increase in HK2 expression was responsible for the over-activation of BV2 cells. In addition, in in vivo experiments, suppressed activity of microglial HK2 by pretreatment with lonidamine reduced the size of ischemic injury in the rat brain in the middle carotid artery occlusion model (MCAO) [29]. MCAO followed by 40 min of hypoxia in 9-day-old neonatal mice after 24 h brought about a decrease in PDHC activity but no changes in the acetyl-CoA level in crude mitochondrial fraction. Animals pretreated with dichloroacetate (DCA), a pyruvate dehydrogenase kinase (PDK, EC 2.7.11.2) inhibitor, showed increased activity of the PDHC and levels of acetyl-CoA above the control values, as well as reduced cell apoptosis [34] (Table 1). MCAO-induced failure of energy metabolism in the brain is claimed to be overcome with application of Shengui Sanhseng San extract [35] (Table 1). This preparation has been used for over 300 years by Traditional Chinese medicine to treat stroke. It consists of a mixture of rhizome and roots of three plants. Its application for 7 days to MCAO animals increased, in a dose-dependent manner, the density of GLUT1 and 3 and the levels of pyruvate, citrate, acetyl-CoA and ATP and decreased phosphorylation of E1 subunit PDHC in the post-ischemic region of the brain [35]. A pleiotropic, neuroprotective effect of resveratrol preconditioning was also reported. It increased ischemic tolerance in in vivo and in vitro models ^[36][37][36,37]. Neuronal–astrocytic co-cultures obtained from 14 d resveratrol-treated mice displayed increased expression of mitochondrial pyruvate carriers and citrate synthase levels, yielding an increase in citrate and ATP synthesis and a delay of excitotoxic injury during oxygen–glucose deprivation [36]. Such conditions increased the passage of citrate into the nucleus, which augmented its conversion to acetyl-CoA by nuclear ACLY and activated acetylation of histones H3K9ac and H4K16ac by HAT [36]. It should be emphasised that, although the presented data strongly suggested the involvement of acetyl-CoA in resveratrol-mediated neuroprotection, no direct findings concerning its mechanisms have been presented ^[36][37][36,37] (Figure 1). On the other hand, modest inhibition of PDHC activity in cultured N9 microglial, C6 astroglial and primary glial cells by 0.15 mmol/L Zn or NO excess brought about no significant decreases in their acetyl-CoA/ATP levels or their viability ^[11][38][39][11,38,39]. On the contrary, under similar culture conditions, Zn caused over 50% suppression of PDHC activity and acetyl-CoA content in differentiated cholinergic SN56 neuronal cells (Table 1). These alterations were accompanied by a significant loss in rates of ACh synthesis/release and neuronal viability ^[11][39][40][11,39,40]. The presented data indicate that adaptive overexpression of glycolytic pathway in glial cells, after hypoxic insult, may overcome inhibition of acetyl-CoA synthesis caused by suppression of PDHC activity due to provision of additional amounts of pyruvate [30]. Another cause of relative resistance of neuroglia to neurotoxic conditions may be their lower energy demands for maintenance of plasma membrane potential and non-utilisation of acetyl-CoA for neurotransmitter or NAA synthesis ^[11][14][30][11,14,30]. On the other hand, energy requirements for neurons are much higher than those for glial cells. Therefore, inhibition of PDHC activity exerted deeper suppressive effects on acetyl-CoA content and viability in neuronal than in glial cells ^[38][39][38,39] (Table 1). There are data showing that disturbances in brain Zn homeostasis may be the primary cause of neurodegeneration. Open-head traumatic injury of the brain cortex caused immediate hyperglycaemia lasting up to 3 h, followed by severe hypoglycaemia in both male and female rats. In the ninth post-trauma hour, loss of respiratory control was observed in isolated brain mitochondria, along with suppression in acetyl-CoA and ATP levels in the peri-contusional ipsilateral cortex of male rats [41] (Table 1). These changes could be evoked by excitotoxic effects of Zn released in excess from impaired presynaptic terminals ^[42][43][42,43]; then, Zn was taken up by postsynaptic neurons causing the inhibition of the PDHC and several enzymes of the TCA cycle, ultimately leading to their death ^[39][44][45][39,44,45]. These pathologic alterations could be corrected by early i.v. infusion of lactate or a delayed one of BHB that maintained their stable blood concentrations of 1.2 and 2.0 mmol/L, respectively. In female rats, contusion brought about inhibited respiration rates but no changes in acetyl-CoA and ATP levels. Early infusion of 2 mol/L BHB increased acetyl-CoA and ATP over control levels, whereas 100 mmol/L lactate was without effect. On the contrary, early infusion of BHB appeared to be harmful, resulting in severe decreases in acetyl-CoA and ATP levels in peri-contusional tissue. This indicates that, in traumatic brain injury, therapeutic i.v. applications of BHB or lactate as complementary to glucose direct precursors of acetyl-CoA should consider the post-trauma time and sex of the patient to avoid negative side effects of such treatment [41]. However, there is no rational explanation for post-trauma time and sex-linked differences in the beneficial or harmful effects of exogenous lactate or HB on acetyl-CoA-mediated post-traumatic recuperation of the brain. Nevertheless, irrespective of the particular pathomechanism, the beneficial effects of lactate or BHB were accompanied by an increase in the whole tissue acetyl-CoA level, indicating a key role of this intermediate in brain healing [40]. Such a thesis is supported by the results of clinical trials that revealed beneficial effects of infusion hyperosmic sodium lactate, being a direct acetyl-CoA precursor. It bypasses glycolysis as the ATP consuming pathway and alleviates reperfusion injury in patients after focal cerebral ischemia [46]. The positive influences of hypertonic lactate infusion in brains of TBI patients or those with acute cardiac failure may be extended beyond its role as an alternative energy precursor, as it is also an anti-oedematous agent, scavenger of free radicals, Zn/Ca chelator and suppressor of reperfusion-evoked glutamate/Zn/NO excitotoxicity ^[46][47][46,47]. For obvious reasons, the effects of lactate/HB on the acetyl-CoA status in injured brains were not investigated in humans. Cardiac arrest is a prevalent cause of death worldwide. Cardiopulmonary resuscitation has improved survival, but many patients die soon after due to anoxic brain injury and cardiac instability [48]. In rats subjected to 6 min anoxia and then resuscitated, intraperitoneal injection of dichloroacetate (DCA) caused a two-fold increase in the survival rate and alleviated neurological deficits during the 72 h post-resuscitation period. DCA decreased levels of proinflammatory IL-1β and TNF-α and blood lactate concentration, along with partial restoring decreased PDHC activity and ATP and pyruvate levels in the hippocampus and brain cortex [48]. The inhibition of the PDHC may be caused by an increase in intraneuronal Ca levels due to its excessive influx through voltage-gated calcium channels and NMDA receptors in depolarised plasma membranes. Ca excess activated PDK, yielding inhibitory phosphorylation of the E1 PDHC subunit ^[49][50][49,50]. These data suggest that the beneficial effects of DCA resulted from the inhibition of PDK, yielding the dephosphorylation of the E1 subunit and an increase in PDHC activity, followed by increases in pyruvate oxidation and acetyl-CoA and ATP levels ^[48][49][50][48,49,50]. Moreover, in mice, therapeutic hypothermia or DCA application inhibited the PDK and reactivated the PDHC, improving outcome after cardiac arrest [51]. Cardiac arrest in mice lasting 8 min followed by cardiopulmonary resuscitation resulted in a rapid and deepening-with-time decrease in thiamine diphosphate (TDP) and ATP levels in the brain cortex, resulting in a high rate of animal mortality. This was due to phosphorylation-induced inhibition of PDHC activity, yielding a deficit of acetyl-CoA in neurons ^[46][52][46,52]. Daily i.v. supplementation of thiamine increased TDP and ATP levels in brains, decreasing the rate of animal mortality [52]. Moreover, in humans, cardiac arrest caused a marked acute decrease in PDHC activity in mononuclear blood cells, which was partially restored within 72 h after resuscitation. In addition, 30 min of severe hypoglycaemia in rats after 7 days brought about a several-fold increase in PDK2 and inhibition of PDHC activities in the brain, causing extensive neuronal death, along with activation of astroglia and microglia [53]. These pathologic alterations were alleviated in part by injections of DCA. In summary, these data prove that the PDHC is a common target for several neurodegenerative signals. Therefore, maintenance of the activity of this complex should be considered as the potential goal for neuroprotective interventions. Moreover, these data suggest that the inhibition of PDK or an increase in thiamine supplementation may benefit neuroprotective treatment of patients after cardiac arrest ^[51][52][51,52]. Other in vivo experiments revealed mechanisms that may partially compensate for shortages of acetyl-CoA evoked by deficits of the PDHC. In brain-specific heterozygotic Pdha1 knockdown mice, PDHC activity was reduced by 68%, but the acetyl-CoA level was not significantly decreased against wild-type controls [54] (Table 1). Such a discrepancy may be explained by the compensatory activation of the ACSS1 pathway, which increased acetate incorporation into acetyl-CoA in animal brains. Systemic administration of acetate to PDHC-deficient mice stimulated metabolic flux through the TCA cycle and normalised glutamatergic neurotransmission, yielding suppression of gamma oscillations and epileptiform discharges ^[54][55][54,55]. Beneficial effects of dietary supplementation with glyceryl triacetate—as a source of acetate—were also observed in an experimental autoimmune encephalomyelitis (EAE) mouse model of multiple sclerosis. Such treatment prevented the loss of ethanolamine, phosphatidyl choline and cholesterol in the myelin of EAE mice compared to EAE controls treated with water [56]. These data suggest that exogenous acetate or its donors may be used as a complementary precursor of acetyl-CoA, bypassing the PDHC step, to increase lipid deposition in oligodendrocytes and neurons impaired by demyelinating diseases [56] (see chapter 6 for details). Such a conclusion is supported by the finding that, in mice with specifically deleted Pdha1 gene, Schwann’s cells and brain oligodendrocytes retain capacity for myelinisation. This may be caused by the existence of compensatory overexpression of the ACSS1 pathway providing acetyl-CoA, which bypasses the PDHC step in oligodendrocytes ^[54][57][54,57]. On the other hand, animals with Pdha1 deletion in all brain cells displayed reduced fibre density and signs of axonal degeneration. This suggests that acetyl units in PDHC-deficient oligodendrocytes are provided by adjacent PDHC-competent astroglial and neuronal cells ^[54][57][54,57]. TBI suppressed PDHC expression in the peri-contusional area, and the rate of this decline depended on the potency of the impact. Mild or severe TBI brought about differential effects on genes expression, protein levels and the activities of several enzymes linked with energy production [58] (Table 1). Mild TBI within 5 days post-trauma caused increases in gene expression of catalytic E1, E2 and E3 PDHC subunits, along with decreases in PDK and gradual, delayed increases in PDP expression. Such a profile of PDHC subunits would be compatible with a stable level of acetyl-CoA in a mildly insulted brain. On the other hand, severe TBI did not affect expression of PDHC catalytic subunits but strongly suppressed PDP (down to 5% of controls) and elevated PDK gene expression. Such a pattern promoted inhibition of the PDHC E1 subunit due to PDK-dependent inhibitory phosphorylation [50]. Accordingly, marked decreases in acetyl-CoA/CoA-SH levels were found in cases of severe TBI [58]. Hence, post-traumatic differential changes in brain acetyl-CoA levels in severe vs. mild TBI result from variations in phosphorylation/dephosphorylation levels yielding the inhibition or activation of the PDHC, respectively. A higher level of acetyl-CoA may be a predictive marker for positive outcome following TBI [58]. Inhibition of energy metabolism is an early sign of mitochondrial dysfunction in AD and other neurodegenerative diseases; it is caused by decreases of activities or expression of the PDHC, KDHC and some other mitochondrial enzymes of the TCA cycle evoked by accumulating Aβ and hyperphosphorylated tau ^[59][60][59,60]. Aβ1-42 in sub-micromolar concentrations was found to inhibit activities of the PDHC and KDHC in vitro in synaptosomes isolated from the brain of WT rats or in synaptosomes from 2756 Tg AD mice and in clonal neuronal cells ^[14][61][62][14,61,62]. However, contradictory results were presented by Gandbhir and Sundaram [63]. They showed that a relatively high 0.004 mM concentration of pre-aggregated Aβ1-42 markedly increased levels of PDHC and KDHC proteins, causing tau hyperphosphorylation and impairment of the SH-SY5Y cholinergic cell line. These effects were partially reversed by AmyTrap, a homo-tetrameric peptide synthesised from D-amino acids, which could remove Aβ1-42 from neuronal cells [63]. However, it remains unknown as to how Aβ-induced increases of PDHC and KDHC expression resulting in elevations of acetyl-CoA and ATP levels could exert neurotoxic effects. These data contradict the well-documented view that high levels of ATP and acetyl-CoA are markers of the wellbeing of neurons and other cells ^{[13][14][15][53][64]}[13,14,15,53,64]. For an explanation of this discrepancy, determinations of in situ enzyme activities, as well as estimations of ATP and acetyl-CoA levels in different culture conditions, are necessary. Such information is indispensable in light of the claimed therapeutic potential of the AmyTrap compound [63].