Glucose is the preferred fuel in the brain and is primarily catabolized via aerobic glycolysis (Dienel, 2018). During oxidative glycolysis, glucose first enters the cell and is irreversibly phosphorylated by hexokinase to glucose-6-phosphate (G-6-P). The G-6-P metabolite can then continue down the glycolytic pathway, shuttle into the pentose phosphate pathway (PPP), or be stored as glycogen (Figure 1). Both the glycolytic and PPP pathways are involved in the production of pyruvate, a critical metabolite that is converted to acetyl coenzyme A (AcCoA) by the pyruvate dehydrogenase (PDH) enzyme complex. AcCoA enters the citric acid cycle (TCA) in the mitochondria to generate the electron carrier metabolites NADH and FADH₂ upon cycle completion. Both NADH and FADH₂ enter the electron transport chain to produce an estimated net 30 ATP molecules (net amount is lower than theoretical estimates after accounting for proton leak) via oxidative phosphorylation (OXPHOS). Notably, the amount of ATP produced by OXPHOS per molecule of glucose is substantially greater than the net 2 ATP molecules generated solely through glycolysis (Dienel, 2018). Remarkably, OLs have been shown to prefer glycolysis over OXPHOS in spite of their high energetic demands. A recent study focused on characterizing the distinct bioenergetic properties of both human-derived adult OPCs and OLs, demonstrated that under favorable and nutrient-rich conditions, OPCs and OLs both rely significantly more on glycolysis than OXPHOS to generate ATP (Rone et al., 2016). Additionally, when subjected to stress conditions like those observed in the demyelinating disease, multiple sclerosis (MS), OLs begin retracting their processes to destabilize the compact axonal myelin sheath, which is thought to limit cell death by entering a less-energy intensive, low metabolic state (glycolysis). This response is perhaps favorable in the short-term, preserving existing OLs should they experience conditions conducive for a metabolic rebound. In contrast, OPCs are not capable of sufficiently reducing their metabolic rate as these cells require substantially more ATP. As a result, OPCs begin to rapidly die under stress. Surprisingly, developing rodent OPCs have been found to be more metabolically active compared to rodent OLs, capable of elevating both OXPHOS and glycolysis, with both rodent OPCs and OLs producing more ATP than their human counterparts (Rone et al., 2016).

A central role for glycolysis in OLs was firmly established by studying mice with an inducible knockout for the Cox10 gene in immature OLs from birth. Cox10 encodes for cytochrome c oxidase (COX—also known as complex IV), which is the terminal electron acceptor in the electron transport chain of the mitochondria that helps to establish the proton gradient needed for ATP generation via OXPHOS. Despite confirmed mitochondrial dysfunction, mutant mice (Cox10^flox/flox*Cnp1^Cre/+) displayed no signs of demyelination, oligodendroglial pathology, or white matter abnormalities, even at 9 months of age (Fünfschilling et al., 2012). Lactate levels, however, were found to be significantly increased. Lactate can be generated from the reaction between pyruvate and NADH, catalyzed by lactate dehydrogenase (LDH). Given that the majority of lactate comes from glycolytic pyruvate and that it is secreted from the cell (via monocarboxylate transporters, MCTs), lactate is often used as a readout for glycolytic activity (Rogatzki et al., 2015). In other words, OLs were demonstrated to survive and adequately function by using glycolysis to generate their needed energy, although it should be noted that the energy required for mature OL maintenance is lower than developing OLs. Nevertheless, even with lower overall ATP yields, the kinetics of glycolysis is much faster than complete glucose oxidation (glycolysis + OXPHOS; Dienel, 2018). Additionally, the increase in lactate production from enhanced glycolysis can benefit the underlying ensheathed axons since myelin wrapping enables the formation of “myelinic channels” for which lactate and other trophic factors can be shuttled to axons to maintain axonal health (Simons and Nave, 2015). Thus, the high energy needs of neurons during action potential signaling, calculated to be 4.1 × 10⁵ ATP/s for an axon diameter of 0.76 μm and 1.8 × 10⁶ ATP/s for a larger 1.26 μm axon diameter firing at physiological rates of 3 Hz and 8 Hz, respectively (Harris and Attwell, 2012), can be supported by this influx of lactate, which can be readily converted to pyruvate via LDH, followed by conversion to AcCoA by pyruvate dehydrogenase. AcCoA can then enter the TCA cycle to ultimately produce the byproducts NADH and FADH₂ that will feed into complex 1 and complex 2 of the electron transport chain, respectively, fueling OXPHOS-linked ATP production, ultimately offering a faster way for neurons to produce ATP that bypasses the upstream glycolytic pathway altogether (Figure 1). The ability of lactate-derived ATP to meet energy demands in the brain is reminiscent of astrocytic glycogen-derived lactate that has been shown to help maintain ATP in an exhaustive-exercising brain model, thus protecting neurons from cell death (Matsui et al., 2017). Lactate’s role in OLs may go beyond supporting neurons as evidence suggests that OLs can also import this metabolite through MCT1 transporters, providing a way to support oligodendrocyte development and myelination when glucose concentrations are low (Rinholm et al., 2011). OLs also oxidize lactate at a greater rate than neurons and astrocytes do for lipid synthesis, which may contribute to the preference for glycolysis in these cells (Sánchez-Abarca et al., 2001).

However, it is important to note that other compounds regulate axonal metabolism through their actions within OLs. In particular, sirtuin 2 (SIRT2), a member of the sirtuin family of signaling proteins involved in metabolic regulation, has recently been shown to be present in OL exosomes that readily cross the periaxonal space and enter axons, where the release of SIRT2 (which is undetectable in neurons) acts on mitochondrial proteins to deacetylate them and increase axonal ATP levels, even in the presence of lactate and high glucose containing media (Chamberlain et al., 2021).

Disruption in OL glycolytic activity is also thought to contribute to many diseases, including Alzheimer’s disease (AD). Early AD is accompanied by white matter loss, OL death, and axonal degeneration in both human AD patients and AD animal models; these early events have all been linked to mitochondrial dysfunction. Further investigation into the role OL stress plays in AD neuropathology determined that dynamin-related protein-1 (Drp1) was activated, leading to mitochondrial fission and fragmentation, causing impaired bioenergetics and suppression of glycolytic activity in OLs via inhibition of the rate-limiting enzyme hexokinase-1. In turn, the NLRP3 inflammasome (a multimeric protein complex) was activated, which promotes proinflammatory programmed cell death. However, a genetic knockout of Drp1 restored glycolytic rates in mature OLs of AD mice, which prevented myelin loss, reduced axonal degeneration, and improved cognitive performance. Thus, this study provided strong evidence that correctly regulated OL glycolysis plays an important role in CNS disease (Zhang et al., 2020).

The PPP is a biosynthetic pathway (non-ATP producing) in cells that branches from glycolysis after glucose is metabolized to G-6-P by the enzyme hexokinase. Once committed, G-6-P enters the oxidative, irreversible state of the PPP pathway where in a series of stepwise reactions, it forms 6-phosphogluconate, ribulose-5-phosphate (Ru-5-P), and 2 molecules of NADPH. In the non-oxidative, reversible arm of the PPP, Ru-5-P is converted to ribose-5-phosphate (R-5-P) to provide the carbon backbone needed for nucleotide synthesis (Patra and Hay, 2014). Additionally, other PPP metabolites such as fructose-6-phosphate (F-6-P) and glyceraldehyde-3-phosphate (G-3-P) can shuttle back into the glycolytic pathway or, in the case of F-6-P, can be converted back into G-6-P and replenish oxidative PPP by generating more NADPH (Figure 1), which in turn can provide redox power for myelin lipid synthesis or bind with the enzyme catalase to enhance antioxidant activity within the cell (Kirkman et al., 1987; Berg et al., 2002; Kowalik et al., 2017).

The metabolic flux of specific metabolites can be analyzed within cells using 13-carbon (¹³C) labeling techniques. With this approach, Amaral et al. (2016) assessed [1,2-¹³C] glucose in differentiated OLs to better understand the metabolic pathways involved in glucose catabolism. They found significant glucose metabolism in OLs, with 10%–15% being metabolized via the PPP pathway, a rate similar to that seen in astrocytes. Additionally, they noted that the F-6-P and G-3-P derived from R-5-P of the PPP can enter the glycolytic pathway to produce pyruvate, which gets carboxylated to oxaloacetate (the last metabolite generated in the TCA that forms citrate after condensation with AcCoA) by the enzyme pyruvate carboxylase (PC; Berg et al., 2002; Amaral et al., 2016). The generation of oxaloacetate by pyruvate carboxylation allows alpha-ketoglutarate (α-KG, one of the intermediates in the TCA) to leave the cycle and enter the cytosol to form glutamate, or in the presence of glutamine synthetase (GS), make the amino acid glutamine. This finding has a profound impact on OL metabolic functions as this glutamate-glutamine cycle has previously only been described in astrocytes, where it aids neurons in synthesizing the most abundant excitatory and inhibitory neurotransmitters in the brain, glutamate, and GABA, respectively (Hertz, 2013). Although it is yet to be seen if glutamine release occurs in OLs, it is possible that myelinic channels formed during wrapping can shuttle glutamate/glutamine to neurons and assist in neuronal signaling and more specifically provide feedback for adaptive myelination. It is important to note that in addition to elevated PPP, Amaral et al. (2016) observed high OL mitochondrial activity (TCA activity) but the level of OXPHOS was not reported.

Similar to OLs, OPCs display high glucose catabolism in the PPP pathway. When labeling OPCs with ¹⁴Carbon glucose (¹⁴C-glucose) to assess the flux of glucose through different metabolic pathways in the postnatal rat brain, the PPP was reported to be two-fold higher than in type 1 astrocytes, 10-fold higher than in type 2 astrocytes, and at least four-fold higher than in neurons (Sánchez-Abarca et al., 2001). This is not surprising given that OPCs are highly proliferative cells, especially during development and in the subventricular zone of adult brains (Spitzer et al., 2019), and thus require ample nutrients for cell cycle progression. This requirement is most pronounced within the late G1/S phase of the cycle, where the non-oxidative branch of the PPP synthesizes R-5-P, the precursor for nucleotides adenine and guanine, to aid in DNA synthesis before mitosis and complete cell division can occur (Khedkar et al., 2016).

Inhibition of PPP in OLs with supra concentration of the G-6-P inhibitor dehydroepiandrosterone (DHEA) compromises NADPH supply and causes significant OL death and decreases in myelin. Similar results are also observed when treating both OPCs and OLs with the NADPH antimetabolite 6-aminonicotinamide (6AN). These findings were attributed to a disruption in NADPH-linked antioxidant activity and treatment with reduced glutathione (a naturally occurring antioxidant) was able to rescue the deficits observed, thereby, highlighting the importance of PPP activity and the many metabolites it produces in oligodendroglia (Kilanczyk et al., 2016).

Cholesterol is a critical component of the myelin sheath needed to preserve the complex arrangement of lipids and proteins of the myelin membrane. Along with GalC, cholesterol helps to form specialized domains along the membrane, the so-called “lipid rafts”, which contribute to protein trafficking and signaling, as well as myelin fluidity and stability (Decker, 2004; Saher et al., 2011; Montani, 2021). Because the blood-brain barrier prevents the uptake of circulating lipoprotein bound dietary cholesterol, cholesterol synthesis occurs in situ (Petrov et al., 2016). The rate of cholesterol synthesis in the CNS is at the highest during the first few weeks to months after birth, when OLs are engaged in active myelination. However, within the myelin membrane, there is long-term stability of cholesterol such that turnover occurs at the relatively slow pace of ~1 year in the young adult mouse brain and ~5 years in the human adult brain (Ando et al., 2003; Russell et al., 2009).

OL cholesterol synthesis is reliant on AcCoA generated from either glucose or ketone sources, although OLs show a preference for ketone bodies (Koper et al., 1981). AcCoA is formed from pyruvate (an end product of glycolysis) via pyruvate dehydrogenase or, in the case of ketones, the ketone body acetoacetate can be converted back into AcCoA via the enzyme β-ketoacyl-CoA. AcCoA then gets transformed into 3-hydroxyl-3-methylglutaryl-coenzyme A (HMG-CoA) and subsequently into mevalonate, which then goes on to produce lanosterol followed by a series of reactions that ultimately generates cholesterol. The intermediary reactions in this process are catalyzed by specific enzymes that require NADPH (a byproduct of the PPP) as a cofactor. Notably, the formation of mevalonate by HMG-CoA reductase is the irreversible, rate-limiting step for cholesterol biosynthesis (Petrov et al., 2016). HMG-CoA reductase is targeted by statin drugs to lower cholesterol production in the body. Given the widespread use of statins (e.g., simvastatin), several studies have used statins to assess the impact of cholesterol synthesis in OLs. Both OPCs and OLs that underwent prolonged treatment with simvastatin in vitro exhibited significant process retraction, increased cell death, and decreased mitochondrial activity whereas treatment with mevalonate or cholesterol itself was able to rescue the simvastatin-induced phenotype (Miron et al., 2007). Furthermore, severe hypomyelination has been observed in both white and gray matter tissue in mice engineered with OL-specific loss of Fdft1, the gene encoding squalene synthase, which catalyzes the first step of the cholesterol synthesis pathway. These Fdft1 conditional knockouts displayed ataxia, tremors, and premature death (Saher et al., 2005). These and other studies have concluded that cholesterol is essential for myelin membrane growth and function, as well as for the survival of OPCs and OLs, which may be possibly mediated through signaling proteins contained within membrane cholesterol-enriched lipid rafts. In the absence of cholesterol, signaling proteins can become decoupled from lipid rafts, suppressing necessary signal transduction that mediates survival and other cellular processes in oligodendroglia (Bryant et al., 2009).

Galactosylceramide (GalC) is synthesized from a sphingolipid ceramide molecule. The UDP-galactose: ceramide galactosyltransferase (CGT) enzyme transfers UDP-galactose (UDP: uridine diphosphate) to ceramide, forming GalC. Interestingly, in CGT global knockout mice, no GalC production was detected and myelin instead contained high levels of glucocerebroside. Despite this compensatory mechanism, significant decreases in the number of myelinated axon segments in the spinal cord were observed. Unexpectedly, CGT deficient mice demonstrated a 50% increase in mature OLs, revealing a possible negative regulatory role for GalC in OL differentiation. A reduction in conduction velocity and insulative capacity of the myelin sheath was also noted, which led to progressive hindlimb paralysis, tremors, and even premature death (Coetzee et al., 1996; Bansal et al., 1999; Marcus et al., 2000). Problems with myelin compaction were also reported with the appearance of outward facing paranodal loops, outfolding of the myelin membrane, and incorrect myelin compaction (Bosio et al., 1998; Ishibashi et al., 2002).

The molecular structure of GalC has provided useful insights into its function. GalC and its sulfated form, sulfatide, are the most abundant glycolipids in the myelin membrane, accounting for 23% and 4% respectively, of the total myelin lipid mass (Norton and Cammer, 1984). GalC, like other galactosphingolipids, have longer, more saturated fatty acids than phospholipids and can participate in intermolecular hydrogen bonding, allowing them to form ordered membrane domains (Dicko et al., 2003). In fact, GalC is believed to interact with other sulfur enriched glycolipids such as sulfatide to form “glycosynapses” during the wrapping of the myelin sheath, contributing to the stability of myelin. These glycosynapses can interact with axonal ligands, extracellular matrix proteins, or even with each other across the apposed membranes to facilitate the transmission of transmembrane signals (Hakomori, 2002; Boggs et al., 2010). Liposomes containing GalC and sulfatide were added to OLs in culture to mimic facing pairs of extracellular surfaces of myelin or interactions between OLs processes. These liposomes were able to interact with glycosynapses to cause a disruption in the microtubule network and a concentrated redistribution of GalC on the extracellular side and MBP on the cytosolic side. However, as there is no clear role for the seGalC-MBP rich domains in OL signaling, future work is needed to understand their purpose (Boggs and Wang, 2001).

In the myelin membrane, the most abundant phospholipid belongs to the subclass plasmalogen. Plasmalogens are a type of glycerophospholipid that exhibit a vinyl-ether group on the first carbon of the glycerol molecule, with the predominant plasmalogen in OLs being ethanolamine plasmalogen (EP). EP synthesis begins in the peroxisome and is completed in the endoplasmic reticulum (Messias et al., 2018; Montani, 2021). Peroxisomal dihydroxyacetone phosphate acyltransferase (DHAPAT) enzyme is necessary for plasmalogen synthesis and loss of this enzyme results in ethanolamine deficiency, dysmyelination of the cerebellum, reduction in total MBP protein in both cortex and cerebellum, paranode disorganization, as well as a ~40% decrease in myelinated fibers of the EP deficient corpus callosum (Teigler et al., 2009). Future studies are still needed however to fully decipher the precise role EP has in OLs.

Understanding how both intrinsic and extrinsic signaling events can affect oligodendroglial metabolism will be key to the development of therapeutic agents to either enhance metabolic control of oligodendroglial functions or counter metabolic dysregulation. Several recent studies have shed light on some of the signaling pathways impacting OL metabolism. For example, netrin-1, a secreted protein that binds to the transmembrane receptor DCC (deleted in colorectal cancer), has recently been implicated in regulating OL mitochondrial dynamics and bioenergetics. Proteomic analysis and confocal imaging of netrin-coated beads in association with OLs revealed binding to paranodal enriched proteins DCC and neurofascin-1 (NF155). Similarly, binding to the dynein-1 heavy chain and kinesin light chain-1 (two major mitochondrial microtubule motor proteins) was observed, indicating OL netrin-1 may be associated with mitochondria trafficking. Indeed, super-resolution microscopy and transmission electron microscopy confirmed that netrin-1 was involved in mitochondria movement to distal regions of OL processes and paranodes. Furthermore, mitochondria elongation was enhanced by netrin-1 induced src family kinase (SFK) activation, and downstream inactivation of rho-associated protein kinase (ROCK). Mitochondrial elongation results from the fusion of multiple mitochondria organelles and as such can alter the metabolic dynamics of the cell. Live cell metabolic profiling showed that prolonged netrin-1 signaling (through SFK activation and ROCK inhibition) hyperpolarizes the mitochondrial inner membrane and increases glycolysis but not OXPHOS in primary OLs (Nakamura et al., 2021). Whether these netrin-1/SFK induced changes in OL metabolism can be beneficial in a demyelination paradigm with dysfunctional OLs remains to be determined but nevertheless, these findings present new pathways to pursue in the quest for next-generation oligodendroglia therapeutics.

OLs are known to provide trophic and metabolic support to the axons they ensheath, yet variations in synaptic strengths and neuronal firing rates can greatly alter the metabolic demand of axons. Thus, a key unanswered question is how OL adjust their ability to metabolically support axons in response to environmental cues (Mount and Monje, 2017; Bonetto et al., 2020). In a recent study, researchers were able to demonstrate that OLs “sense” neuronal activity through OL NMDA receptors, which can respond to the neurotransmitter glutamate (released upon neuronal firing) and then adjust the OLs ability to regulate axonal energy metabolism (Saab et al., 2016). For instance, stimulation of NMDA receptors with NMDA-acid and glycine (NMDA-A/Gly) in OL cultures triggered an increase in GLUT1 expression and trafficking to the membrane, leading to increased intracellular glucose and lactate concentrations as well as high lactate release. Selective deletion of the NR1 subunit [the glycine binding subunit critical to NMDA receptor function (Catts et al., 2015)] from OLs (NR1 cKO) conversely resulted in a strong reduction of GLUT1 transporters in OLs (e.g., an ~80% reduction in the optic nerve). GLUT1 deficiency in OLs negatively impacted the rate of myelin growth, which is metabolically controlled. Thinner and fewer myelinated axons were observed in the optic nerve of NR1 cKO mice by P18–P20. By adulthood (P70) myelination “caught up”, however, in the context of correct circuitry formation, delays in myelination may have detrimental effects on learning and memory in the postnatal brain by leading to aberrant synapse formation. Additionally, adult NR1 cKO mice (P50–P70) that were subjected to metabolic stress in the form of oxygen-glucose deprivation, had reduced conduction velocity (CV) in the optic nerve by as much as 50% compared to WT control, further demonstrating the role OL metabolism plays in regulating both OL dynamics and axon function. Remarkably, treatment with lactate for reperfusion of the nerve promoted CV similar to control levels, demonstrating axons readily use lactate to generate ATP which in itself can regulate axon CV (Massicotte et al., 2001; Saab et al., 2016).

All mammalian cells require sufficient nutrients for proper growth and maintenance. A major player in sensing and regulating cellular growth based upon nutritional availability is the mechanistic target of the rapamycin (mTOR) pathway. mTOR consists of two complexes, mTORC1, and mTORC2, both of which are capable of integrating nutrients such as glucose, oxygen, amino acids, and insulin to influence growth and proliferation (Sabatini, 2017). mTORC1 is functionally dependent on the regulatory associated protein of the mTOR (raptor) subunit whereas, mTORC2 activity is modulated by the rapamycin insensitive companion of the mTOR (rictor) subunit (Lebrun-Julien et al., 2014). The mTOR complex interacts with Akt kinase both in upstream and downstream signaling. Interestingly, overexpression of constitutively active Akt in mouse OLs resulted in significant CNS hypermyelination and as expected, elevated mTOR activity. However, when mTOR signaling was inhibited via chronic administration of rapamycin, the observed hypermyelination was significantly reduced in young adult mice (Narayanan et al., 2009). Recent findings have been able to further unravel the contributions each mTOR complex plays in CNS myelination with an emphasis being placed on mTORC1 for its part in modulating OL dynamics and myelin lipids (Lebrun-Julien et al., 2014).

Conditional knock-out mice lacking raptor, rictor, or a combination of raptor and rictor (raptor/rictor) in adult OLs were generated to study the effects of mTORC1 (raptor-dependent) and mTORC2 (rictor-dependent) in CNS myelin. Both raptor and rictor OL-knockouts resulted in reduced myelination when compared to control but the raptor knockout mice were persistently hypomyelinated into late adulthood whereas, rictor knockout mice underwent transient hypomyelination and restored their spinal cord myelin by p60, partly due to the compensation of raptor-driven mTORC1 activity. In double raptor/rictor knockouts, hypomyelination remained throughout adulthood (6 months) indicating mTORC1 is indeed critical for myelination. However, when analyzing OL dynamics in the raptor knockouts, the number of Olig2+, CC1+Olig2+, and PDGFRa+Olig2+ cells did not differ significantly from control as they did in the rictor knockouts. Interestingly, the double knockout raptor/rictor group did exhibit reductions in OPCs, mature OLs, and late progenitor to pre-myelinating OLs (characterized by Tcf-4+/Olig2+), revealing that the total mTOR complex (mTORC1 and mTORC2) contributes to oligodendroglia maturation. When analyzing levels of known transcription factors involved in OL differentiation (Sox10 and Myrf), both the raptor/rictor and raptor mutant mice displayed strong decreases in transcript levels. In addition, the phosphorylation of 4E-BP1 and S6 kinase (downstream mTOR targets) was drastically reduced in the raptor and raptor/rictor mutants indicating mTORC1 activity is capable of regulating OL dynamics more than mTORC2 (Lebrun-Julien et al., 2014).

Because mTOR has a major role in lipogenesis, it was postulated that mTORC1 could also be in part responsible for controlling proper lipid biosynthesis in adult myelinating OLs (Soliman, 2011; Lamming and Sabatini, 2013). Similar to the hypomyelination phenotype observed during development, tamoxifen-inducible loss of raptor and raptor/rictor mutants in adult OLs displayed a significant reduction in myelin even 12 months post induction. Axon diameter was also reduced in these mutant mice. Although no changes were seen in sterol regulatory element-binding protein-1 (SREB1), transcription factors that regulate gene expression involved in lipid synthesis, the enzymatic activity of their main targets fatty acid synthase (FAS) and stearoyl-CoA desaturase-1 (SCD1), decreased in raptor and raptor/rictor mutants (Lebrun-Julien et al., 2014; Shimano and Sato, 2017). Interestingly, alterations in lipid composition were observed in the raptor mutant group with total levels of cholesterol, ceramide, sphingomyelin, and glycosylceramide (critical component of myelin—see Section “Reactive Oxygen Species”) severely reduced in the whole spinal cord (Lebrun-Julien et al., 2014). Taken together, these results indicate that mTORC1 activity of the mTOR complex is predominantly responsible for affecting OL lipid biosynthesis as well as myelin growth and maintenance.

The ability to exogenously regulate oligodendroglial metabolism has implications for a host of neurodegenerative diseases. For instance, in sporadic AD, a strong correlation exists between impaired brain glucose metabolism and disease onset and progression (Saito et al., 2021). More specifically, when analyzing gene expression data in post-mortem AD brains via RNA sequencing, impairments in both glycolytic and ketolytic genes were observed, more so in OLs than in neurons, astrocytes, or microglia, suggesting that regional OL hypometabolism may contribute much more to disease pathology than originally believed (Saito et al., 2021). In support of a more central role for OL dysfunction in AD, there is evidence that the demyelination precedes the toxic beta-amyloid and tau pathologies that lead to synaptic dysfunction and symptoms of cognitive decline in AD (Papuć and Rejdak, 2018). One drug approved for mild to moderate AD that has been shown to positively influence AD outcomes is donepezil. A potent and selective inhibitor of acetylcholinesterase (AchE), donepezil helps to reduce the breakdown of acetylcholine in the body thereby, prolonging the activity of cholinergic neurons in the brain and thus slowing down the neuronal loss normally observed in AD (Chen and Mobley, 2019). Recent research has revealed that donepezil might also exert its anti-AD therapeutic effects via its impact on OLs. Cui and colleagues conducted experiments on neural progenitor cell (NPC)-derived OPCs treated with varying concentrations of donepezil and found that donepezil promoted OPC-to-OL differentiation in a dose-dependent manner with higher concentrations nearly doubling the number of differentiated OLs relative to those in control cultures. Furthermore, this effect was limited to donepezil as other AchE inhibitors used in AD therapy such as rivastigmine, tacrine, and huperzine A did not result in any significant changes in the number of OLs. When donepezil was administered to mice undergoing active remyelination following cuprizone-mediated demyelination, by 2 weeks the donepezil treated group had a significant increase in the number of myelinated axons per field as well as thicker myelin, as defined by a significant decrease in the g ratio (the inner axonal diameter/outer axonal diameter; Cui et al., 2019). Interestingly, in a separate randomized, double-blinded, placebo-controlled study assessing the effect of donepezil in 28 patients with mild to moderate AD over a 24-week period, overall brain glucose metabolism was maintained within 0.5% of baseline levels in the treatment group whereas, the placebo group experienced an average 10.4% decline in brain glucose metabolism. In the parietal, temporal, and frontal lobes (areas where white matter lesions are often observed in AD patients), a slight but significant increase in glucose metabolism was also observed in AD patients on donepezil, which may contribute to the therapeutic effect of the drug (Tune et al., 2003). Whether or not donepezil’s pro-myelination effects are mediated through metabolic changes in OLs remains to be seen but the data gathered thus far presents a hypothesis worth investigating.

OL dysfunction in SCZ has recently been linked to cerebral dysconnectivity, where white matter anomalies, decreased OL number, and the breakdown of neuronal circuitry across varying brain regions are observed (Schmitt et al., 2011; Steiner et al., 2014). Moreover, elevated glucose levels and reduced lactate were identified in the cerebrospinal fluid of first onset, drug-naive SCZ patients (Holmes et al., 2006). Additionally, decreased expression in glycolytic and glycogen synthesis enzymes was observed and genetic studies have identified linkages between genes involved in glucose metabolism with an elevated risk of SCZ (Prabakaran et al., 2004; Stone et al., 2004). Taken together, SCZ is a disease closely tied to both OL dysfunction and impaired glucose metabolism of the brain. Clozapine, an atypical antipsychotic drug used to mitigate symptoms in SCZ, has antagonistic effects on serotonin (5-HT) and dopamine receptors, yet new evidence suggests that clozapine’s therapeutic efficacy may in part stem from its ability to upregulate key metabolic mediators in OLs, possibly by disinhibiting decreased activation of the MAPK/ERK pathway following 5-HT stimulation (Meltzer, 1994; Masson et al., 2012; Steiner et al., 2014). In support of this idea, clozapine treatment of OLN-93 cells (an OL cell line) led to increased glycolytic activity and lactate production as well as increases in GalC and myelin lipid synthesis (Steiner et al., 2014). A notable and significant increase in intracellular glucose was observed in the clozapine treatment group but the number of GLUT1 transcripts was not changed. Similarly, lactate concentrations were increased in both intracellular homogenates and extracellular media, suggesting upregulation of glycolysis in treated cells. However, no significant changes in mitochondrial oxygen consumption were noted. Interestingly, clozapine treatment led to elevated acetyl-CoA-carboxylase (ACC) activity, a key rate-limiting enzyme that converts AcCoA into the fatty acid malonyl-CoA needed for downstream lipogenesis (Brownsey et al., 2006). Although pre-clinical data with clozapine looks promising for treating white matter pathologies, a recent phase 1, randomized, blinded, placebo controlled clinical trial in patients with progressive MS (pMS), concluded that clozapine was not well tolerated as patients experienced spastic ataxic gait and worsening mobility, which resolved after drug cessation (La Flamme et al., 2020). Thus, additional studies are needed to achieve a better understanding of this promising drug’s pharmacodynamics and other parameters in the context of human disease.

The anti-hyperglycemic drug metformin has been used extensively to treat Type-2 Diabetes Mellitus and is globally the most prescribed anti-diabetic drug. Metformin is believed to exert its therapeutic effect by indirectly altering the metabolic state of cells. Due to its polarity, metformin entry into the cell is facilitated by members of the organic cation transporter (OCT) family. Following entry, metformin binds to complex I in the mitochondria to inhibit its function, resulting in decreased ATP output. This causes a shift in the cellular ATP:AMP ratio, which is detected by the enzyme complex AMP-kinase (AMPK; Zhou et al., 2001). In addition to its role as the cells’ “metabolic sensor”, AMPK is a master regulator of many signaling pathways, including the mTOR pathway (Sabatini, 2017). When activated, AMPK will suppress many energy intensive anabolic processes that require ATP (e.g., gluconeogenesis and protein synthesis) and turn on catabolic processes such as beta-oxidation to liberate more cellular ATP. In the context of type II diabetes, metformin-induced activation of AMPK leads to increased trafficking of GLUT transporters to increase glucose uptake and glycolysis, thereby, lowering blood glucose (Pernicova and Korbonits, 2014). Additionally, AMPK activation also decreases cyclic AMP, which in turn through downstream signaling increases the activity of phosphofructose kinase-1, thus leading to elevated levels of glycolysis to generate ATP and downregulation of gluconeogenesis (Pernicova and Korbonits, 2014). However, it is important to note that metformin’s effect on cellular metabolism is only well understood in hepatocytes where a large concentration of OCT1 transporters is localized in the body. Nevertheless, active research is exploring the potential other uses for metformin in metabolopathies such as cancer. Given the presence of other OCT transporters in the brain, mainly OCT3 (believed to be a functional but less potent transporter of metformin; Shirasaka et al., 2016; Koepsell, 2021), several groups have begun to investigate metformin’s effect on the CNS. New findings regarding the relationship between metformin and myelin are discussed below (Rotermund et al., 2018).

Systemic metabolic changes can also be achieved through alterations in dietary intake. For instance, a diet high in glucose, carbohydrate, and fat can lead to impaired glucose metabolism, dyslipidemia, and hyperinsulinemia, which collectively can cause a metabolic syndrome with symptoms such as hypertension, fibrosis, and inflammation in the heart, liver, and kidneys (Panchal et al., 2011). Brain development is also often affected by dietary changes. In fact, when subjecting piglets to sustained iron deficiency for the first month of life, white matter development was significantly impaired throughout with the greatest reduction seen in the corpus callosum as evident by changes in its fractional anisotropy and anatomical width (Leyshon et al., 2016). In contrast, iron homeostasis is critical for proper CNS function, particularly for OPCs and mature OLs, which have the highest cellular iron in the brain, utilizing iron as a cofactor for enzymes involved in glucose oxidation and lipid and cholesterol synthesis that make up the myelin sheath (e.g., lipid saturases and lipid dehydrogenases (Stephenson et al., 2014; Cheli et al., 2020).

Dietary fasting is also capable of inducing changes in oligodendroglia. In the previously mentioned study (see “Metformin Rejuvenates Aged OPCs to Increase CNS Remyelination” Section), Neumann and colleagues investigated the potential alternate day fasting (ADF) may have on OPC dynamics. As noted previously, aged OPCs were unable to undergo efficient differentiation as well as have decreased ATP production and basal respiration rates. ADF is capable of altering the aging processes by restricting caloric intake both of which induce changes to the body’s metabolism (Heilbronn and Panda, 2019). The ADF group was given access to food on alternate days for 6 months compared to the control group which was able to feed ad libitum. Upon cessation of the ADF fasting period, demyelination was induced in cerebellar white matter by means of focal ethidium bromide injection. Immunohistological analysis revealed a two-fold increase in the number of mature myelinating cells (identified with the markers CC1 and MBP) in the ADF group when compared to the “old” controls. This finding supports the idea that ADF may prove to be beneficial for WM deficiencies, but more studies are warranted to fully assess its therapeutic efficacy in humans.

Other dietary therapies such as the ketogenic diet (KD) have been recognized for their positive effects on disease. The KD, first described in 1921, is well known for its antiseizure effects and as such is used to treat intractable epilepsy in both adults and children (Vining, 1999). The traditional KD consists of high fat, low carbohydrate, and limited but adequate protein to allow the body to transition away from using glucose for “fuel” and instead to using ketones from fat. While the exact underlying cellular and molecular mechanisms of the KD remain elusive, it is being explored for treatment in AD, PD, and amyotrophic lateral sclerosis (ALS) because of its positive effects on the brain. Accruing evidence, however, suggests that the KD may mechanistically be working through activation of ATP sensitive potassium channels in cells of the CNS and/or by inhibition of the mTOR pathway which interestingly is linked to myelination and AMPK activity (see Section “mTORC1 Signaling Regulates OL Myelination and Lipid Metabolism”). Therefore, it seems plausible that the KD should be able to alter OL dynamics as well, a concept Stumpf and others set out to investigate using a mouse model of Pelizaeus-Merzbacher disease (PMD), a hereditary leukodystrophy. PMD most often occurs via a duplication in the X-linked myelin gene, proteolipid protein 1 (PLP1), causing PLP-cholesterol accumulation and subsequently impaired protein-lipid transport to the myelin sheath. As a result, PMD presents with pathological hypomyelination and symptoms of nystagmus, spasticity, ataxia, and impaired cognitive development. Furthermore, late stage PMD is accompanied by progressive loss of OLs, axonal degeneration, and cortical atrophy, some of which were reversed after starting a KD (Stumpf et al., 2019).

PMD and control mice were fed either a high fat/low carbohydrate KD or a standard control diet for 2–12 weeks while monitoring ketone bodies such as beta-hydroxybutyrate, which in the brain helps facilitate cholesterol synthesis and is essential for myelin membrane growth (Koper et al., 1981; Saher et al., 2005). As expected, the mice on the KD saw a five-fold increase in blood ketone levels and a 70% drop in serum glucose in as little as 7 days from KD onset when compared to control mice. Remarkably, PMD mice on the KD significantly increased PLP1 mRNA expression followed by increases in the number of Olig2+ cells and mature OLs. Furthermore, these changes in OL dynamics also resulted in more myelinated axons in the PMD-KD mice with the g-ratio matching that seen in WT control mice. Most importantly, these positive changes in myelination were able to restore normal function in the PMD mice. When assessing motor performance with the elevated beam test and rotarod test, control diet PMD mice exhibited progressive worsening of motor functions over time whereas, the PMD-KD mice retained motor fitness at WT levels. Independent of PMD pathology, the researchers confirmed the pro-OL/pro-myelin effects of the KD diet were also applicable in the cuprizone model of adult remyelination where during the cuprizone off/remyelination phase, mice on a KD diet saw accelerated repair 2 weeks into remyelination, evident by significant increases in mature OLs and reduction in the number of slips on the beam motor test (Stumpf et al., 2019). Hence when taken together, entering into a state of ketosis may prove beneficial for patients coping with WM diseases and in fact several clinical trials are currently underway to determine if the KD diet can reverse and repair WM damage in the CNS (ClinicalTrials.gov Identifier: NCT03718247, NCT04820478, NCT03509571, CT04530032, and NCT04701957).