Oligodendrocyte-derived transcellular signaling regulates axonal energy metabolism

Abstract

The unique morphology and functionality of central nervous system (CNS) neurons necessitate specialized mechanisms to maintain energy metabolism throughout long axons and extensive terminals. Oligodendrocytes (OLs) enwrap CNS axons with myelin sheaths in a multilamellar fashion. Apart from their well-established function in action potential propagation, OLs also provide intercellular metabolic support to axons by transferring energy metabolites and delivering exosomes consisting of proteins, lipids, and RNAs. OL-derived metabolic support is crucial for the maintenance of axonal integrity; its dysfunction has emerged as an important player in neurological disorders that are associated with axonal energy deficits and degeneration. In this review, we discuss recent advances in how these transcellular signaling pathways maintain axonal energy metabolism in health and neurological disorders.

INTRODUCTION

The mammalian CNS is a well-integrated network comprised of two main cell classes: neurons and glial cells; the latter subclassified as astrocytes, microglia, and OLs. These glial cells provide physical and metabolic support to neurons, modulate neuronal communication, and maintain microenvironment homeostasis [1]. In the CNS, OLs enwrap lipid-rich myelin sheaths in concentric spirals around axons with diameters larger than 0.2 μm [2]. Nodes of Ranvier, which are periodic gaps between myelin segments that are exposed to the extracellular space, accelerate the conduction of nerve impulses [3,4]. The bulk of myelinated axons localizes within the white matter, which is organized into tracts for fast transmission of information to connect different brain regions into functional circuits [5]. While myelin sheaths cover most of the axonal surface for electrical insulation, it also limits the rapid access of metabolites and nutrients from the extracellular environment to the axonal cytoplasm. Because of their unique association with CNS axons, OLs are well-positioned to act as metabolic intermediaries between myelinated axons and the extracellular matrix. Generic studies of OL-specific proteins provide early evidence that OLs support axonal integrity and survival beyond acting as passive electrical insulators. For example, knockout (KO) mouse models of proteolipid protein 1 (PLP1) or 2’–3’-Cyclic nucleotide 3’-phosphodiesterase (CNP) exhibit severe axonal degeneration in the absence of alterations in the electrical conduction properties of the axon [6,7]. The myelin basic protein (MBP)-null “shiverer” mice maintain intact axons despite demyelination [8,9]. The deficits in information processing upon demyelination in MBP-deficient mice were similar to those resulting from a reduction in axoglial metabolic support [10]. These studies highlight important roles for OLs in supporting axonal integrity independent of electrical insulation.

The brain is one of the most energy-demanding organs in the body: it consumes nearly 20% of the body’s total cellular energy in the form of adenosine triphosphate (ATP) [11]. Glucose is the predominant source of carbon for the generation of ATP under physiological conditions, whereas alternative energy substrates (such as lactate, acetate, and ketone bodies) are also supplied to neurons for generating ATP when glucose supply is limited [12–15]. Glucose metabolism generates ~95% of the ATP in the brain jointly through glycolysis and oxidative phosphorylation (OXPHOS). Each glucose is catabolized by glycolysis into two ATP molecules and two pyruvate molecules which are subsequently transferred into the mitochondrion for entry into the tricarboxylic acid cycle, producing 30–36 ATP molecules through OXPHOS [16]. The complete oxidation of glucose gives a near stoichiometric ratio of oxygen to glucose consumption of 6:1, which is the most efficient metabolic pathway in neurons to fulfill their energy demand (Figure 1).

Figure 1. Energy metabolism in axons.

Glucose is the predominant carbon source for synthesizing adenosine triphosphate (ATP). Glucose uptake into the axonal cytoplasm occurs via the major neuronal glucose transporters GLUT3/4. Glycolysis converts one glucose molecule into two pyruvate molecules, with a net production of two ATP molecules. Under aerobic conditions, these pyruvate molecules are subsequently transferred into the mitochondrion for entry into the tricarboxylic acid cycle, where they are converted into acetyl-CoA molecules to fuel oxidative phosphorylation (OXPHOS), leading to a net production of ~32 ATP molecules. The complete oxidation of glucose by the combined action of glycolysis and OXPHOS is the most efficient bioenergetic source of the cellular ATP that maintains axonal energy homeostasis. When glucose availability in axons is limited, other metabolites, such as lactate, acetate, and ketone bodies, serve as alternative energy substrates for axonal ATP synthesis. Lactate is converted to pyruvate by lactate dehydrogenase (LDH). Acetate activation to acetyl-CoA is catalyzed by the ADP-forming acetyl-CoA synthetase (ACD). Although glial ketogenesis in supporting axonal energy metabolism requires further validation in neuronal systems, ketone bodies could be oxidized to acetyl-CoA molecules through a series of enzymatic reactions involving d-3-Hydroxy-n-butyrate dehydrogenase (BDH1), 3-oxoacid CoA-transferase (SCOT), and acetyl-CoA acetyltransferase (ACAT). Through conversion to pyruvate or acetyl-CoA molecules, these alternative metabolic substrates could enter OXPHOS within mitochondria for ATP generation to maintain axonal energy metabolism.

Axonal terminals utilize approximately 55% of the total ATP in neurons, where most energy is used to maintain resting potentials, fire action potentials, and sustain synaptic vesicle (SV) recycling [17–20]. An energy budget estimates that optic nerve axons with diameters smaller than 0.9 μm have a 38% deficit in ATP production compared to their ATP consumption [21]; these axons likely require energetic support from neighboring cells. This provides a theoretical basis for OLs to support axonal energy metabolism, thus opening an emerging theme relevant to a range of neurological disorders, as axonal mitochondrial dysfunction and energetic failure have been observed in human OL-deficient diseases and animal models for hereditary spastic paraplegia (HSP), multiple sclerosis (MS), and amyotrophic lateral sclerosis (ALS) [22–25]. In this short review, we limit our discussion to current knowledge and new insights into the transcellular OL-neuron signaling in support of axonal energy metabolism through metabolite shuttling and exosome signaling. We also summarize emerging lines of evidence showing how deficient OLs impair axonal energy metabolism in neurological disorders. For additional insights from different perspectives of OL-derived support in neuronal metabolism and function, we refer readers to other in-depth reviews [26–31].

Neuronal adaptation to energetic stress

Neurons adopt special mechanisms maintaining local ATP supply through coordinating glycolysis and OXPHOS in distal axons and synapses, where bioenergetic failure is particularly vulnerable and relevant to neurodegenerative diseases [32]. Revealing these mechanisms is an emerging frontier for therapeutic investigation. For instance, intensive synaptic activity places a large burden on the axonal energy budget. Neurons rapidly respond to activity-induced energetic stress by activating the AMP-activated protein kinase (AMPK) energy-sensing pathway, which recruits glucose transporter type 4 (GLUT4) to presynaptic membranes to enhance glucose uptake and glycolytic activity [33,34]. To maximize ATP production capacity, activity-dependent activation of AMPK further recruits axonal mitochondria to presynaptic terminals via myosin VI and mitochondrial anchoring protein syntaphilin (SNPH) [18], and/or remodels mitochondrial cristae ultrastructure to support the energy-expensive SV recycling under long-term synaptic activation [35]. In the CNS, acute axonal injury causes severe damage to the cellular environment and mitochondrial bioenergetics. In mature neurons, axonal mitochondria largely remain stationary, and those damaged mitochondria at injury sites cannot be replaced by healthy ones, leading to local energy crisis [32,36]. The insufficient ATP supply in injured axons is among one of the key intrinsic factors leading to regenerative failure [37,38]. Thus, turning off SNPH-anchoring through an energy-signaling cascade in injured axons allows healthy mitochondria to be delivered into injured areas, thereby facilitating axonal regeneration by rescuing the energy crisis [39]. Although glycolysis is an important aspect of axonal maintenance and growth cone dynamics [40], it remains elusive whether glycolytic activity alone is sufficient to maintain axonal energy metabolism after CNS injury. By trafficking healthy mitochondria to the injury sites, neurons can partially restore axonal energy supply to power regeneration, which requires high levels of energy. Considering the intricate networks in the human brain where billions of neurons and glial cells wire together, this raises an urgent question of whether additional energetic support from neighboring glial cells is beneficial for neurons to maintain axonal energy homeostasis under energetic stress. With myelin sheaths covering up to 98% of the CNS axonal surface, OLs are well-positioned to provide metabolic support to maintain axonal integrity [26,27].

Transcellular OL-neuron metabolic signaling

The energetic metabolic strategies employed in OLs

A single OL can myelinate up to 50 axons, wrapping as many as 150 layers of myelin membrane around each axon [28]. This large cellular expansion places enormous energetic demands on OLs. Analytical calculations predict that ~6.84×10²¹ ATP molecules are needed to produce 1 gram of myelin protein and 48-fold that per gram of myelin lipid [21]. To meet this energetic demand, OLs employ a metabolic strategy heavily relying on mitochondria before and during myelination, with a high density of long, tubular mitochondria to support high OXPHOS rates [41]. Upon completion of myelination, the number and size of mitochondria in the mouse OL processes are dramatically reduced, with only half the density and length of neuronal mitochondria [42–44]. Consistently, a metabolic profiling study using stable isotope tracing revealed a higher glycolytic than oxidative metabolic activity in the white matter of rat brains [45], suggesting a metabolic shift towards glycolysis in mature OLs. This metabolic switch in OLs is in the opposite direction as neuronal differentiation: a switch from glycolysis to mitochondrial OXPHOS occurs during differentiation of human neural stem cells to mature neurons [46]. Therefore, OLs and neurons exhibit large diversity in transcriptomic, proteomic, and metabolomic landscapes [47–49]; OLs serve as an ideal metabolic supplement for axonal energy production by providing metabolites and proteins that are in short supply in neurons generally through two pathways: (1) shuttling energy substrates, such as glucose, lactate, and pyruvate, that can be directly metabolized by neurons for energy production and (2) secreting exosomes containing proteins, RNAs, and other materials that boost neuronal energy metabolism.

Metabolite shuttling

OLs serve as metabolite reservoirs for neurons by uptaking glucose as their main carbon source from blood circulation through glucose transporters type 1 (GLUT1) [50]. Under high levels of glucose, the rate of glycolytic pyruvate production in the cytosol exceeds its rate of mitochondrial utilization, thus excess pyruvate is converted to lactate, the end-product of anaerobic glycolysis. The reversible conversion of lactate to pyruvate is catalyzed by lactate dehydrogenase (LDH). Evidence for OL-axon metabolic signaling initially came from studies of mouse optic nerve explants where axons are completely myelinated [23,51]. As intracellular lactate accumulates, it can pass from the OL soma through myelinic channels to reach the inner tongue processes enwrapping an axon. Lactate is released into the periaxonal space through the monocarboxylate transporter 1 (MCT1), and subsequently taken up by axons via the neuronal isoform MCT2, presumably fueling neuronal mitochondrial OXPHOS [23] (Figure 2). It is worth noting that pyruvate is passively transported to axons through the same mechanism as lactate through MCTs [51]. The lactate/pyruvate exchange between OLs and neurons are determined by axonal energetic demands. Overproduction of lactate could result in lactic acidosis, which causes irreversible cell damage and is detrimental for both grey and white matter functions [52]. Therefore, the glucose utilization and the glycolytic rate in OLs is tightly adapted to meet different axonal energy demands and to avoid excessive lactate production.

Figure 2. Transcellular metabolite shuttling and metabolic signaling between oligodendrocytes and neurons.

Oligodendrocytes (OLs) and neurons exhibit large diversity in transcriptomic, proteomic, and metabolomic landscapes, and thus OLs serve as an ideal metabolic supplement for axonal energy production by providing metabolites and proteins that are in short supply in neurons through two pathways: (1) shuttling energy substrates, such as lactate and pyruvate, that can fuel axonal energy metabolism and (2) secreting exosomes containing proteins and RNAs that can boost neuronal energy metabolism. Mature OLs are highly glycolytic; the glycolytic end products pyruvate and lactate can pass from OLs to axons through the myelinic monocarboxylate transporter 1 (MCT1) and its neuronal isoform MCT2. After entering axons, lactate is converted to pyruvate by LDH in the axon cytoplasm. Pyruvate fuel mitochondria ATP production to support axonal energy metabolism. In addition, astrocytes participate in metabolite shuttling between OLs and axons. Astrocytes are highly glycolytic and almost exclusively store brain glycogen, where glycogen is broken down to glucose, which is metabolized by glycolysis to produce pyruvate and lactate. Astrocyte-derived glucose and lactate/pyruvate are transported to OLs through gradient-diffusion along MCTs and/or connexin channels between the astrocyte-OL interface, thus maintaining the metabolite shuttling pathway when glucose availability is restricted. OLs also release exosomes containing proteins, lipids, and RNA molecules to support axonal energy metabolism. One of these components within OL-derived exosomes is NAD-dependent deacetylase sirtuin 2 (SIRT2), which is undetectable in neurons but highly enriched in mature OLs and released within exosomes. After being released from exosomes within axons, SIRT2 targets mitochondria and deacetylates adenine nucleotide translocase 1 and 2 (ANT1/2). SIRT2-dependent deacetylation of ANT1 robustly enhances ANT1’s affinity for ADP binding, facilitating the conversion of ADP to ATP and the export of mitochondrial ATP. Thus, transcellular OL-axon metabolic signaling is critical for maintaining or boosting axonal bioenergetics. OLs release exosomes by sensing neuronal activity via NMDARs in the myelinating processes of OLs. NMDAR activation also induces GLUT1 expression in the myelin sheath, resulting in increased glucose uptake and glycolytic activity, which eventually enhances gradient-based shuttling of pyruvate and lactate from OLs to axons. These pathways allow neurons to maintain axonal energy homeostasis and fulfill activity-enhanced energy demand.

N-methyl-D-aspartate receptors (NMDARs) are expressed in the myelinating processes of OLs and activated by neuronal glutamate release, thus relaying neuronal energy demand to nearby OLs. The activation of NMDARs mobilizes GLUT1 to myelin sheaths, resulting in increased glucose uptake and glycolytic activity leading to lactate build-up in OLs [50]. This OL-derived lactate shuttling through MCT1 provides on-demand metabolic support to axonal compartments. The role of MCT1 in maintaining axonal integrity has been elucidated in animal models, where heterozygous MCT1-null mice develop widespread axonopathy with no change in overall myelination by 8 months of age [23]; mice with deletion of MCT1 develop late-onset hypomyelination and significant axonal degeneration [53]; blockade of glial MCT function in Drosophila cause age-dependent neuron loss [54]. While these studies highlight MCT1 in the maintenance of axonal integrity independent of myelination, the effect of MCT1 deletion on axonal ATP availability or axonal lactate content has not been directly addressed. Although it is well established that OLs deliver lactate to axons and neurons employ lactate as an energy substrate [55–59], the evidence that neurons can metabolize OL-derived lactate is still indirect. An ex vivo imaging study in transgenic mice expressing a FRET-based ATP sensor revealed that electrical activity firing in acutely isolated optic nerves reduces axonal ATP content, suggesting that axonal glycolysis products are not sufficient to maintain activity-enhanced mitochondrial energy metabolism on their own [60]. This study strongly argues for a critical role of oligodendrocytes in delivering energy substances to axons of CNS neurons.

The brain also adapts its metabolism by switching from glucose-based to lipid-based fuel sources under low glucose availability [15,61]. Lipids are key molecules of the brain’s structure and function and comprise ~50% of the brain’s dry weight [62]. It is estimated that up to 20% of total brain energy required for the maintenance of basal activity is provided by fatty acids [63]. While neurons do not store a significant pool of lipid droplets, astrocytes and OLs are the main cell types reserving lipid components in the brain. Astrocytes metabolize fatty acids through mitochondrial β-oxidation and export ketone bodies such as acetoacetate and β-hydroxybutyrate through MCTs, which can be taken up and utilized by neighboring neurons for energy production [64]. In the Drosophila starved brain, the cortex astrocyte-like glia fuel neurons with ketone bodies as an energy substrate to support memory function [15], suggesting an alternative metabolite shuttling pathway between glia and neurons. As myelination is largely dependent on continuous lipid synthesis [65], OLs are suppliers of lipid metabolites, which have been proposed to serve as energy reserves when circulating glucose is low [66]. During normal myelin turnover, myelin lipids are recycled by the autophagy-lysosomal pathway, thus liberating fatty acids for synthesis of new myelin lipids. When glucose availability is reduced, OLs break down myelin, allowing free fatty acids to undergo β-oxidation to create acetyl coenzyme A (acetyl-CoA), an entry substrate for mitochondrial OXPHOS. By utilizing ATP generated through myelin catabolism, OLs are capable of sharing more glucose-derived pyruvate/lactate with the axonal compartments in a mouse optic nerve model system [66]. Such a shift from normal myelin turnover to lipid-based ATP generation is essential for OL survival, as well as transiently reversing axonal energy deficits under low glucose conditions. Based on assumptions about metabolite shuttling coupled with lipid catabolism, this study provided new insights into how OLs adapt to nutrient stress to protect axons from hypoglycemia-induced degeneration. The metabolic switch from glucose-based to lipid-based fuel serves as a systematic chronic adaptation during the progression of aging-related neurodegenerative disease, including Alzheimer’s Disease (AD). Glucose hypometabolism and a compensatory shift to ketone body-based energy production has been reported as a metabolic phenotype of the AD brain in both clinical and preclinical studies [67,68]. Mitochondrial function is compromised in the early stages of AD [69]. Declined mitochondrial respiration in the mouse brain leads to the catabolism of myelin sheaths by activating phospholipase-A₂ sphingomyelinase, thus myelin catabolism provides ketone bodies to fuel an energetically-compromised brain during glucose hypometabolism [70]. However, the extended breakdown and catabolism of myelin lipids is detrimental to myelin integrity, resulting in white matter demyelination and axonal degeneration, which is a common pathological hallmark of AD.

Metabolite shuttling also engages panglial networks to coordinate glial metabolite provisions to meet the energetic demands of the brain [71]. Both astrocytes and OLs uptake glucose from blood circulation through GLUT1 and supply lactate/pyruvate to neurons through MCTs [29,72]. Intriguingly, brain glucose is stored as glycogen almost exclusively in astrocytes, where glycogen is broken down to glucose, which is metabolized by glycolysis to produce pyruvate and lactate [73,74]. Astrocyte-derived glucose and lactate/pyruvate is transported to OLs through gradient-dependent diffusion along GLUTs, MCTs, and connexin channels during glucose deprivation [75,76]. One can speculate that OLs further deliver the glucose and lactate/pyruvate to myelin-covered axons, which astrocytes have limited access to. Although the flow direction and mechanisms regulating metabolite transfer from astrocytes to neurons through OLs remain largely unclear, these studies suggest a role for panglial networks in metabolite provisions to axons (Figure 2).

Extracellular vesicle (EV) signaling

EV signaling is increasingly recognized for intercellular communication. EVs are heterogenous membranous structures released by cells into the extracellular space, allowing exchange of proteins, lipids, and RNA molecules. EVs can be subclassified as exosomes which are produced through the endocytic pathway with a diameter of 30–100 nm, and ectosomes which are shed from the plasma membrane with a diameter of 50–1000 nm [77]. Exosomes are secreted via exocytosis into the extracellular space and taken up by the recipient cell. In the CNS, both glia and neurons release exosomes under physiological and pathological conditions, and these exosomes play critical roles in supporting myelination, synaptic function, and energy metabolism [78–81]. Rapid and targeted transcellular communication through exosomes in the CNS is crucial for optimal brain function. OL-derived exosomes were first found to be internalized by neurons in vitro and in vivo and to carry major proteins involved in myelination [82,83]. A comprehensive proteomic analysis further revealed that OL-derived exosomes carry a multitude of proteins involved in cellular regulation and cytoskeletal dynamics [84]. The secretion of exosomes from OLs is triggered by activity-dependent release of neuronal glutamate, which binds OL-NMDA and -α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors [83]. OL-derived exosomes promote neuronal survival under oxygen-glucose deprivation, elevate action potential firing rate, enhance metabolism, activate pro-survival signaling pathways, and alter gene expression patterns and the transcriptome [83,85]. Therefore, OL-derived exosomes have remarkable potential in supporting neuronal metabolism, function, and health; however, the underlying mechanisms remain largely elusive.

A recent study in mouse neurons revealed that OL-derived exosomes boost axonal mitochondrial bioenergetics by delivering NAD-dependent deacetylase sirtuin 2 (SIRT2) to axons [81]. SIRT2 is undetectable in neurons but highly enriched in mature OLs and released within exosomes [82,84,86]. The transfer of SIRT2 to axons adds an additional layer of deacetylation modification on neuronal mitochondrial proteins. After being released from exosomes within axons, SIRT2 targets mitochondria and deacetylates adenine nucleotide translocase 1 and 2 (ANT1/2) [81] (Figure 2). ANT1/2 play an essential role in bioenergetics by exchanging ADP and ATP across the mitochondrial inner membrane, therefore converting ADP to ATP and catalyzing the export of mitochondrial ATP [87]. ANT1 is acetylated at lysine (Lys) 10, 23, and 92; these positively charged lysines are critical for binding and transport of negatively charged ADP [88]. Acetylation at these lysines neutralizes positive charges, leading to lower binding affinity to ADP and reduced ATP production [89]. SIRT2-dependent deacetylation of ANT1 at Lys23 robustly enhances ANT1 for ADP binding, thus facilitating the conversion of ADP to ATP [90]. Intriguingly, knockdown of SIRT2 in OLs or deletion of the sirt2 gene in mice abolishes the OL-axon crosstalk in boosting axonal bioenergetics. Injection of OL-derived exosomes rescues axonal mitochondrial deficiency in the spinal cord of sirt2 KO mice [81]. These studies provide a new concept that exosome-mediated delivery of SIRT2 from OLs to axons is an efficient and robust mechanism boosting axonal mitochondrial bioenergetics. It remains unknown whether this transcellular crosstalk is regulated in response to axonal energy stress or sustained synaptic activity. It was reported that OLs release EVs by sensing neuronal activity via NMDARs [83]. NMDAR activation also induces GLUT1 expression in the myelin sheath and stimulates production of lactate/pyruvate, thereby supporting axonal energy metabolism during increased synaptic activity [50]. Thus, further investigations are necessary to determine whether SIRT2-filled exosomes are secreted from OLs in response to energetic stress in local axonal compartments. Motor-driven axonal transport is an energy-consuming process [21]. It was recently reported that OL-derived exosomes rescue impaired axonal transport under nutrient deprivation [84]. It remains unclear whether SIRT2 or other energy-signaling molecules mediate this rescue effect.

The composition and secretion of EVs are influenced by the disease progression of neurological disorders [91]. Exosomes can carry and transfer misfolded proteins such as α-synuclein, amyloid precursor protein (APP) and its cleavage products; these pathological forms are associated with neurodegenerative pathology [92]. For example, multiple system atrophy (MSA) is a demyelinating disease characterized by glial inclusions of α-synuclein. A nanoparticle tracking analysis revealed that α-synuclein was significantly lower in OL-derived EVs from the blood of MSA patients than in those with Parkinson’s disease. In addition, the concentration of OL-derived exosomes was reduced in MSA patients compared to healthy controls [93]. These suggest that the content and secretion of OL-derived exosomes are likely related to the pathological α-synuclein aggregation in MSA. Glutamate excitotoxicity is a prominent pathophysiological feature of multiple sclerosis (MS) [94]. Pathological glutamate accumulation in experimental autoimmune encephalomyelitis (EAE), an inflammation-driven disease model of MS, can trigger excessive release of exosomes from OLs [83], thus raising an urgent question of whether altered composition and secretion of OL-derived EVs is reflected by or contributes to the disease progression.

Deficient OLs impair axonal energy metabolism in neurological disorders.

A range of neurological disorders are characterized by axonal energetic failure that leads to progressive axonal and neuronal degeneration. In the past two decades, we have witnessed a rapid growth in knowledge of OL-linked axonal pathogenesis [6,7,23,25,30,95–100]. Here we provide a summary of the pathological roles of deficient OLs on axonal energy metabolism (Table 1).

Table 1.

Axonal mitochondrial phenotypes in rodent and human disease models associated with deficient OLs.

OL-deficient models	Metabolic phenotypes	Refs
PLP-DM20-deficient mice	Focal axonal swellings containing mitochondria in white and gray matter from the age of 6–8 weeks.	[6]
PLP1-deficient patient	Axonal swellings with accumulated mitochondria.	[95]
Postmortem human optic nerve	High cytochrome-c oxidase activity in unmyelinated regions.	[103]
Cnp1-null mice	Mitochondrial accumulation at axonal swellings of spinal cord.	[7]
Shiverer mice (MBP-null)	Increased mitochondrial activity and number within dysmyelinated axons.	[9]
MS patient motor cortex	Reduced mitochondrial gene expression and respiratory chain activity.	[107]
Pattern III MS	Defects in respiratory chain complex IV in axons.	[97]
Active chronic MS	Reduced complex IV activity.	[97]
Inactive chronic MS	Mitochondrial accumulation at axonal swellings and higher complex IV activity in large diameter axons.	[97]
Plp1-overexpressing mice	Increased mitochondrial density in spinal cord axons.	[108]
Diphtheria Toxin subunit A mice	Increased mitochondria numbers in demyelinated axons.	[98]
Spinal experimental encephalomyelitis lesions	Swollen mitochondria in focal axonal degeneration.	[99]
MCT1 heterozygous KO mice	Swollen and enlarged mitochondria in optic nerve.	[23]
P0-CNS mice (expressing P0 instead of PLP in CNS myelin)	Lower axonal ATP levels in optic nerve.	[104]
Plpnull/y mice	Increased GLUT1 and MCT1 and faster recovery of ATP levels upon reperfusion of glucose.	[106]
OL-specific MCT1 KO	Enlarged mitochondria with irregular cristae structure.	[53]

Deficient OLs induce mitochondrial pathology

A wealth of evidence in demyelination models has shown that deficient OLs impair axonal mitochondrial function and trafficking. PLP is the most abundant protein in CNS myelin that is associated with a variety of neurodegenerative diseases, including HSP, SPG2, MS, and others [22–25]. PLP-null mice develop a late-onset axonal degeneration, characterized by mitochondrial accumulation at distal paranodes [6]. Accumulation of mitochondria at axonal swellings is also involved in several other OL-deficient neurodegeneration models, including mice lacking CNS myelin protein CNP [7] and shiverer mice which harbor a mutation in MBP [9]. These mouse models recapitulate the mitochondrial pathological phenotypes found in human MS patients [95,97,99,100]. The irregular build-up of axonal mitochondria is attributed to impaired transport caused by OL dysfunction. In addition, expression of the axonal mitochondrial anchoring protein SNPH is elevated within chronic MS lesions [97]. SNPH-mediated anchoring is required for the volume increase of axonal mitochondria following demyelination [101]. Notably, deleting SNPH from the shiverer mouse prolongs survival and reduces cerebellar degeneration, presumably by replenishing healthy mitochondria to distal axons [102], suggesting that remobilizing mitochondria benefits axonal energy maintenance. While mitochondrial accumulation is one of the most prominent features of axonal neurodegeneration, whether their bioenergetics are also impaired by deficient OLs remains debatable [9,97,103,104].

Transcellular metabolic signaling in neurodegeneration

OL-derived and MCT1-mediated lactate and pyruvate shuttling provides metabolic support to axons. However, MCT1 expression in OLs is significantly reduced with aging [53]. Loss of MCT1 is observed in human neurodegenerative diseases such as ALS and Creutzfeldt-Jakob disease, and their animal models [23,105]. MCT1 ablation in aging mice leads to axonal degeneration [53]. Plp^null/y mice display altered myelin structures associated with increased GLUT1 and MCT1; ATP levels in optic nerves from Plp^null/y mice recover faster and more completely from glucose deprivation [106]. This study highlights that OL-axon metabolite shuttling is enhanced in response to energetic stress, thus protecting axons from degeneration. It remains elusive whether impaired OL-axon metabolite shuttling contributes to axonal energy deficits during neurodegenerative progression. PLP- and CNP-deficient mice develop a secondary progressive axonal degeneration characterized by the formation of axonal swellings. Electron microscopy analysis in PLP-deficient mice demonstrated a 3.6-fold higher number of OL-derived multivesicular bodies (MVBs), which resemble the cytoplasmic storage sites of exosomes before their release. Exosomes are mis-localized in myelinated tracts of PLP- and CNP-deficient mice, and released exosomes are significantly reduced in these models. Consistently, PLP- and CNP-deficient exosomes fail to support axonal transport or neuronal metabolism during nutrient deprivation [84]. This study highlights that OL-axon transcellular signaling through exosomes is a promising therapeutic intervention for protecting axons from neurodegeneration.

CONCLUDING REMARKS

Recent studies have started to uncover the emerging role of transcellular OL-axon metabolic crosstalk in the support of axonal energy metabolism and axonal integrity, which is highly relevant to a range of neurological disorders. These studies also raise several urgent questions to be addressed: (1) whether defective OL-axon energy crosstalk plays a convergent pathological role leading to axonal energy failure and axonal degeneration; (2) whether OLs employ diverse forms of metabolic support across different brain regions where OLs and astrocytes coordinate the glial metabolite provisions to meet the enhanced energetic demands; (3) whether other exosomal proteins also modulate axonal energy metabolism by targeting to mitochondrial OXPHOS, glycolytic activities, or other bioenergetic pathways; and (4) how OLs sense axonal energy stress or adapt to neuronal signaling under physiological and pathological conditions. Current methodology has limited capacity for labeling and tracing metabolites at the cellular level in vivo. Metabolic imaging in higher resolution thus needs to be developed for monitoring transcellular glia-axon metabolic crosstalk. This will advance our understanding of how energy substrates are transferred among OLs, neurons, and astrocytes, and how they are metabolized in axons under physiological and pathological conditions. Future investigations into signaling mechanisms regulating OL-derived metabolic support will benefit the development of therapeutic interventions to restore axonal bioenergetics, and thus protect axons from degeneration.

Highlights.

• Neurons adopt special mechanisms maintaining local ATP supply in distal axons

• Oligodendrocytes provide metabolic support to axons by shuttling metabolites

• Oligodendrocytes release exosomes to boost axonal mitochondrial bioenergetics

• Deficient oligodendrocytes impair axonal energy metabolism in neurodegeneration