Glial-neuron crosstalk in health and disease: A focus on metabolism, obesity, and cognitive impairment

Abstract

Dementia is a complex set of disorders affecting normal cognitive function. Recently, several clinical studies have shown that diabetes, obesity, and components of the metabolic syndrome (MetS) are associated with cognitive impairment, including dementias such as Alzheimer’s disease. Maintaining normal cognitive function is an intricate process involving coordination of neuron function with multiple brain glia. Well-orchestrated bioenergetics is a central requirement of neurons, which need large amounts of energy but lack significant energy storage capacity. Thus, one of the most important glial functions is to provide metabolic support and ensure an adequate energy supply for neurons. Obesity and metabolic disease dysregulate glial function, leading to a failure to respond to neuron energy demands, which results in neuronal damage. In this review, we outline evidence for links between diabetes, obesity, and MetS components to cognitive impairment. Next, we focus on the metabolic crosstalk between the three major glial cell types, oligodendrocytes, astrocytes, and microglia, with neurons under physiological conditions. Finally, we outline how diabetes, obesity, and MetS components can disrupt glial function, and how this disruption might impair glia-neuron metabolic crosstalk and ultimately promote cognitive impairment.

1. Introduction

The burden of dementia, defined as an impairment in mental capacity that interferes with daily function, is growing at a rapid pace as the population increases in size and age. Dementia is not one specific disorder but rather a constellation of signs and symptoms, which include poor judgement, faulty executive function, poor memory, and declining social skills. Alzheimer’s disease (AD) is a complex neurodegenerative brain disease, which results in symptoms of dementia and constitutes the bulk of dementia cases. However, cognitive impairment occurs on a continuum, from mild cognitive impairment (MCI) leading up to the more serious loss of cognitive function present in frank AD. Although well-known AD risk genes exist, most notably apolipoprotein E ε4 (APOE ε4) (Serrano-Pozo, Das et al. 2021), the vast majority of AD cases are sporadic and lack a known genetic determinant. Multiple clinical studies have identified associated AD risk factors, including components of the metabolic syndrome (MetS), such as diabetes (Biessels and Despa 2018) and obesity (O’Brien, Hinder et al. 2017). AD itself is also characterized by dysfunctional metabolism, including insulin resistance (Kim and Feldman 2015, Kellar and Craft 2020) and impaired glucose and lipid metabolism in the brain (Butterfield and Halliwell 2019, Zhu, Zhang et al. 2019). Thus, impairment of both systemic and brain metabolism are intimately linked to neurodegeneration and are important areas of ongoing inquiry.

This relationship between systemic and brain metabolism with neuronal health is intuitive. Neurons rely on well-coordinated bioenergetics to transmit signals, turnover neurotransmitters, and regulate synaptic and dendritic spine formation. Thus, breakdown in bioenergetics will impair each of these functions and would eventually lead to neurodegeneration and AD. Moreover, neurons are aided in these roles by glia, and AD is now increasingly believed to progress in a non-cell autonomous manner (Heneka, Carson et al. 2015), suggesting that AD entails a breakdown in axo-glial communication. However, the precise mechanisms remain to be elucidated.

In this review, we present the studies supporting the evolving idea that disrupting axo-glial metabolic crosstalk can promote neurodegeneration and AD. First, we outline the clinical evidence demonstrating that systemic metabolic dysfunction increases the risk of AD. Next, we summarize the known mechanisms of axo-glial metabolic crosstalk under physiological conditions. Finally, we discuss how disruption of this crosstalk under pathological conditions promotes cognitive impairment.

2. MetS, diabetes, obesity, and cognitive impairment

Systemic metabolic dysfunction in humans is characterized by the presence of the metabolic syndrome (MetS). The MetS is defined as three out of five metabolic criteria: elevated waist circumference (≥102 cm males, ≥88 cm females; i.e., obesity), increased systolic (≥130 mmHg) or diastolic blood pressure (≥85 mmHg), increased triglycerides (≥150 mg/dL), elevated fasting blood glucose (>100 mg/dL; i.e., prediabetes, diabetes), and lower high-density lipoprotein cholesterol (HDL-c; <40 mg/dL males, <50 mg/dL females) (Grundy, Cleeman et al. 2005). Over the past decade, several clinical studies have shown that the components of MetS are dementia risks.

In diabetes patients with hyperglycemia, impairment commences very subtly as so-called diabetes-associated cognitive decrements but can progress to MCI or overt AD (Biessels and Despa 2018). Cognitive impairment secondary to diabetes is accompanied by structural brain changes and pathological processes, such as central insulin resistance (Arnold, Arvanitakis et al. 2018), inflammation, and oxidative stress (Biessels and Despa 2018). Cross-sectional analysis of the Spanish PREDIMED-PLUS study (n=6,823) found that older (mean 65 years), overweight or obese type 2 diabetes (T2D) participants with glycated hemoglobin (HbA1c)<53 mmol/mol (7%), a hyperglycemia surrogate, had better executive function versus participants above this level, after various adjustments, including education level (Mallorquí-Bagué, Lozano-Madrid et al. 2018). The US ARIC study of older participants (n= 5,099), spanning four states, identified risks for incident MCI at a 5-year median follow-up, which include diabetes (adjusted hazard ratio [HR] 1.14 [95% confidence interval (CI) 1.00, 1.31]), poor glycemic control based on HbA1c in diabetic individuals (HR 1.31 [95%CI 1.05, 1.63]), and longer diabetes duration (≥5 vs. <5 years; HR 1.59 [95%CI 1.23, 2.07]), adjusted for several covariates and education (Rawlings, Sharrett et al. 2019). Diabetes already in midlife can have far-reaching consequences in later-life; analysis of the Japan Public Health Center-Based Prospective Study (n=12,219, aged 40–59 years) found that diabetes correlated positively with incident dementia risk (odds ratio [OR] 2.60 [95%CI 1.12, 6.03]. Thus, overall diabetes associates with dementia or MCI in cross-sectional and longitudinal studies and across diverse populations.

Several studies also report an association between dyslipidemia and central obesity with cognitive impairment and dementia. As with diabetes, obesity has pathological processes in common with AD, such as inflammation and mitochondrial dysfunction (O’Brien, Hinder et al. 2017). In a cross-sectional US Michigan cohort (n=184), obese, but normoglycemic, participants performed more poorly on the NIH Toolbox versus lean normoglycemic controls, after multiple adjustments, including education level (Callaghan, Reynolds et al. 2020). This study indicates that elevated waist circumference is a risk for cognitive impairment independent of hyperglycemia. A meta-analysis of 21 longitudinal studies with a minimum 2-year follow-up found that being overweight or obese correlated positively with incident dementia (risk ratio [RR] 1.41 [95%CI 1.20, 1.66] in participants less than 65 years old; interestingly, the trend reversed above 65 years of age (RR 0.83 [95%CI 0.74, 0.94]) (Pedditzi, Peters et al. 2016). Another analysis of ARIC (n=13,997) found that elevated midlife total cholesterol, low density lipoprotein-c, and triglycerides correlated with more extensive cognitive decline at a 20-year follow-up, after adjusting for education along with multiple clinical and demographic variables (Power, Rawlings et al. 2018).

These findings indicate that early disruptions to systemic metabolism have long-lasting and progressive effects on cognition. However, the precise metabolic mechanisms leading to cognitive impairment remain incompletely understood. Since AD develops non-cell autonomously, we propose early events leading to cognitive impairments occur in concert with a breakdown in both neuronal and glial function, particularly their metabolic crosstalk. To set the framework, we will first review the homeostatic functions of glia-neuron interactions, followed by the mechanisms leading to their breakdown during pathological conditions of metabolic dysfunction.

3. Glia-neuron interactions in the healthy brain

Under homeostatic conditions, neurons are supported by central nervous system (CNS) glia, which mainly comprise oligodendrocytes, astrocytes, and microglia. Conventionally, oligodendrocytes myelinate CNS axons to expedite signal transmission, astrocytes primarily regulate CNS blood flow and recycle neurotransmitters, and microglia, the resident immune cells, protect neurons from invading pathogens or brain damage via the inflammatory response. Recent evidence indicates, however, that glia, e.g., oligodendrocytes, also nurture axons by carefully orchestrated axo-glial metabolic crosstalk, in addition to their more traditional roles, e.g., myelin formation (Philips and Rothstein 2017). The need to metabolically aid axons is intuitive; a resting cortical neuron in the human brain expends 4.7 billion ATP molecules per second, so the energy requirements are massive (Zhu, Qiao et al. 2012). Although neurons have among the highest percent mitochondrial mass versus other cell types to meet these energy needs (Yu and Pekkurnaz 2018), they lack significant energy storage capacity and rely on continuous glucose uptake. Thus, glia supplement axons with energy substrates during periods of especially high energy need (González-Gutiérrez, Ibacache et al. 2020), e.g., during high firing rates or during neurodevelopment, making glia-neuron metabolic crosstalk a central tenet of healthy brain functioning.

3.1. Oligodendrocyte-neuron interactions

Oligodendrocytes primarily function to myelinate CNS axons, facilitating saltatory signal transmission. Oligodendrocytes wrap around axons, bringing oligodendrocytes into very close proximity to axons (Figure 1). Each oligodendrocyte can give rise to several myelin segments and support multiple axons (Philips and Rothstein 2017). Oligodendrocytes develop from oligodendrocyte progenitor cells, a process called oligodendrogenesis, which starts in embryonic stages and slows with aging (Bergles and Richardson 2015). However, oligodendrocyte progenitor cells are continuously present during life (Rivers, Young et al. 2008) and de novo myelin deposition is a dynamic process, which, if impaired, can lead to cognitive impairment (Arai 2020, Chen, Liu et al. 2021). Thus, oligodendrocyte-neuron interactions are critical to neuronal function and cognition.

Figure 1. Oligodendrocyte-neuron interactions under homeostatic conditions.

(A) Oligodendrocytes develop from the differentiation and maturation of oligodendrocyte precursor cells (OPCs) during embryogenesis and, to a lesser extent, in adulthood during myelin turn-over. The primary oligodendrocyte function is axon myelination in the CNS, which brings oligodendrocytes and axons into close contact. (B) Oligodendrocytes also metabolically support neurons. Glucose is taken up by oligodendrocytes through GLUT1 and undergoes glycolysis to pyruvate followed by conversion to lactate to provide metabolic aid to axons. Lactate is released from oligodendrocytes through MCT1 into the periaxonal space, where it enters the axon through MCT2. Oligodendrocytes also communicate with neurons by releasing EVs, which contain enzymes, such as SIRT2, a gluconeogenesis regulator, potentiating neuronal metabolic activity. (C) Oligodendrocytes use gap junctions to metabolically interact with both neurons and astrocytes by enhancing the exchange of lactate through connexins.

astro, astrocyte; CNS, central nervous system; EVs, extracellular vesicles; glu, glucose; GLUT1, glucose transporter 1; lac, lactate; MCT, monocarboxylate transporter; oligo, oligodendrocyte; OPC, oligodendrocyte precursor cell; pyr, pyruvate; SIRT2, NAD-dependent deacetylase sirtuin 2.

Although the concept of metabolic support from glia to neurons was first advanced in astrocytes, the direct intimate contact of oligodendrocytes to axons renders them especially suitable for fulfilling the metabolic requirements of neurons (Philips and Rothstein 2017). Studies in oligodendrocyte protein-specific knockout animals highlight oligodendrocyte support function beyond myelination. Mice lacking proteolipid protein (PLP) suffer axonal degeneration without changes to myelin compaction (Griffiths, Klugmann et al. 1998). In contrast, myelin basic protein (MBP) knockout does not induce axonal degeneration in mice but impairs myelin compaction (Griffiths, Klugmann et al. 1998). Further, identification of myelin-to-axon cytoplasmic channels, which are dependent on the oligodendrocyte protein 2’,3’-cyclic nucleotide 3’-phosphodiesterase (CNP), suggests a potential route for transfer of contents to the axon (Snaidero, Velte et al. 2017). Overall, these findings demonstrate that axon degeneration can occur independent of myelin compaction, indicating a purpose for oligodendrocyte proteins, such as PLP, beyond myelin structure (Griffiths, Klugmann et al. 1998).

This view was reinforced by the discovery of the essential role of monocarboxylate transporters (MCTs) to axon health (Figure 1). In a seminal paper, Lee et al. found MCT1 was highly enriched in oligodendrocytes, which they suggested was an important route oligodendrocytes leverage to metabolically support axons (Lee, Morrison et al. 2012). In vivo, oligodendrocyte-specific MCT1 knockdown induces axon degeneration. Further, MCT1 inhibition ex vivo in organotypic spinal cord slices induces motor neuron death under glucose starvation, which is rescued by supplementing the culture media with lactate. Since lactate supplementation rescues the effect of MCT1 inhibition, the authors suggested that MCT1 transport of lactate from the oligodendrocyte to the axon is essential to neuronal health.

Inhibiting MCT1, 2, and 4 in the hippocampus in vivo impairs long-term memory formation, which can be rescued by lactate supplementation in the case of MCT1 and 4 inhibition, but not MCT2 inhibition (Suzuki, Stern et al. 2011). These data indicate a substantial reliance of neurons on MCT-mediated lactate uptake for memory retention (Figure 1). Indeed, additional studies suggest neurons may even prefer lactate as an energy source over glucose, in both the intact (Wyss, Jolivet et al. 2011) and injured brain (Glenn, Martin et al. 2015). Additionally, inhibiting MCT1/MCT2 lowers axonal ATP levels in electrically stimulated axons (50 and 100 Hz) and alters the ATP to compound action potential ratio, lowering the capacity for electrical conduction (Trevisiol, Saab et al. 2017). Therefore, a growing body of evidence supports the concept that MCT-mediated lactate shuttling occurs from oligodendrocytes to axons and that lactate is an important energy source in the CNS, both in oligodendrocytes and neurons.

There are multiple examples of the role of lactate as an essential energy source in the brain (Hu and Wilson 1997), both for oligodendrocytes as well as for neurons, directly and through oligodendrocyte-mediated lactate transfer. For example, lactate supplementation can rescue low-glucose-induced impairment in oligodendrocyte differentiation in rat cerebellar or cortical slices, suggesting lactate is an important substrate during oligodendroglial myelination (Rinholm, Hamilton et al. 2011). In parallel, blocking oxidative phosphorylation in mature oligodendrocytes by COX10 knockout shifts metabolism to glycolysis and lactate utilization as a readily available energy source, which has no effect on myelination (Fünfschilling, Supplie et al. 2012). Additionally, oligodendrocytes may import lactate independent from MCT1 through gap junctions composed of connexin hemichannels (Philips and Rothstein 2017). Knocking out the expression of connexin hemichannels, which are mainly expressed by oligodendrocytes, induces neuronal vacuolation and variable degrees of dysmyelination (Odermatt, Wellershaus et al. 2003). Even connexin hemichannel knockout in astrocytes triggers myelination defects and cognitive impairment in mice, suggesting the presence of gap junction crosstalk between astrocytes and oligodendrocytes (Lutz, Zhao et al. 2009).

In addition to MCTs and the lactate axis, oligodendrocyte N-methyl-D-aspartate receptors (NMDARs) sense neuron-derived glutamate, a surrogate of signal transmission activity, which regulates the metabolic support offered by oligodendrocytes (Saab, Tzvetavona et al. 2016). Specifically, oligodendroglial NMDAR activation in optic nerves ex vivo enhances calcium influx and glucose uptake by oligodendrocytes via glucose transporter 1 (GLUT1), in turn augmenting lactate transfer to axons. Thus, an increase in neuronal transmission activates biochemical pathways in oligodendrocytes, which stimulates lactate transfer to energy-requiring axons through metabolic crosstalk.

Oligodendrocytes can additionally communicate with axons by secreting extracellular vesicles (EVs), which enter the periaxonal space and are internalized by neurons via endocytosis (Figure 1) (Frühbeis, Kuo-Elsner et al. 2020). This EV-mediated oligodendrocyte-axon crosstalk enhances neuronal metabolism (Frühbeis, Fröhlich et al. 2013), firing rates (Fröhlich, Kuo et al. 2014), and axonal transport (Frühbeis, Kuo-Elsner et al. 2020). Interestingly, PLP- and CNP-deficient oligodendrocytes secrete dysfunctional EVs, which are incapable of neuronal or axonal support (Frühbeis, Kuo-Elsner et al. 2020). Oligodendrocyte-derived EVs also transfer SIRT2 cargo, an NAD+-dependent deacetylase and regulator of gluconeogenesis, to axons, increasing axonal basal respiration and ATP production and levels in maturing neurons (Chamberlain, Huang et al. 2021). SIRT2 then boosts mitochondrial bioenergetics by deacetylating axonal mitochondria proteins, indicative of oligodendrocyte-axon metabolic crosstalk independent of myelin function.

Overall, evidence indicates that oligodendrocytes support neurons beyond myelin formation through metabolic coupling. This enforces the critical dependence of neurons on oligodendrocytes for the required energy substrates to maintain normal neuronal function and cognition.

(Frühbeis, Fröhlich et al. 2013, Fröhlich, Kuo et al. 2014, Frühbeis, Kuo-Elsner et al. 2020)

3.2. Astrocyte-neuron interactions

Astrocytes primarily function to regulate CNS blood flow and recycle neurotransmitters. They help neurons maintain synaptic transmission and excitability via the primary astrocytic recycling pathway, the glutamate-glutamine cycle (Figure 2) (Bélanger, Allaman et al. 2011). Using this pathway, astrocytes take up glutamate via glutamate transporter 1, and convert it to glutamine which is transferred to neurons through the glutamate aspartate transporter. Neurons then convert glutamine back to glutamate, replenishing the neuronal neurotransmitter pool. Astrocytes interact directly with neuronal synapses, so-called tripartite synapses, to facilitate synaptic plasticity and transmission (Mederos, González-Arias et al. 2018). Hippocampal astrocytes also express functional acetylcholine receptors (Gahring, Persiyanov et al. 2004, Shen and Yakel 2012) and modulate neuronal cholinergic firing rates (Pabst, Braganza et al. 2016). Thus, astrocytes may modulate neuronal activity through glutamate and cholinergic neurotransmitters (Maurer and Williams 2017).

Figure 2. Astrocyte-neuron interactions under homeostatic conditions.

(A) The primary function of astrocytes is to regulate CNS blood flow and recycle neurotransmitters. (B) Astrocytes take up glucose from blood vessels, which is metabolized to pyruvate by glycolysis, and then, in turn, to lactate. Astrocytes leverage the MCT1/MCT2 shuttle to transfer lactate to neurons, providing metabolic support to axons. (C) Moreover, astrocytes serve as an energy reserve through glycogen storage. (D) Astrocytes also play important roles through the recycling of glutamate/glutamine. First, astrocytes take up glutamate via GLT-1, which is converted to glutamine and transported to neurons via GLAST. Neurons can then convert glutamine back to glutamate to replenish their neurotransmitter pool. (E) Finally, astrocytes provide antioxidants to neurons, which prefer oxidative metabolism and generate relatively high reactive oxygen species levels.

astro, astrocyte; CNS, central nervous system; GLAST, glutamate aspartate transporter; gln, glutamine; glt, glutamate; glu, glucose; GLT-1, glutamate transporter 1; lac, lactate; MCT, monocarboxylate transporter; oligo, oligodendrocyte; pyr, pyruvate.

Astrocyte-neuron synapse interactions also couple neurons to the vasculature, facilitating astrocyte-coordinated changes to CNS blood flow and metabolite supply to the CNS through the vasculature (Bélanger, Allaman et al. 2011). Astrocyte-mediated changes in blood flow are influenced, at least in part, by brain metabolism (Gordon, Choi et al. 2008), supporting metabolic links between neurons and astrocytes. Astrocytes residing near blood vessels increase glucose uptake from circulation in response to neuronal transmission and disseminate glucose and metabolites through connexins across astrocytic networks (Rouach, Koulakoff et al. 2008).

In terms of metabolism, both astrocytes and neurons can use glucose and lactate as energy sources; however, they prefer different, though complementary, energy sources and pathways. Astrocytes primarily utilize glycolysis and produce lactate (Lovatt, Sonnewald et al. 2007), whereas neurons prefer oxidative metabolism (Bélanger, Allaman et al. 2011). Complementary energy generating pathways are critical to astrocyte-neuron metabolic interactions. Astrocytes consume a large portion of the brain glucose supply, especially during neural activation (Chuquet, Quilichini et al. 2010). However, neurons require a disproportionally large amount of energy to function versus other CNS cell types. Metabolic interaction partially explains this discrepancy; astrocytes sense glutamate, a surrogate of neuronal activity, which enhances their glucose consumption and subsequent lactate release to neurons through the astrocyte-neuron lactate shuttle (Pellerin and Magistretti 1994). Thus, elevated astrocytic glucose metabolism serves to enhance lactate substrate transport into neurons, amplifying their energy reserves and aiding electrical activity. It is important to note that neuron and astrocyte populations are heterogeneous, varying in number and type by brain region and multiple additional factors; thus, their interactions likely also vary (Khakh 2019).

While neurons prefer oxidative metabolism and generate relatively high amounts of reactive oxygen species, they are limited in intrinsic antioxidant production rendering them particularly susceptible to oxidative damage (Figure 2). Astrocytes are less vulnerable to oxidative stress and also assist neurons partly by providing them with antioxidant precursors, e.g., glutathione (Bélanger, Allaman et al. 2011), underscoring another instance of astrocyte-neuron interactions involving metabolism. The antioxidant glutathione is regenerated from its oxidized form using electrons from NADPH, which is itself generated from glucose processing via the pentose phosphate pathway (Dringen 2000). Oxidative stress upregulates pentose phosphate pathway activity and NADPH levels in astrocytes (García-Nogales, Almeida et al. 2003), augmenting their antioxidant capacity and ability to support neurons. Additionally, increased brain activity stimulates a shift in energy utilization from primarily glucose to lactate consumption in neurons concomitant with enhanced antioxidant ascorbic acid release from astrocytes (Castro, Beltrán et al. 2009), providing an additional route to astrocyte-mediated metabolic support.

Primarily located in astrocytes, glycogen is an important energy storage reserve in the brain that can protect neurons under hypoglycemic conditions (Brown 2004). These astrocytic glycogen reserves are also a lactate source to neurons via the astrocyte-neuron lactate shuttle, further providing metabolic support to neurons and synaptic activity (Tekkök, Brown et al. 2005). Blocking astrocytic glycogenolysis impedes memory consolidation and long-term memory formation, implying bioenergetic support from glycogen is critical to cognitive function (Suzuki et al., 2011). Exogenous lactate reverses these effects, unless the neuronal lactate transporter MCT2 is blocked, indicating a critical role for astrocyte-neuronal metabolic lactate signaling long-term memory consolidation. Lastly, in addition to the significant metabolic interaction between astrocytes and neurons, astrocytes are also key mediators of metabolic support of oligodendrocytes to neurons, forming a neuron-oligodendrocyte-astrocyte axis (Amaral, Meisingset et al. 2013). The integrity of this critical trio is paramount to normal brain function.

3.3. Microglia-neuron interactions

Microglia are resident innate immune cells of the brain; however, their function extends well beyond immune surveillance and response to pathogens. During development, microglia support neurons by regulating neuronal survival and differentiation, eliminating non-viable neurons and pruning synapses (Figure 3) (Schafer and Stevens 2015, Mosser, Baptista et al. 2017). Microglia-neuron crosstalk continues in the adult brain, serving a variety of purposes, including sensing neuronal activity and regulating synaptic plasticity (Marinelli, Basilico et al. 2019). Microglia-neuron crosstalk is mediated by both direct contact (Cserép, Pósfai et al. 2021) and secreted signals (Marinelli, Basilico et al. 2019). Microglia extend cellular processes that constantly surveil their environment, secret soluble signals that bind neuronal receptors, and express receptors to receive neuronal signals (Pósfai, Cserép et al. 2019, Uweru and Eyo 2019). Microglial processes communicate with neurons at multiple cellular anatomical locations, including somas, axons, and dendrites, allowing microglia to sense and respond to neuronal activity directly (Cserép, Pósfai et al. 2021). Here, we highlight the main mediators and key functions of microglial-neuronal crosstalk in development and the healthy adult brain.

Figure 3. Microglia-neuron interactions under homeostatic conditions.

The primary microglia function is as the resident innate immune cells of the brain; however, their function extends well beyond immune surveillance. (A) Microglia clear cellular debris from the CNS and eliminate non-viable neurons. (B) Microglia also prune neuronal synapses through C3-C3R interactions. (C) In addition, microglia-neuron interactions occur through other receptor-ligand interactions (CX3CL1-CX3CR1, CD200-CD200R, anti-inflammatory and house-keeping functions) and microglia-secreted neurotransmitters, neurotrophic factors, and cytokines, which bind to cognate receptors on neurons. (D) Regarding metabolic communication, one putative mechanism is through junction formation of neuronal somas to microglial purinergic receptors, e.g., P2RY₁₂, which are activated by ATP and may constitute a sensing mechanism of neuronal activity since these junctions are enriched with neuronal mitochondria and endoplasmic reticulum contacts. (E) Microglia prefer glucose as an energy substrate, which they take up through GLUTs, and metabolize through oxidative phosphorylation (OXPHOS) under homeostatic conditions (pink microglia). However, even though less efficient, activated microglia (red microglia) shift their metabolism to glycolysis while mounting a pro-inflammatory response. One putative reason is that microglia may utilize glycolysis-derived NADPH to generate reactive oxygen species for host defense.

C3, complement component 3; C3R, C3 receptor; CNS, central nervous system; CD200, Cluster of Differentiation 200 ligand; CD200R1, CD200 receptor 1; CX3CL1, fractalkine also called chemokine (C-X3-C motif) ligand 1; CX3CR1, fractalkine receptor also called CX3C chemokine receptor 1; glu, glucose; GLUT, glucose transporter; OXPHOS, oxidative phosphorylation; P2RY₁₂, purinergic receptor.

Complement signaling though complement receptor 3 (C3R) on microglia and C3 from neurons contributes to developmental synaptic pruning (Figure 3) (Schafer, Lehrman et al. 2012). In adulthood, aberrant complement activation is implicated in neurodegenerative disease (Dalakas, Alexopoulos et al. 2020). Other receptor-ligand pairs essential for homeostatic microglial-neuronal crosstalk involve fractalkine (CX3CL1; neurons) and the fractalkine receptor (CX3CR1; microglia), and CD200 ligand (neurons) and CD200 receptor (CD200R; microglia), which perform anti-inflammatory and house-keeping roles (Marinelli, Basilico et al. 2019).

Microglia and neurons also communicate via secreted neurotransmitters, neurotrophic factors, cytokines, purines, and the purine derivative ATP (Marinelli, Basilico et al. 2019). Microglia express serotonin, GABA_B, and glutamatergic receptors, which allows them to sense neuronal activity (Marinelli, Basilico et al. 2019). They also express acetylcholine receptors, which, upon activation, exert anti-inflammatory and neuroprotective effects (Suzuki, Hide et al. 2006, Egea, Buendia et al. 2015, Li, Liu et al. 2019). Microglia also secrete neurotrophic factors (e.g., brain-derived neurotrophic factor) and cytokines (e.g., tumor necrosis factor alpha [TNF-α], interleukin-1β [IL-1β]), which bind to cognate receptors expressed on neurons (Pósfai, Cserép et al. 2019). Purinergic receptor activation, e.g., P2XRs and P2YRs, evoke diverse microglial responses, such as migration upon ATP stimulation of P2RY₁₂ (Figure 3) (Calovi, Mut-Arbona et al. 2019, Illes, Rubini et al. 2020). Microglial purinergic receptors form purinergic junctions with neuronal somas (Cserép, Pósfai et al. 2020). The exact function of these purinergic junctions remains under investigation, but microglia may leverage them to sense neuronal activity, since these junctions are comprised of neuronal mitochondria and endoplasmic reticulum contacts.

Diverse interactions between microglia and neurons contribute to proper nervous system development and help maintain homeostasis in the adult brain. In addition to interactions concerning neurotransmitters, neurotrophic factors, cytokines, and purinergic receptors, recent evidence also suggests potential for a metabolic aspect to microglia-neuron interactions. Most studies have defined the relationship between microglial activation state and metabolism, whereas less is known about microglia-neuron metabolic crosstalk. Microglia express multiple GLUT receptors and utilize glucose as their main fuel source, preferentially using oxidative phosphorylation when in a homeostatic state (Bernier, York et al. 2020, Lauro and Limatola 2020). However, although glucose is preferred, microglia can utilize glutamine, and possibly free fatty acids and amino acids, as fuel sources, making them metabolically versatile when glucose levels are low or under stressful conditions, for example when mounting an inflammatory response (Bernier, York et al. 2020, Bernier, York et al. 2020).

When microglia respond to a pro-inflammatory challenge, such as lipopolysaccharide stimulation, they transition from oxidative phosphorylation to glycolysis (Lauro and Limatola 2020). This metabolic switch is also observed in activated peripheral macrophages (Liu, Xu et al. 2021). In a recent review on the topic, Bernier et al. proposed microglia require the glycolytic shift to mount a pro-inflammatory response (Bernier, York et al. 2020). Specifically, microglia might benefit from a glycolytic shift because NADPH can be used to generate reactive oxygen species for host defense, and metabolic intermediates can contribute to proliferation and cytokine production. Indeed, inhibiting glycolysis blocks lipopolysaccharide-induced primary microglial TNF-α, IL-1β, and IL-6 expression (Hu, Mai et al. 2020). Thus, the metabolic state of microglia regulates inflammatory signals, which can communicate with surrounding cells, including neurons.

How microglia might sense neuronal metabolism remains largely unknown. Microglia sense ATP- and activity-dependent neurotransmitter release from neurons, so, perhaps, they indirectly sense metabolic demands. One potential source of direct metabolic crosstalk exists at the microglial to neuronal somatic purinergic junction, where neuronal mitochondria aggregate, suggesting a potential mechanism of metabolic communication (Cserép, Pósfai et al. 2020).

Overall, microglial metabolic shifts are critical for pro-inflammatory activation, whereas microglia-neuron metabolic crosstalk remains poorly understood. However, given emerging evidence of potential metabolic communication, e.g., via purinergic receptors, research into microglial-neuron metabolic crosstalk constitutes an interesting research direction.

4. Glia-neuron interactions in the brain in the context of the MetS and dysfunctional metabolism

In the healthy brain, carefully orchestrated metabolic glia-neuron interactions occur to sustain normal brain functioning (Philips and Rothstein 2017). Under conditions of the MetS, rising insulin resistance and excess energy substrates, both elevated glucose and lipid levels, perturb glial homeostasis and metabolism and induce neuroinflammation (Van Dyken and Lacoste 2018, de la Monte and Grammas 2019, O’Grady, Dean et al. 2019, Langley, Yoon et al. 2020, Bouhrara, Khattar et al. 2021). Additionally, metabolic dysfunction lowers brain expression of neurotransmitter receptors, e.g., acetylcholine receptors (Xu, Cao et al. 2020, Martinelli, Tomassoni et al. 2021, Martins, Contieri et al. 2021), which would lead to cognitive impairment. In parallel to MetS-induced glial disruption, we posit perturbations in glia-neuron interactions occur with a loss of energy substrate transfer to neurons in need of metabolic support. Ultimately, this would lead to failure of neuronal bioenergetics and signal transmission, and, eventually, neuronal loss and cognitive impairment. The following section summarizes the available studies on glia-neuron metabolic crosstalk occurring in the context of dysfunctional metabolism.

4.1. Oligodendrocyte-neuron interactions

As discussed in the section on homeostatic glia-neuron interactions, oligodendrocytes provide two pivotal supporting functions for neurons, myelination and metabolic support (Philips and Rothstein 2017). With reference to myelination, diabetes, obesity, and MetS components negatively impact oligodendrocytes, leading to oligodendrocyte loss and loss of myelin integrity (Yoon, Kleven et al. 2016, Kim, Langley et al. 2020). These pathological changes correlate with cognitive impairment in preclinical studies. In mouse models of obesity, high-fat diet (HFD) decreases myelin thickness, which correlates with poorer cognitive performance (Graham, Grabowska et al. 2019). Similarly, in type 2 diabetic mice, loss of white matter, a marker of brain demyelination, is associated with worsened cognitive function (Li, Guo et al. 2019). In human clinical studies, patients with type 2 diabetes, like their murine counterparts, exhibit disrupted white matter networks that correlate with cognitive impairment (Zhang, Liu et al. 2016, Biessels and Despa 2018). In parallel, obesity and insulin resistance are linked to lower myelin content in cognitively unimpaired adults (O’Grady, Dean et al. 2019, Bouhrara, Khattar et al. 2021).

With reference to metabolic support, metabolomics analysis of central nervous system tissue from MetS animals fed a HFD demonstrates impaired metabolism occurs concurrently with oligodendrocyte loss, with a drop in tricarboxylic acid (TCA) cycle intermediates and changes in protein biosynthesis, glutathione metabolism, and the mitochondrial electron transport chain (i.e., oxidative phosphorylation) (Langley, Yoon et al. 2020). These detrimental changes worsen over time. HFD feeding also promotes the loss of oligodendrocyte progenitor cells and reduces their differentiation in mouse models (Langley, Yoon et al. 2020). In a db/db mouse model of obesity and type 2 diabetes, early changes in myelin and mitochondrial lipids occur in the brain prior to the onset of overt structural alterations (Palavicini, Chen et al. 2020). Although these studies in mouse models on oligodendrocytes are correlative and not causative, they suggest that systemic metabolic dysfunction adversely impact oligodendrocyte health, and possibly neuronal health in turn, and constitute interesting research avenues. Importantly, there is discordance across studies, which may arise from the rodent age, location of the sampled CNS matter, and differences in diet, which also require further inquiry.

The impact of hyperglycemia on oligodendrocytes is unclear. Chronic hyperglycemia does not affect viability, oxidative stress, or differentiation of oligodendrocyte progenitor cells in vitro (da Rosa, Meira et al. 2019). Yet a high-fat high-sucrose diet decreases the number of mature myelinating oligodendrocytes in mouse spinal cord in tandem with impaired TCA metabolism (Kim, Langley et al. 2020). Therefore, it is possible that mature oligodendrocytes, but not their progenitors, are susceptible to hyperglycemia. This idea is supported by data from type 1 diabetic rats with hyperglycemia. In these animals, the optic nerve is characterized by disorganized myelin and some demyelinated zones, indicating a potential oligodendrocyte loss and impaired function (Dorfman, Aranda et al. 2015).

Regarding potential effects of metabolic dysfunction on oligodendrocyte-neuron metabolic crosstalk, neuronal expression of components of the lactate shuttle, MCT1 and MCT2, increase in the brain of obese mice after 12 weeks of HFD, as well as in the brain of type 2 diabetes and obesity mouse models (ob/ob and db/db mice) (Pierre, Parent et al. 2007). These findings likely represent an initial compensatory mechanism by the brain to overcome the bioenergetic crisis produced by systemic MetS conditions (Chomova 2022). More recent MCT research has addressed changes in AD murine models over time, but data are lacking on HFD and MetS animals. MCT1, MCT2, and MCT4 expression and lactate levels decrease in APP/PS1 AD mice with cognitive impairment (Zhang, Cheng et al. 2018), supporting an association between aberrant oligodendrocyte-neuron coupling and cognition. This association is further supported by studies showing brain MCT1 and MBP levels progressively decrease in older versus younger APP/PS1 AD mice, paralleling the trajectory of cognitive decline (Dong, Zhang et al. 2018). Finally, AD also directly affects oligodendrocyte metabolism by altering the expression of glycolytic and ketolytic genes, impairing their ability to provide metabolic support and energy substrates to neurons (Saito, Miller et al. 2021). Analogous studies are required in MetS animal models to improve our understanding of systemic metabolic dysfunction on oligodendrocyte-neuron metabolic crosstalk.

It is also possible to draw parallels between oligodendrocyte-neuron interactions with Schwann cell-neuron coupling in the peripheral nervous system. Schwann cell-restricted MCT1 knockout triggers sensory neuropathy and hypomethylation in aging mice along with perturbed myelin lipid composition (Jha, Lee et al. 2020). Heterozygous MCT1 knockout worsens sensory and motor neuropathy and causes a thinning of myelin in STZ type 1 diabetic mice (Jha, Ament et al. 2020). In models of peripheral nerve injury, Schwann cells shift their metabolism to glycolysis to supply neurons with energy fuel during repair, which specifically occurs through MCT1 and Schwann cell-axon metabolic coupling (Babetto, Wong et al. 2020). Indeed, heterozygous MCT1 knockout (Morrison, Tsingalia et al. 2015) or pharmacological MCT1 inhibition (Babetto, Wong et al. 2020) impairs axon repair and/or accelerate degeneration. Cumulatively, these studies demonstrate how Schwann cells metabolically support neurons during pathological conditions, be it during diabetes- or mechanical injury-induced axon degeneration. Whether similar mechanisms operate in the CNS has not been investigated in this level of detail, to our knowledge.

In summary, diabetes, obesity, and MetS components disrupt oligodendrocyte metabolism and myelin integrity (Yoon, Kleven et al. 2016, Kim, Langley et al. 2020), which we posit leads to a failure to provide adequate metabolic support to neurons and cognitive impairment (Figure 4).

Figure 4. Oligodendrocyte-neuron interactions under pathologic conditions of metabolic dysfunction.

(A) Diabetes, obesity, and MetS components negatively impact oligodendrocytes, which correlate with cognitive impairment. HFD promotes obesity and the MetS and induces oligodendrocyte loss and perturbed myelin structure and lipid composition. (B) Additionally, transcriptomics and metabolomics changes occur to multiple targets related to metabolism (drop in TCA cycle intermediates), mitochondrial biogenesis and function (changes in PGC-1α expression, drop in OXPHOS), and protein biosynthesis, along with upregulated endoplasmic reticulum stress and oxidative stress pathways (glutathione metabolism). (C) We hypothesize HFD also stimulates changes to MCT1, 2, and 4 expression and that oligodendrocytes experience a glycolytic shift to supply neurons with energy fuel under conditions of stress via metabolic coupling. (D) Lastly, HFD also promotes the loss of OPCs and reduces their differentiation, which would impair the ability to regenerate HFD-induced damage to myelin.

glu, glucose; HFD, high-fat diet; lac, lactate; MCT, monocarboxylate transporter; MetS, metabolic syndrome; OPC, oligodendrocyte precursor cell; OXPHOS, oxidative phosphorylation; PGC-1α, peroxisome proliferator-activated receptor gamma coactivator 1-alpha; TCA, tricarboxylic acid.

4.2. Astrocyte-neuron interactions

Astrocytes are central to normal brain physiology, glucose homeostasis, and energy regulation. Glucose is the primary energy source for the brain, and while neurons have high energy requirements, astrocytes are the principal cells responsible for glucose uptake and transfer of metabolic substrates, particularly lactate, to neurons. Although insulin is not required for glucose uptake by neurons, the brain is an insulin sensitive organ (Cai, Xue et al. 2018, Kim, Elzinga et al. 2019) and develops insulin resistance in response to MetS, with changes in brain structure (Lu, Aziz et al. 2021) (Figure 5).

Figure 5. Astrocyte-neuron interactions under pathologic conditions of metabolic dysfunction.

Diabetes, obesity, and MetS components negatively impact astrocytes and correlate with cognitive impairment. (A) Obesity promotes astrocyte reactivity (red astrocytes), morphologic changes, and inflammation. (B) Insulin resistance in the brain impairs the ability of astrocytes to take up glucose and maintain CNS glucose homeostasis. (C) Diabetes perturbs astrocytic metabolism. (D) Obesity and diabetes also impair the ability of astrocytes to clear glutamate, which subsequently suppresses neuronal transmission, and leads to excitotoxicity.

CNS, central nervous system; GLAST, glutamate aspartate transporter; gln, glutamine; glt, glutamate; glu, glucose; GLT-1, glutamate transporter 1; IR, insulin resistance; lac, lactate; MCT, monocarboxylate transporter; pyr, pyruvate; TCA, tricarboxylic acid.

Astrocytes express insulin receptors and are insulin responsive, increasing glycogen storage following insulin treatment (Heni, Hennige et al. 2011). In vivo studies using astrocyte-specific insulin receptor knockout mice show that astrocytes lacking insulin receptors are less capable of maintaining CNS glucose homeostasis, secondary to loss of hypothalamic astrocyte function, disrupting normal physiological responses to brain glucose levels (García-Cáceres, Quarta et al. 2016). This is paralleled by a loss of normal glucose transport across the blood-brain barrier in these same animals, further emphasizing the crucial role astrocytes play in sensing systemic glucose (García-Cáceres, Quarta et al. 2016). Insulin receptor knockout in astrocytes lowers ATP release and disrupts astrocyte-neuron energy exchange, which decreases dopamine release from dopaminergic neurons (Cai, Xue et al. 2018). This decreased dopamine release impacts cognition since knockout animals display depressive and anxiety. These preclinical observations support a growing literature on how systemic insulin resistance, a hallmark of dyslipidemia and the MetS, in parallel with aforementioned brain insulin resistance, develops over time and is associated with cognitive impairment and dementia (Kellar and Craft 2020).

In the context of the MetS and astrocytes, two recent reports provide additional insight into this critical association. First, treating primary astrocytes with fatty acids to simulate dyslipidemia present in the MetS, lowers autophagic flux in astrocytes, a response likely dependent upon the brain region from which cells are isolated (Ortiz-Rodriguez and Arevalo 2020). This response, along with blocking autophagy in astrocytes, is toxic, both to astrocytes and neurons. The second report highlights an intriguing response of the sympathetic nervous system to HFD animal models. HFD fed rats exhibit sympathetic neuron excitotoxicity, with increased astrocyte leptin receptor expression and decreased glutamate receptor and transporter expression (Liu and Zheng 2019). Interestingly, changes in sympathetic nervous system function are common in obesity, components of the MetS, and dysfunctional metabolism (Liu and Zheng 2019). The link between elevated sympathetic activity, cognitive impairment, and increased astrocyte leptin receptor expression is an interesting one deserving of further study (Knight, Giuliano et al. 2020). These two studies open new avenues of research as the field pursues a deeper understanding of the metabolic crosstalk between astrocytes and neurons during the MetS.

In comparison to the few studies outlined above, there is more established literature on the pathogenesis of AD and related dementias regarding changes in bioenergetics, mitochondrial function, and inflammation in astrocytes (Rodríguez-Arellano, Parpura et al. 2016, Arranz and De Strooper 2019). These changes in astrocyte physiology parallel those reported in astrocytes in response to obesity and systemic dysfunction, with well documented increases in inflammation and astrocyte reactivity, coupled with changes in morphology and function (Tomassoni, Nwankwo et al. 2013, Koga, Kojima et al. 2014, Tomassoni, Martinelli et al. 2020). What is less well studied is astrocyte biology in the context of diabetes-mediated cognitive impairment. While published research strongly supports our contention that diabetes will induce dysfunctional astrocyte-neuron metabolic coupling, there are only a few studies directly addressing this question. One such study is in the db/db mouse model of type 2 diabetes. Zhang et al. report increases in brain lactate and alanine levels and speculate these findings may signal a breakdown of the lactate-alanine shuttle between astrocytes and neurons, while concurrent changes in TCA metabolites suggest a metabolic switch in neurons from oxidative metabolism to anaerobic glycolysis (Zheng, Zheng et al. 2017). The authors also report a decrease astrocyte clearance of glutamate, which subsequently suppresses GABA transmission to neurons and impairs synaptic plasticity. These findings associate with cognitive impairment in the db/db animals, as assessed by the Morris water maze (Zheng, Zheng et al. 2017).

In summary, accumulating data support the hypothesis that metabolic imbalance between astrocytes and neurons promotes cognitive impairment in obesity and the MetS. Multiple preclinical and clinical studies also confirm that obesity, diabetes, and MetS components promote cognitive impairment, and that changes in astrocyte-neuron metabolic interactions are critical in other neurodegenerative diseases, including AD (Zulfiqar, Garg et al. 2019). Furthermore, several reports point to astrocyte-neuron metabolic coupling as critical for learning and memory under physiologic conditions. While more studies are needed to fully understand astrocyte-neuron metabolic interactions in cognitive impairment related to the MetS and metabolic dysfunction (Figure 5), astrocytes are attractive therapeutic targets and the source of current and planned interventions for the treatment of cognitive impairment and associated dementias (Arranz and De Strooper 2019).

4.3. Microglia-neuron interactions

Components of the MetS promote CNS neuroinflammation (Van Dyken and Lacoste 2018), with microglial activation, elevated cytokine levels, oxidative stress, and blood-brain barrier disruption along with peripheral immune trafficking into the brain. Neuroinflammation contributes to pathology through activated microglia interactions with hypothalamic neurons leading to leptin and insulin resistance (Van Dyken and Lacoste 2018, Robb, Morrissey et al. 2020) and with hippocampal neurons leading to cognitive decline (Cope, LaMarca et al. 2018). The role of microglia-neuron crosstalk in regulating metabolic disease is relatively well-established in the hypothalamus. However, the role of microglial metabolism and metabolic crosstalk in the hippocampus and on cognitive performance is not as well described.

Hypothalamic microglial pro-inflammatory activation contributes to neuronal stress and ultimately drives feeding behaviors and diet-induced obesity (Valdearcos, Robblee et al. 2014, Valdearcos, Douglass et al. 2017). Although the mechanism of the crosstalk remains elusive, Valdecaros et al. demonstrated that diphtheria toxin depletion of hypothalamic microglia prevents saturated fatty acid-induced neuronal stress (Valdearcos, Robblee et al. 2014). Additionally, pharmacological microglial depletion in the context of saturated fatty acid treatment increases neuronal leptin responses and decreases intake of chow. In a later study, the authors show that inhibiting microglia-specific NF-κB inflammatory activation by genetic manipulation prevents HFD-induced hyperphagia and obesity (Valdearcos, Douglass et al. 2017). Microglia, therefore, regulate systemic metabolic physiology by interacting with neurons in the medial basal hypothalamus (Robb, Morrissey et al. 2020). Along those lines, microglia also express leptin receptors, and knockout of myeloid leptin receptors reduces microglial morphological ramification in the hypothalamic paraventricular nucleus, disrupts hypothalamic neuronal circuitry, and induces hyperphagia and weight gain (Gao, Vidal-Itriago et al. 2018). This advocates a potential role for microglial leptin sensing in regulation of hypothalamic neurons and subsequent systemic metabolism. Dysregulated mitochondrial metabolism has further been implicated in HFD-induced hypothalamic microglial activation and subsequent obesity (Kim, Yoon et al. 2019). Finally, although microglia are the predominant activated immune cell contributing to hypothalamic neuroinflammation (Boura-Halfon, Pecht et al. 2019), macrophages, the peripheral equivalent of microglia, also traffic into the CNS during the MetS (Van Dyken and Lacoste 2018, Yang, Graham et al. 2019), including the hypothalamus based on CD45^high expression (Lainez, Jonak et al. 2018).

The influence of HFD consumption and obesity on microglia-neuron crosstalk in brain regions responsible for cognitive function and memory has not been as well studied as in the hypothalamus. The hippocampus, a limbic structure contributing to memory and learning tasks, displays microglial activation in mouse models of diet-induced obesity (Hao, Dey et al. 2016, Cope, LaMarca et al. 2018). Obesity is associated with hippocampal-dependent cognitive impairment in rodent models (Sobesky, Barrientos et al. 2014, Sims-Robinson, Bakeman et al. 2016, Cope, LaMarca et al. 2018). Moreover, microglia phagocytose synaptic spines, contributing to hippocampal-dependent cognitive impairment (Cope, LaMarca et al. 2018). While microglia play a role in obesity-induced cognitive impairment, the mechanisms of microglia-neuron crosstalk and the role of metabolism in this communication are unknown. Neuronal fractalkine and the microglial fractalkine receptor are decreased in a cognitively impaired diet-induced obese mouse models, and this dysregulated microglia-neuron interaction may contribute to cognitive impairment (Kawamura, Katsuura et al. 2021). However, fractalkine receptor deficiency using CX3CR1^+/− mice in a diet-induced obesity model prevents microglial activation and hippocampal dependent deficits (Cope, LaMarca et al. 2018). Lastly, macrophages also infiltrate into the hippocampus and contribute to neuroinflammation (Buckman, Hasty et al. 2014, Erion, Wosiski-Kuhn et al. 2014). Obesity-induced activation of NLR family pyrin domain containing 3 (NLRP3), a macrophage inflammasome component, in peripheral visceral adipose depots stimulates hippocampal microglia, contributing to cognitive impairment (Guo, Yamamoto et al. 2020).

While the data collectively support a critical role for microglia in cognition (Figure 6), future studies are needed to assess the role of microglial-neuron metabolic interactions on cognitive impairment.

Figure 6. Microglia-neuron interactions under pathologic conditions of metabolic dysfunction.

(A) Diabetes, obesity, and MetS components activate microglia, which leads to neuronal stress. (B) In the hippocampus, HFD induces aberrant and excessive microglial phagocytosis of synaptic spines, contributing to hippocampal-dependent cognitive impairment. (C) HFD decreases neuronal CX3CL1 and microglial CX3CR1, contributing to hippocampal dependent cognitive impairment.

CX3CL1, fractalkine also called chemokine (C-X3-C motif) ligand 1; CX3CR1, fractalkine receptor also called CX3C chemokine receptor 1; HFD, high-fat diet.

5. Conclusions

Multiple clinical studies report diabetes, obesity, and MetS components are associated with cognitive impairment ranging from MCI to dementias, such as AD (Mallorquí-Bagué, Lozano-Madrid et al. 2018). These findings underscore the importance of metabolism in maintaining healthy cognitive function. Under normal physiological conditions, glia perform various supportive functions for neurons ranging from myelination and lactate supplementation by oligodendrocytes (Philips and Rothstein 2017), replenishing of the neurotransmitter pool and energy storage and antioxidant supplementation by astrocytes (Bélanger, Allaman et al. 2011), and synaptic pruning and immune functions by microglia (Mosser, Baptista et al. 2017). Further, since neurons cannot store a significant amount of energy, they rely on glia for continuous metabolic support. States of diet-induced obesity or dysregulated metabolism lead to multiple pathologic changes in glia, including oligodendrocyte loss and impaired myelination (Kim, Langley et al. 2020), changes in astrocyte autophagy (Ortiz-Rodriguez and Arevalo 2020) and neurotransmitter release (Zheng, Zheng et al. 2017), and microglial activation (Valdearcos, Douglass et al. 2017). Collectively, these pathological alterations impair glia-neuron metabolic interactions and lead to a failure in the energy supply chain to neurons, which may potentially result in neuronal damage leading to cognitive impairment.