Role of astroglia in diet-induced central neuroplasticity

Abstract

Obesity, characterized by increased adiposity that develops when energy intake outweighs expenditure, is rapidly becoming a serious health crisis that affects millions of people worldwide and is associated with severe comorbid disorders including hypertension, cardiovascular disease, and type II diabetes. Obesity is also associated with the dysregulation of central neurocircuits involved in the control of autonomic, metabolic, and cognitive functions. Systemic inflammation associated with diet-induced obesity (DIO) has been proposed to be responsible for the development of these comorbidities as well as the dysregulation of central neurocircuits. A growing body of evidence suggests, however, that exposure to a high-fat diet (HFD) may cause neuroinflammation and astroglial activation even before systemic inflammation develops, which may be sufficient to cause dysregulation of central neurocircuits involved in energy homeostasis before the development of obesity. The purpose of this review is to summarize the current literature exploring astroglial-dependent modulation of central circuits following exposure to HFD and DIO, including not only dysregulation of neurocircuits involved in energy homeostasis and feeding behavior, but also the dysregulation of learning, memory, mood, and reward pathways.

INTRODUCTION

Over the past two decades, the rates of obesity and associated metabolic disorders have increased dramatically, both in the United States and worldwide (Hammond and Levine 2010). Obesity and its comorbid disorders, including hypertension, type II diabetes, heart disease, stroke, and osteoarthritis, represent a serious health risk and imposes a significant strain on the economy, costing the United States an estimated $147 billion annually in healthcare and associated productivity costs (Hammond and Levine 2010; Paeratakul et al. 2002). While obesity is recognized as a multifactorial disorder with strong genetic and environmental components, it ultimately results from disordered and dysregulated energy balance, where caloric intake exceeds energy expenditure (Levin 2006, 2010a, 2010b). Prolonged exposure to a high-fat diet (HFD) increases food and caloric intake per meal, resulting in weight gain and increased adiposity in both humans and animal models alike (Daly et al. 2011; de Lartigue et al. 2011). Interestingly, prolonged HFD exposure and diet-induced obesity (DIO) are associated with not only dysregulation of autonomic and metabolic functions related to energy homeostasis (Chaar et al. 2016; Kentish et al. 2016; Little and Feinle-Bisset 2011; Little et al. 2007; McMenamin et al. 2018; Troy et al. 2016), but also the disruption of higher order functions such as learning, memory, mood, reward processes, and hippocampal activity (Cano et al. 2014; Hao et al. 2016; Spencer et al. 2017; Wu et al. 2018). One common thread between DIO and its wide range of comorbid disorders is the inflammation associated with increased adiposity (Ávalos et al. 2018; Belegri et al. 2018; Dalvi et al. 2017; Guillemot-Legris et al. 2016; Spencer et al. 2017). Whether inflammation plays a significant role in the development of obesity, however, appears to be a more challenging and unanswered question—one that is rapidly becoming an area of great interest.

DIO has a strong association with systemic inflammation, which has been assumed to lead to the development of neuroinflammation and the dysregulation of autonomic and metabolic functions (Ávalos et al. 2018; Bastard et al. 2006; Belegri et al. 2018; Purkayastha and Cai 2013; Tilg and Moschen 2006; Trayhurn 2005). Several groups have demonstrated recently, however, that neuroinflammation is detectable in central regions responsible for the regulation of food intake and energy homeostasis after only 1 day of HFD exposure, long before systemic inflammation is detected (Belegri et al. 2018; Waise et al. 2015), and well in advance of dysregulation of brainstem and hypothalamic neurocircuits (Astiz et al. 2017; Belegri et al. 2018, Buckman et al. 2015; Clyburn et al. 2018). This raises the question whether diet-associated neuroinflammation can occur independently of increased adiposity and its associated systemic inflammation, and whether neuroinflammation-induced neurocircuit dysregulation may also contribute directly toward the development of obesity and its associated comorbid disorders. The mechanisms by which HFD exposure can induce central neuroinflammation in the absence of increased adiposity and circulating proinflammatory cytokines have not been examined thoroughly, however. The primary immunoregulatory cells of the central nervous system (CNS), astrocytes and microglia, are able to modulate synaptic strength and neuronal excitability when they are phenotypically active, and usually observed during neuroinflammation, suggesting that astroglial activation following HFD exposure may be involved in the dysregulation of central neurocircuits associated with food intake and energy homeostasis following HFD and DIO exposure (Bonansco et al. 2011; Clasadonte and Prevot 2018; Fellin et al. 2004).

Understanding how altered diet composition and increased caloric intake affects neurosignaling to promote the development of obesity is critically important to elucidating novel therapeutic strategies to limit food intake and weight gain. The purpose of this review is to summarize the current literature exploring astroglial-dependent modulation of central circuits following exposure to HFD and DIO, in addition to highlighting evidence that suggests acute alterations in diet cause neuroinflammation that result not only in the dysregulation of energy homeostasis and feeding behavior, but also in the dysregulation of learning, memory, mood, and reward pathways. While, to date, only a few studies have examined the direct involvement of astroglia in the regulation of food intake during HFD exposure, there is a strong body of literature detailing the effects of HFD on astroglial modulation as well as the corresponding downstream effects of astroglia on local neurocircuits suggesting that astroglia may, at the very least, play a prominent role in diet-induced central neuroplasticity.

NEUROINFLAMMATION AND ASTROGLIAL FUNCTION

Canonically, DIO has a strong association with systemic inflammation in adipose and hepatic tissue (Bastard et al. 2006; Das 2010; Trayhurn 2005). While adipose tissue is responsible for the storage of energy in the form of lipid droplets, adipocytes are closely related to macrophages and participate in the inflammatory cascade (Grant and Dixit 2015; Tilg and Moschen 2006) and several other adipose tissue cell types, including lymphocytes and fibroblasts, serve as modulators of the immune system (Grant and Dixit 2015; Schäffler et al. 2007). Because of its secondary function as an immunoregulatory tissue, the level of adiposity has been assumed to correlate directly with serum levels of inflammatory cytokines (Bastard et al. 2006; Naznin et al. 2015). Following chronic HFD exposure and the development of DIO, levels of critical circulating immunoregulatory proteins, including the proinflammatory cytokines tumor necrosis factor-α (TNF-α) and interleukin-6 and -1 (IL-6 and IL-1, respectively), increase (Fain 2006; Wisse 2004). Chronic adipose-associated systemic inflammation is associated with CNS neuroinflammation and with the consequent dysregulation of multiple neurocircuits, including those involved in caloric intake and energy homeostasis (Ávalos et al. 2018; Belegri et al. 2018; Dalvi et al. 2017; Guillemot-Legris et al. 2016; Li et al. 2018; Purkayastha and Cai 2013; Wu et al. 2014). While circulating cytokines have been shown to enter the central parenchyma through saturable transport mechanisms or via selective uptake in discrete brain regions such as the septum, emerging evidence suggests that neuroinflammation in response to HFD per se may actually appear long before the increase in circulating cytokines and the development of systemic inflammation associated with DIO (Banks and Kastin 1991; Bauer et al. 2007; Gutierrez et al. 1993).

Immune responses within the CNS involve a specialized group of inflammatory cells, composed primarily of microglia and astrocytes (Barone and Kilgore 2006; Cartier et al. 2005). In healthy, uninflamed tissue, astroglia (i.e., astrocytes and microglia) were described originally as being a merely passive and supportive cell type responsible primarily for maintaining a healthy neuronal population (Kimelberg 2004, 2007; Wang and Bordey 2008); they have since been shown to fulfill multiple critical roles in CNS function (Chowen et al. 2013; Pérez-Alvarez and Araque 2013; Pérez-Alvarez et al. 2014; Um 2017). Astrocytes, named for their star-shaped morphology, are derived from neural progenitor cells in the neuroectoderm and are responsible for maintaining the blood-brain barrier (BBB), promoting neuronal survival, and formation and maintenance of the synapse (Kimelberg 2007; Wang and Bordey 2008). Astrocytes that come in contact with and maintain the BBB, for example, referred to as tanycytes, play an important role in controlling BBB permeability to peptides and proteins in addition to acting as important biosensors, modulating neuronal activity in response to circulating factors (Langlet 2014). In contrast, microglia originate from hematopoietic stem cells and are the innate immune cells of the CNS (Eglitis and Mezey 1997; Wang and Bordey 2008).

While astrocytes and microglia have been considered to fulfill distinct and separate primary functions, their roles are closely intertwined (Pascual et al. 2012; Rothhammer and Quintana 2015). Recent evidence suggests that each cell type has multiple subtypes and activity states, making their functional classification more complex (Kimelberg 2004; Nakajima and Kohsaka 2001). Microglial activation via LPS administration, for example, can trigger astrocyte activation through the release of ATP (Pascual et al. 2012) and, once activated, astrocytes are then able to modulate excitatory neurotransmission (Bonansco et al. 2011; Fellin et al. 2004; Jourdain et al. 2007; Pascual et al. 2012). Astrocytes also form regional specific networks through gap junction coupling (Contreras et al. 2002; Lee et al. 1994), which also allow for direct cytoplasmic coupling through the docking of hemichannels composed of membrane-located connexin 43 dodecamers (Contreras et al. 2002). This unique feature allows not only for the transmission of ionic and metabolic signals within a local network, but the amplification and coordination of astrocytic responses, including the modulation of excitatory neurotransmission (Contreras et al. 2002; Lee et al. 1994). Other studies have indicated that microglia may modulate synaptic strength both directly, through the release of ATP and glutamate, to activate purinergic and both ionotropic and metabotropic glutamate receptors, respectively (Angulo et al. 2004; Bessis et al. 2007; Roumier et al. 2004), as well as indirectly, through the release of d-serine (which acts as a coagonist of NMDA receptors) and brain-derived neurotrophic factor (BDNF) which has multiple actions to modulate both excitatory and inhibitory synaptic transmission (Bessis et al. 2007; Rose et al. 2018). BDNF, for example, which acts on the tyrosine kinase B (TrkB) receptor, can increase neuronal excitability not only by downregulating the K⁺-Cl⁻ cotransporter (KCC2), thus altering chloride homeostasis and inducing a depolarizing shift in the reversal potential for fast GABAergic transmission (Coull et al. 2005; Gomes et al. 2013), but also by phosphorylation and activation of NR1 subunit of NMDA receptors (Fig. 1) (Liu et al. 2015).

Fig. 1.

Astroglial modulation of synaptic strength. A representative diagram illustrating key mechanisms by which astrocytes and microglia may modulate synaptic strength. Astrocytes are able to modulate synaptic strength through both neurotransmitter reuptake as well as by gliotransmitter release. By physically moving closer or further away from the synapse, thereby increasing or decreasing the rate of neurotransmitter reuptake of GABA through GAT, glutamate through EAAT1/2, and glycine through GlyT1, astrocytes are able to alter the neurotransmitter concentration in the synaptic cleft. Astrocytes are also responsible for the nonvesicular release of gliotransmitters such as glutamate, ATP, TNF-α, d-serine (coagonist for the synaptic NMDA receptors), and glycine (coagonist for the extrasynaptic NMDA receptors) through the functional reversal of the GlyT1 transporter. Astrocytes can also communicate to nearby glia through gap junctions and calcium signaling to coordinate responses in distinct networks. Microglia modulate synaptic strength through the release of proinflammatory cytokines and other neuroactive substances, including glutamate, d-serine, TNF-α, BDNF, IL-6, and IL-1. Indeed, microglial release of TNF-α has been shown to phosphorylate AMPA receptors resulting in their internalization, while BDNF release activates neuronal TrkB receptors hence modulates neuronal excitability. Cytokine release from microglia can also modulate neuronal networks by promoting neurogenesis, and may be responsible for the activation of local astrocytes, leading to further adaptations in synaptic plasticity. While further investigations are required to determine the exact cascade of events that results in diet-induced astroglial modulation of synaptic strength, several lines of evidence suggest that high-fat diet exposure leads to microglial activation and acute inflammation, which then induces astrocyte activation. Once activated, both astrocytes and microglia are able to modulate their control over synaptic strength through the mechanisms described above. AMPA, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid; BDNF, brain-derived neurotrophic factor; EAAT1/2, excitatory amino acid transporters; GlyT1, glycine transporter; GAT, GABA transporter; LTD, long-term depression; NMDA, N-methyl-d-aspartate; P2X, puringergic P2X receptor; -R, receptor; TrkB, tyrosine kinase B.

While astrocytic modulation of excitatory neurotransmission can modulate the gain and efficacy of nearby neuronal synapses, and subsequently modulate and alter physiological output, microglial release of glutamate and d-serine can also induce excitotoxicity and neurodegeneration throughout the CNS, including the hippocampus and cortex (Bessis et al. 2007; Bonansco et al. 2011; Carson et al. 2006; Hao et al. 2016; Nakajima and Kohsaka 2001). Inhibition of microglial glutamate and d-serine release following ischemic stroke rescues the observed loss of cortical neuron viability, while stimulation of astroglia in the hippocampus with soluble amyloid precursor protein (sAPP) results in hemichannel-dependent release of glutamate, and subsequent neuronal excitotoxicity (Barger and Basile 2001; Kim et al. 2007; Yrjänheikki et al. 1998). While astroglial modulation of synaptic strength will be discussed in more detail below; it is important to note that both activated astrocytes and microglia are responsible for gliotransmitter release and the modulation of neuronal activity, suggesting that both cell types share some similar functions, further complicating their functional classification.

The importance of both astrocytes and microglia in central inflammation is well recognized (Hanisch 2002; Norden et al. 2016). While many central proinflammatory cytokines may be derived from infiltrating blood-borne macrophages, studies using cultured astrocytes and microglia have found that they exhibit a stimulation method-dependent and cell-type specific release of cytokines (Aloisi et al. 1992; Chung and Benveniste 1990; Lee et al. 1993); astroglial networks, therefore, may themselves also induce and support central neuroinflammation (Colonna and Butovsky 2017, Hanisch 2002; Norden et al. 2016). Stimulation of astrocytes with interleukin 1β (IL-1β), for example, induces the release of tumor necrosis factor-α (TNF-α) and IL-6 in protein kinase C (PKC)-dependent manner (Aloisi et al. 1992, Chung and Benveniste 1990; Lee et al. 1993). Stimulation of microglia with lipopolysaccharide (LPS), commonly observed during ischemia (Lee et al. 1993), in contrast, increases release of TNF-α through a mechanism dependent on the transcription and translation of p38 mitogen-activated protein (MAP) kinase. Interestingly, TNF-α has both neuroprotective and deleterious effects in regards to local neuronal health (Czeh et al. 2011), having been shown to reduce neuronal oxidative stress in the cortex following traumatic brain injury (Sullivan et al. 1999), but also being responsible for neuronal apoptosis and inhibiting neurite growth in the hippocampus (Hallenbeck 2002; Neumann et al. 2002). Microglia have been shown to be capable of releasing an array of inflammatory cytokines including, but not limited to, IL-1β, IL-3 IL-6, IL-8, gamma interferon inducible protein-10 (IP-10) and transforming growth factor (TGFβ), all of which may have unique effects on neuronal function (Hanisch 2002; Lee et al. 1993). Thus, astrocyte- and microglia-dependent release of proinflammatory cytokines is a dynamic and responsive system with many potential functions and outcomes (Fig. 1). Defining whether neuroinflammation-induced neuroplasticity is caused by activation of microglia or astrocytes, however, is complex and difficult to separate into discrete cell-type specific mechanisms.

Regardless of the cell type involved, however, it is clear that central neuroinflammation can occur independently of peripheral inflammation, including activation of astroglia, although the mechanism(s) responsible are not well understood. Astroglia are capable of sensing and responding to the excitability of nearby neurons (Clasadonte and Prevot 2018; Fellin and Carmignoto 2004; Pérez-Alvarez and Araque 2013), suggesting that altered neuronal activity may be responsible, least in part, for alterations in astroglial function and subsequent release of gliotransmitter (Guillemot-Legris et al. 2016; Gutiérrez-Martos et al. 2018).

ASTROGLIAL MODULATION OF SYNAPTIC STRENGTH

Mechanisms by Which Astroglia Alter Synaptic Efficacy

It has become clear that astroglia play critical roles in several areas of CNS function, and their role in the maintenance and modulation of synaptic strength has profound implications on our understanding of neurophysiology and neurocircuit plasticity in response to different sensory inputs, including dietary modulation. As discussed earlier, astroglia are themselves able to release neurotransmitters (gliotransmission), including glutamate and adenosine triphosphate (ATP), which can modulate the excitability of nearby synapses directly, in addition to being capable of releasing receptor coagonists such as glycine and d-serine that are required for NMDA receptor activation (Araque et al. 2014; Beltrán-Castillo et al. 2017; Bonansco et al. 2011; Choe et al. 2012; Clasadonte and Prevot 2018; Covelo and Araque 2018; Rose et al. 2018; Savtchouk and Volterra 2018; Stern and Filosa 2013; Zorec et al. 2012). Using dual whole cell patch-clamp electrophysiology and calcium imaging, work from several groups has demonstrated that selective stimulation of astrocytes causes glutamate exocytosis and the activation of presynaptic NMDA receptors; this increases presynaptic terminal excitability and subsequently increases the probability of neurotransmitter release, providing direct evidence for astroglial modulation of neuronal excitability and transmission, and illustrating one mechanism by which this modulation may occur (Fig. 1) (Angulo et al. 2004; Jourdain et al. 2007).

Astrocytes also regulate synaptic strength by virtue of their ability to uptake neurotransmitters such as glutamate and GABA from the synaptic cleft via the excitatory amino acid (EAAT1 and EAAT2), GABA (GAT), and glycine (GlyT1) transporters (Schousboe 2003; Schousboe et al. 2013). Under normal physiological conditions, for example, GlyT1, a sodium- and chloride-dependent symporter (Huang et al. 2004; Shibasaki et al. 2017), allows astrocytes to transport glycine intracellularly from the synaptic cleft (Aroeira et al. 2014). Several studies in the brainstem and spinal cord have shown that increasing intracellular sodium concentrations can functionally reverse this transporter, allowing for nonvesicular release of glycine (Aubrey et al. 2005; Raiteri et al. 2008), as also demonstrated following dopamine stimulation in the neonatal prefrontal cortex (Shibasaki et al. 2017). Astrocytes may, therefore, be a major source of glycine and allow for the activation of neuronal glycine receptors, found in abundance throughout the neonatal cortex and brainstem. Because of its additional role as a coagonist, release of glycine from astroglia may also contribute to the activation of NMDA receptors and hence be involved in the modulation of excitatory neurotransmission (Fig. 1) (Shibasaki et al. 2017).

Because of their mobility, by physically moving closer or farther away from the synaptic cleft, astrocytes are able to modulate the rate of neurotransmitter uptake and, hence, the concentration of neurotransmitter within the synapse (Langle et al. 2002; Montagnese et al. 1987; Perlmutter et al. 1985). While the effects of diet per se on astrocytic control of synaptic glutamate concentration through physical mobility remains to be elucidated, studies have demonstrated that, within the supraoptic nucleus, lactation or water deprivation causes astrocytic processes to physically retract from the synaptic cleft, increasing the synaptic concentration of neurotransmitter, allowing greater diffusion, and the activation of presynaptic and extrasynaptic receptors (Miyata et al. 1994; Montagnese et al. 1987). Astrocytes, therefore, may exert a tight and dynamic control over the temporal pattern of neurotransmitter and, subsequently, synaptic strength and efficacy in an activity-dependent and on-demand manner (Fig. 1) (Aguado et al. 2002; Dani et al. 1992).

Although astrocytes do not exhibit the same canonical electrical signaling properties as their neighboring neurons and have not traditionally been considered electrically excitable (Perea and Araque 2005; Zorec et al. 2012), astrocytic processes do express ionotropic and metabotropic receptors as well as the signaling machinery required to detect and respond to local neuromodulators, such as acetylcholine and norepinephrine (Bekar et al. 2008; Ding et al. 2013; Paukert et al. 2014; Schipke and Kettenmann 2004; Takata et al. 2011; Zonta et al. 2003). Astrocytes are also able to buffer extracellular calcium and potassium levels, through calcium-dependent uptake of potassium, altering presynaptic excitability and the probability of neurotransmitter release (Lian and Stringer 2004; Wallraff et al. 2006). Alterations in intracellular calcium levels also regulates the production of transcription factors responsible for calcium-dependent gliotransmitter release (Scemes and Giaume 2006) in addition to signaling via gap junctions to neighboring astrocytes allowing for the communication and coordination of an astrocytic network through calcium waves (Evans and Martin 2002; Scemes and Giaume 2006). Astrocytes also express voltage-activated sodium currents, which results in outwardly rectifying current-voltage relationships (Akita et al. 2011; Bevan et al. 1985; Sontheimer et al. 1996) and a subset of astrocytes have also been shown to have a resting sodium conductance (O'Connor et al. 1994; Sontheimer et al. 1996), illustrating that subsets of astrocytes may be electrically excitable, and sensitive to not only the excitability of nearby neurons, but also able to communicate this signal within the astroglia network.

Thus, astrocytes can exert tight control over synaptic strength and efficacy and are able to modulate the gain of neuronal signaling through several passive and active mechanisms such as responding to released neurotransmitters and neuromodulators, releasing cofactors and gliotransmitters, and regulating network activity via gap junctions (Fig. 1). Astrocytes, therefore, should be considered dynamic components in the control, regulation, and modulation of neurocircuits, responsive to environmental and neuronal cues, and crucial components in the coordination of network responses over large distances.

Brainstem

Diet can have profound effects on synaptic strength in many areas of the brainstem, including those in the dorsal vagal complex (DVC), which includes the dorsal motor nucleus of the vagus (DMV), the nucleus tractus solitarius (NTS), and the area postrema (AP) (Bhagat et al. 2015; Clyburn et al. 2018; Kentish et al. 2012, 2016; McMenamin et al. 2018; Travagli and Anselmi 2016; Troy et al. 2016). DVC neurocircuits are responsible for integrating and responding to peripheral sensory information from cardiovascular, respiratory, and gastrointestinal (GI) systems (Lu and Bieger 1998; Travagli and Anselmi 2016). Astroglial modulation of synaptic strength has been observed in many autonomic brainstem neurocircuits, (Dallaporta et al. 2010). In the cardiovascular field, for example, the astrocytic buffering of extracellular glutamate via EAAT2 decreases NTS neuronal baseline activity and tonically decreases atrial pressure, suggesting that astrocytes are critical for the control of cardiovascular function under baseline conditions (Matott et al. 2017). In the respiratory field, activation of the proteinase-activated receptor 1 (PAR1) on NTS astrocytes potentiates neuronal synaptic activity (Beltrán-Castillo et al. 2017), increasing glutamate signaling to the NTS-rostral ventral respiratory group, hence regulating breathing patterns (Beltrán-Castillo et al. 2017). Astrocytes have also been implicated in the regulation of respiration under stress, during exercise, and involved in the determination of exercise capacity (Sheikhbahaei et al. 2018). Although the effects of diet on astroglial modulation of synaptic strength in the central respiratory and cardiovascular neurocircuits have not been examined directly, evidence from the GI field suggests that such dietary-induced changes in astroglial function are likely to occur (Balland and Cowley 2017; Buckman et al. 2015).

Extrinsic neural control of GI function, specifically motility of the stomach and upper GI tract that arises principally from parasympathetic inputs via the efferent vagus nerve, is primarily made up of vago-vagal reflexes (Travagli and Anselmi 2016). Briefly, vagal afferent (sensory) neurons, the cell bodies of which lie in the nodose ganglion, innervate the stomach and upper GI tract and relay this sensory signal to the NTS (Browning 2003; Travagli et al. 2006; Zhang et al. 1998). The NTS integrates this sensory signal with inputs from the brainstem and hypothalamus that are involved in energy homeostasis and sends GABAergic, glutamatergic, and catecholaminergic projections to the preganglionic motoneurons of the DMV (Travagli and Anselmi 2016; Travagli et al. 1991). The DMV sends cholinergic projections to the postganglionic neurons in the myenteric plexus, which makes up two distinct pathways, the excitatory cholinergic and inhibitory nonadrenergic, noncholinergic pathway (Travagli and Anselmi 2016). Diet-induced modulation of GI vagal pathways are apparent in response to both acute and chronic exposure (Bhagat et al. 2015; Clyburn et al. 2018; de Lartigue et al. 2011; Fox and Biddinger 2012; Kentish et al. 2012, 2016; Nefti et al. 2009). DIO is associated with a decreased intrinsic excitability of vagal afferent sensory neurons and fibers (Daly et al. 2011) as well as vagal efferent motoneurons, including decreasing the responsiveness to classical satiety peptides such as cholecystokinin (CCK), glucagon-like peptide 1 (GLP-1), and leptin, as well as the glucose-dependent facilitation of serotonin (5-HT)-mediated responses (Bhagat et al. 2015; Covasa and Ritter 2000; Duca et al. 2013; Troy et al. 2016). Reduced afferent excitability decreases responsiveness to meal-induced mechanical and chemical stimulation, while decreased efferent motoneuron excitability reduces the tone of the stomach and increases gastric and fasting volume (Asakawa et al. 2003; Bhagat et al. 2015; Daly et al. 2011; de Lartigue et al. 2011; Fox and Biddinger 2012; Gallagher et al. 2007; Kentish et al. 2012). Together, these increase the volume of food required to signal satiation, ultimately increasing food intake and meal size, and contributing to the development and maintenance of obesity. Interestingly, the responsiveness of central vagal motoneurons is restored following Roux-en-Y gastric bypass surgery, suggesting that such neuroplasticity is transient and reversible (Browning et al. 2013). While such vagal neurocircuit plasticity may certainly be caused by the increase in adiposity and circulating proinflammatory cytokines associated with obesity, the rapid onset of some alterations, far in advance of weight gain or increased adiposity (Clyburn et al. 2018; Waise et al. 2015), suggests some effects are due to the diet itself, rather than obesity. One day of HFD exposure is sufficient to cause inflammation in the nodose ganglion and hypothalamus, for example, while vagal responsiveness to glucose-dependent facilitation of 5-HT signaling is reduced within 4 days (Troy et al. 2016; Waise et al. 2015). Interestingly, selective celiac branch vagotomy prevented the increase in proinflammatory cytokine expression in both the nodose ganglion and hypothalamus, suggesting that acute HFD-induced neuroinflammation may be vagally dependent and occurs long before peripheral inflammation is observed (Waise et al. 2015).

Both humans and animal models alike respond to HFD exposure with a short (~24-h) period of hyperphagia, before energy homeostasis is restored and food intake returns to an isocaloric level within 3–5 days (Buckman et al. 2015). The mechanism(s) responsible for the homeostatic regulation of caloric intake following HFD exposure, and how this mechanism is lost following continued chronic HFD exposure when hyperphagia returns leading to development of DIO, is a growing area of interest. Recent studies from our laboratory have demonstrated that restoration of caloric balance occurs over the same time period (3–5 days) during which acute HFD exposure increases glutamatergic signaling to neurons of the DMV via increased activation of synaptic NMDA receptors, subsequently increasing DMV neuronal excitability, and vagal efferent control of gastric tone and motility (Clyburn et al. 2018). The timescale of this central vagal neuroplasticity corresponds with an observable increase in neuroinflammatory markers such as IL-1β and TNF-α within the pons and midbrain (Astiz et al. 2017; Balland and Cowley 2017; Dalvi et al. 2017). While the role of astroglia in the acute HFD modulation of brainstem synaptic strength remains to be elucidated, the time course of both the neuroplasticity as well as the increased neuroinflammatory markers suggests the involvement of astroglial-dependent compensatory mechanisms that restores caloric balance temporarily (Astiz et al. 2017; Buckman et al. 2015; Douglass et al. 2017). Indeed, several studies have suggests that acute neuroinflammation is protective against a wide range of insults, including hemorrhage, whereas chronic neuroinflammation induces significant damage and disruption to neurocircuit function (Astiz et al. 2017; Burda and Sofroniew 2014). Chronic HFD exposure and DIO, for example, decrease vagal motoneuron excitability, alter neuronal morphology, and inhibit the CCK-induced modulation of synaptic transmission (Bhagat et al. 2015; McMenamin et al. 2018).

Hypothalamus

While alterations in the brainstem neurocircuits that control gastrointestinal function have an obvious effect on gastric compliance, tone, motility, emptying rates and subsequently, meal size and energy homeostasis, HFD-induced astroglial modulation of hypothalamic signaling has also been observed (Ávalos et al. 2018; Balland and Cowley 2017; Buckman et al. 2015; Chowen et al. 2013; Douglass et al. 2017; Pfuhlmann et al. 2018). Given the critical role of the hypothalamus, especially the tuberal region, in controlling appetite, modulation of these circuits may have profound effects on feeding behaviors, caloric intake, and the development of obesity (King 2005; Velloso et al. 2008). As in the brainstem, modulation of hypothalamic circuits can occur rapidly; an increase in proinflammatory cytokines is observable within 24 h of HFD exposure, suggesting such neuroinflammation occurs via a mechanism independent of an increase in systemic cytokine level (Astiz et al. 2017; Waise et al. 2015). Astrocytes have important roles in the regulation of hypothalamic neurocircuitry, both in basal conditions as well as following HFD (Ávalos et al. 2018; Kälin et al. 2015; Langle et al. 2002). Within the basal medial hypothalamus, for example, astroglial activation decreases both basal and ghrelin-induced caloric intake through the activation of agouti-related protein (AgRP)-positive neurons (Yang et al. 2015). In the ventromedial hypothalamus, in contrast, astroglia release ketone bodies in response to HFD, altering neuronal excitability through the concurrent release of ATP (Le Foll et al. 2014). Thus, the diet-induced astroglial-dependent modulation of neuronal activity within the hypothalamus may even exhibit subtle regional, and nucleus-specific, variability.

Given the important role of leptin in food intake and energy balance, it is perhaps unsurprising that its involvement in the astroglial modulation of food intake has been proposed (Cheunsuang and Morris 2005; Kim et al. 2014; Pan et al. 2008; Yang et al. 2015). As a polypeptide produced by adipose tissue and the gastric mucosa, leptin enters the CNS via a selective, saturable transport system and acts within the hypothalamus to induce satiation and reduce food intake (Ahima and Lazar 2008; Berthoud 2005; Berthoud and Morrison 2008; Burguera et al. 2000; Kim et al. 2014). Hyperleptinemia, caused by an increase in adiposity, induces leptin insensitivity and resistance within the hypothalamus and brainstem, causing dysregulation of energy balance, resulting in increased meal size, weight gain, and the development of obesity (Berthoud 2005; Berthoud and Morrison 2008). While studies have traditionally focused on the neuronal effects of leptin, it should also be noted that astroglia, and hypothalamic astroglia in particular, express leptin receptors (Balland and Cowley 2017; Cheunsuang and Morris 2005; Kim et al. 2014). Indeed, astrocyte-specific deletion of the leptin receptor alters glial morphology as well as hypothalamic proopiomelanocortin and AgRP neuronal signaling, leading to the development of obesity (Jayaram et al. 2013). Interestingly, leptin resistance also decreases astrocytic release of ketone bodies in the ventromedial hypothalamus, attenuating the regulation of caloric intake and energy homeostasis (Le Foll and Levin 2016). Furthermore, in astrocyte-specific leptin receptor knockout mice, phosphorylated signal transducer and activator of transcription 3 (pSTAT3) signaling is decreased, and is accompanied by a mild reactive gliosis (Wang et al. 2015). Following exposure to HFD, these leptin receptor knockout mice increased body fat to a greater extent than mice fed either a control or HFD, suggesting that astrocytic leptin signaling is also important in regulating the hypothalamic response to HFD (Wang et al. 2015). Under both basal conditions as well as acute HFD exposure, therefore, astrocytes not only are sensitive to leptin levels, but they tonically regulate synaptic strength within feeding neurocircuits to regulate caloric intake.

Dorsal Striatum

While rapid neuroplasticity within the brainstem and hypothalamic neurocircuits and astroglial networks responsible for GI function, energy homeostasis, and feeding behavior could be considered to have an adaptive advantage, diet-induced astroglial plasticity can also be observed in other discrete areas of the brain. Modulation of activity within the dorsal striatum, involved with the refinement of motor activity and decision making and implicated in food seeking behaviors, occurs following a high-fat/high-carbohydrate diet (Fritz et al. 2018). Using electrophysiological recordings and fast-scan cyclic voltammetry techniques, the altered glutamatergic signaling apparent in neurons of the dorsal striatum following high-fat/high-carbohydrate diet exposure appears to be dependent on the expression of the glial glutamate transporter-1 (GLT-1 or EAAT2) and the subsequent slower reuptake of glutamate, which then increases neuronal excitability (Fritz et al. 2018). Thus, in addition to acting within autonomic networks to increase food intake, HFD may alter astrocytic signaling in the dorsal striatum to also alter food seeking, decision making, and motor activity.

Nucleus Accumbens

Diet-induced plasticity within astroglia networks may have also been identified within the nucleus accumbens, which plays a key role in the central reward circuits (Blancas-Velazquez et al. 2018; Gutiérrez-Martos et al. 2018). In studies where C57BL6/J mice were allowed to choose between a standard diet and a high-fat/high-carbohydrate food (chocolate bars; free choice model), or were allowed access only to the chocolate bars (binge model), the observed structural alterations in medium spiny neurons were prevented by the microglial inhibitor, minocycline (Gutiérrez-Martos et al. 2018), as were the increased expression of neuroinflammatory markers and the altered response to induced hyperlocomotion (Gutiérrez-Martos et al. 2018). While it is unclear whether these observations were due directly to activated microglia, or indirectly through their subsequent activation of astrocytes, it is evident that diet-induced astroglial activation affects reward pathways, not only through the modulation of neuronal function and signaling, but also through modulation of neuronal morphology. The physiological outcome of such diet-induced astroglial modulation in the nucleus accumbens remains unclear, however, but, given the prominent role of the nucleus accumbens in both reward-related as well as motivated behavior, this modulation may well have important implications in the hedonic control of food intake.

Amygdala and Hippocampus

Perhaps the most convincing evidence supporting HFD-dependent astroglial modulation of neuronal activity involves the amygdala and hippocampus. HFD exposure has long been associated with the disruption of cognitive performance, especially learning and memory, an area of increasing interest in populations already at risk for hippocampal impairment, such as the elderly and those with Alzheimer’s disease (Arcego et al. 2018; Hao et al. 2016; Koga et al. 2014; Spencer et al. 2017). Activated astroglia within the hippocampus become leptin-insensitive following HFD exposure, and leptin-induced modulation of hippocampal synaptic transmission is attenuated under the same conditions (Mainardi et al. 2017). HFD can also increase glutamatergic signaling within the hippocampus through altered regulation of astroglial GLT-1 activity and the inhibition of glutamine synthesis (Cano et al. 2014; Valladolid-Acebes et al. 2012). Other studies have shown that prolonged HFD exposure and DIO increases neuroinflammatory markers and levels of phosphorylated Tau protein (a marker for neurodegeneration) and potentially amyloid beta (Aβ) plaques (the hallmark sign of Alzheimer's disease) in the hippocampus, both of which have a detrimental effect on cognitive performance (Hawkes et al. 2015; Koga et al. 2014; Ledreux et al. 2016; Martino Adami et al. 2017). Interestingly, exposure to Aβ plaques appears to inhibit astroglial control of extracellular glutamate levels, promoting dysregulation within neurocircuits involved in learning and memory (Kawano et al. 2017), suggesting that HFD-induced disruption in learning and memory in pathophysiological conditions may be exacerbated by, and at least in part due to, astroglial network disruption. Interestingly, exercise, weight loss, and astroglial inhibition have all been shown to rescue DIO-associated cognitive decline, indicating that such effects are reversible (Koga et al. 2014). In addition, neuroinflammation and dysregulation of astroglia within the hippocampus are not observed until after the dysregulation of brainstem astroglia (Astiz et al. 2017), suggesting an ascending pattern of temporal diet-induced dysfunction. Further studies would be required to determine whether brainstem astroglial network disruption is responsible for hippocampal astroglial dysfunction, or whether there is a distinct temporal patterning and susceptibility of astroglia in different regions to the effects of HFD exposure.

CONCLUSIONS

Our understanding of the role of astroglia in CNS function has advanced dramatically and rapidly in recent years. From being considered responsible only for providing structural support to nearby neurons, to being recognized for their roles in central inflammation and immune responses, maintenance of the BBB, their responses to circulating factors, and role in the finely tuned regulation of synaptic strength to maintain homeostasis or control of neurocircuit function, it is clear that astroglia play critical roles in the physiology and pathophysiology of central neurocircuits. The effect that diet, particularly HFD, has on astroglial control of synaptic strength appears to be rapid, complex, and region specific. There is evidence for activation of astroglia and inflammatory processes in the hypothalamus and hippocampus within 24 h of HFD exposure, suggesting these may be the first, and most sensitive, responses (Table 1). While the acute modulation of some neurocircuits appears to be compensatory, other studies have shown adverse outcomes in response to chronic HFD exposure and DIO-induced astroglial activation including areas involved in higher functions such as learning and memory. Furthermore, studies in both the brainstem and the hippocampus suggest that diet-induced neuroplasticity may be reversed by reducing neuroinflammation or microglia activation (Table 1). Additional studies will be needed to define the roles of acute versus chronic neuroinflammation in the context of the food intake regulation and energy homeostasis, and whether neuroinflammation can spread from the early affected autonomic areas to higher areas affected by diet-induced inflammation, such as the amygdala, hippocampus, and dorsal striatum. While there is some evidence for the spread of inflammation through astroglial networks mediated via the gap junction-dependent spread of calcium currents, the role and mechanism of these effects in the context of diet-induced neuroinflammation still remains to be elucidated. It is also possible that altered neurocircuit activity per se is responsible for the modulation of astroglial activity in areas with reciprocal neuronal connections. While further investigation is needed to determine the temporal pattern of the event cascade that follows HFD exposure and results in the modulation of synaptic strength, it appears that microglial activation and central inflammation occurs first (within ~24 h), certainly long before the development of systemic inflammation. Such microglial responses may then induce astrocyte activation and the subsequent modulation of synaptic strength within neurocircuits involved feeding behaviors and gastric function. It is tempting to speculate, therefore, that astroglial modulation of central neurocircuits plays a far more significant role in the development of DIO, in contrast to the canonical model which considers that central inflammation and neurocircuit dysregulation occurs in response to obesity itself.

Table 1.

Evidence for diet-induced modulation of regional synaptic strength

Region	Diet-Induced Modulation of Neuronal Function	Diet-Induced Modulation of Astroglial Function	Astroglial Modulation of Synaptic Strength	Diet-Induced Astroglial Synaptic Strength
Brainstem Vagal neurocircuits	Bhagat et al. 2015; Browning et al. 2013; Clyburn et al. 2018; McMenamin et al. 2017; Nefti et al. 2009; Troy and Browning 2016; Troy et al. 2016	Buckman et al. 2015b; Waise et al. 2015	Beltran-Castillo et al. 2017; Dallaporta et al. 2010; Matott et al. 2017; Sheikhbahaei et al. 2018	Not Available
Hypothalamus Medial basal and ventromedial	Avalos et al. 2018; Dalvi et al. 2017	Avalos et al. 2018; Balland and Cowley 2017; Belegri et al. 2018; Buckman et al. 2015b; Buckman et al. 2013; Dalvi et al. 2017; Douglass et al. 2017; Le Foll and Levin 2016; Waise et al. 2015	Dalvi et al. 2017; Kim et al. 2014; Le Foll et al. 2014; Yang et al. 2015	Dalvi et al. 2017; Douglass et al. 2017; Le Foll et al. 2014
Dorsal striatum	Fritz et al. 2018	Fritz et al. 2018	Fritz et al. 2018	Fritz et al. 2018
Nucleus accumbens	Blancas-Velazquez et al. 2018; Gutierrez-Martos et al. 2018	Blancas-Velazquez et al. 2018; Gutierrez-Martos et al. 2018	Blancas-Velazquez et al. 2018; Gutierrez-Martos et al. 2018	Blancas-Velazquez et al. 2018; Gutierrez-Martos et al. 2018
Amygdala and hippocampus	Arcego et al. 2018; Ledreux et al. 2016; Mainardi et al. 2017; Martino Adami et al. 2017; Valladolid-Acebes et al. 2011	Cano et al. 2014; Hao et al. 2016; Spencer et al. 2017a	Angulo et al. 2004; Hao et al. 2016; Rose et al. 2017	Hao et al. 2016; Spencer et al. 2017a

A brief overview of the evidence supporting diet-induced astroglial modulation of synaptic strength. Evidence for diet-induced modulation of neuronal activity, astroglial activity, astroglial modulation of synaptic strength, and diet-induced astroglial modulation of synaptic strength have been categorized by central nervous system (CNS) region. Not Available, CNS regions that do not have any supporting evidence available at this time.

Nevertheless, it is clear that astroglia exhibit rapid and complex responses to both acute and chronic alterations in diet and exert tight control over neuronal functions. Astroglia should be considered key players in the regulation of food intake and energy homeostasis, therefore making them an exciting area and target of interest in the study of diet-induced obesity.

GRANTS

This work was supported by NIH grants DK111667 (K. N. Browning) and DK118833 (C. Clyburn).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

C.C. and K.N.B. conceived and designed research; C.C. prepared figures; C.C. and K.N.B. drafted manuscript; C.C. and K.N.B. edited and revised manuscript; C.C. and K.N.B. approved final version of manuscript.