Astrocyte Mitochondria in White-Matter Injury

Abstract

This review summarizes the diverse structure and function of astrocytes to describe the bioenergetic versatility required of astrocytes that are situated at different locations. The intercellular domain of astrocyte mitochondria defines their roles in supporting and regulating astrocyte-neuron coupling and survival against ischemia. The heterogeneity of astrocyte mitochondria, and how subpopulations of astrocyte mitochondria adapt to interact with other glia and regulate axon function, require further investigation. It has become clear that mitochondrial permeability transition pores play a key role in a wide variety of human diseases, whose common pathology may be based on mitochondrial dysfunction triggered by Ca²⁺ and potentiated by oxidative stress. Reactive oxygen species cause axonal degeneration and a reduction in axonal transport, leading to axonal dystrophies and neurodegeneration including Alzheimer’s disease, amyotrophic lateral sclerosis, Parkinson’s disease, and Huntington’s disease. Developing new tools to allow better investigation of mitochondrial structure and function in astrocytes, and techniques to specifically target astrocyte mitochondria, can help to unravel the role of mitochondrial health and dysfunction in a more inclusive context outside of neuronal cells. Overall, this review will assess the value of astrocyte mitochondria as a therapeutic target to mitigate acute and chronic injury in the CNS.

Introduction

Astrocytes are the most widely distributed glial cells in the Central Nervous System (CNS) and they are located throughout grey and white matter, underneath dura matter, and around cerebral vessels (Fig. 1). Subsequently, astrocytes specialize on numerous key functions based upon their locations such as maintaining brain homeostasis, regulating the extracellular environment, forming and maintaining the blood-brain barrier (BBB) and neurovascular unit (NVU) [1], adjusting cerebral blood flow [2–4], and regulating pH [5]. Because astrocytes balance glutamate uptake and release, they monitor neuronal activity and synaptic function and tri-partite synapse to facilitate learning and memory [6]. Astrocytes can store glucose as glycogen, and maintain glycogen storage to convert it to lactate when there is low glucose or increased activity [7, 8] and shuttle lactate to support neuronal and glial metabolism [9–11]. Controlling ATP release [12] and Ca²⁺ networking, astrocytes adjust sleep-wake cycle [13], and facilitating transfer and exchange of soluble substrates between cerebrospinal fluid and interstitial fluid [14–16]. Astrocytes also actively modulate axon conduction [17] as well as the formation and pruning of synapses [18, 19]. This vast functional repertoire is even more augmented in the human brain due to the significant complexity of human astrocytes compared to rodents [20]. In fact, this increased intricacy of human astrocytes, together with oligodendrocytes and expansion of white-matter volume [21], are proposed to be among the prominent reasons for higher cognitive function in humans compared to rodents.

Fig. 1.

Typical astrocytes in white matter (A) and grey matter (B) show different morphological characteristics. (A) In white matter, GFAP (+) fibrous astrocytes (red) are present with numerous processes elongated parallel along the axons. Collapsed confocal images (z-stacks of 30 each at 1 μm thickness) of corpus callosum obtained from a coronal mouse brain section. (B) In grey matter, protoplasmic astrocytes (magenta) with their intricate branches extending in all directions create the well-known “star-like’” shape. Astrocytic end-feet (yellow asterisk) wrap around the blood vessels to form neurovascular units with endothelial cells and pericytes (not shown). Collapsed confocal images (z-stacks of 30 each at 1 μm thickness) of somatosensory cortex obtained from a coronal mouse brain section. All cellular nuclei are shown in cyan. Scale bars = 20 μm

Because astrocytes integrate into many units serving a specialized function in the brain, it is expected that astrocytes structurally adapt to their location and function [22]. In the CNS, astrocytes are derived mainly from radial glial cells and some from progenitor cells in the spinal cord [23–26], while there is continuous generation of astrocytes in the adult brain in subventricular cells [25]. It is possible that astrocyte origin may also contribute to the morphological heterogeneity and the anatomical destination of the astrocyte, which eventually determines its function. In agreement with this concept, transplantation of astrocytes derived from human immature glial progenitor cells into rodent brains resulted in the formation of astrocytes with the complex morphological complexity of human brain, suggesting that the size and structural architecture of astrocytes are intrinsic to their cell origin [6]. Interestingly, these cells assumed characteristics of their location and function, demonstrating the conducive nature of astrocytes to adaptation [6].

Astrocytes show remarkable differences between brain regions [22]. For example, astrocytes exhibit distinct differences between gray and white matter [22]. Based upon morphological characteristics, astrocytes are named as protoplasmic astrocytes in gray matter and fibrous astrocytes in white matter [1] (Fig. 1). Fibrous astrocytes are characterized by their smaller nuclei, and they extend their branches along the axons in parallel, giving them an elongated morphology [1]. Fibrous astrocytes contain larger amounts of filaments than protoplasmic astrocytes [27] to structurally support their elongated and extended branches [28–30]; therefore, Glial Acidic Fibrillary Protein (GFAP) is more prominent in these cells. Protoplasmic astrocytes are larger in size with fine elaborate branches distributed around the cell body, achieving their characteristic “star-like” shape. Interestingly, this location-dependent specificity of astrocytes is equally well-preserved in the human brain, except that they are bigger and protoplasmic astrocytes have more complex architectural detail [31]. Consequently, the participation of astrocytes in the tripartite synapse is enhanced in the human brain. In agreement with diverse morphology reflecting diverse functions, protoplasmic and fibrous astrocytes have different protein expression profiles. For instance, the cluster of differentiation 44 (CD44; [32] filamentous proteins such as vimentin and GFAP [33] are abundantly expressed by fibrous astrocytes in white matter. The majority of gray-matter astrocytes do not express GFAP [34, 35] unless there is an injury. Expression levels of GFAP are important in that they enable fast repetitive vs. longer but high-fidelity signal conduction in gray matter and white matter, respectively. CD44, on the other hand, is a hyaluronan receptor, suggesting an important interaction between white-matter astrocytes and the extracellular matrix.

Astrocytes also regulate glutamate homeostasis. Gray-matter astrocytes express five major glutamate transporters, while white-matter astrocytes express only GLT-1 and GLAST [36–38]. Despite the fact that expression levels of glutamate transporters are higher in white-matter astrocytes [37], the activity of glutamate transporters is seemingly higher in gray matter due to the higher numbers of synapses [39]. Given the abundant synapse numbers in the cortex, for instance, glutamate transport activity is highest in the corpus callosum [39]. Together with increased capacity for glutamate-to-glutamine cycling in white-matter astrocytes, the need for more effective glutamate clearance becomes apparent to keep glutamate levels at approximately half of gray-matter levels [39]. Excitotoxicity to oligodendrocytes via activation of AMPA and Kainate receptors [40–48], but not NMDA receptors [49], highlights the importance of glutamate clearance by astrocytes in white matter to preserve oligodendrocyte-axon interactions and to sustain conduction.

Activation of a variety of receptors on the astrocyte cell membrane triggers Ca²⁺ release from internal stores, which can spread to nearby astrocytes and initiate a Ca²⁺ wave [50] across astrocytes that propagates via gap junctions through an intricate network [51–53] to deliver a fast-long-distance signal. Astrocytes have non-overlapping domains [35] but act in unison to recruit nearby astrocytes due to their gap junctions. Interestingly, this recruitment reaches a diameter of ~400 μm in gray matter, encompassing ~100 astrocytes. This network is shown to be most elaborate in the optic nerve due to the high coupling of astrocytes [54], although the functional correlation remains unknown.

In addition, Na⁺ signaling plays an important role in maintaining astrocytic homeostasis. Of note, the concentration of cytosolic Na⁺ in astrocytes is typically higher than neurons [1, 55] and the influx of Na⁺ in astrocytes propagates from processes to soma and into adjacent cells through gap junctions [55–57]. Na⁺ can enter astrocytes through either cationic channels (e.g P2X and NMDA receptors, TRP channels, and specific Na⁺ called Na_x channels) or Na⁺-dependent transporters (e.g. excitatory amino acid transporters types 1 and 2, GABA transporter type 1 and 3, glycine transporters type 1, noradrenaline and dopamine transporters, and Na⁺-coupled neutral amino acid transporters) [1, 58, 59]. The majority of plasmalemmal transporters not only act as sensors but also modifiers of cytosolic Na⁺ [1]. On the other hand, Na⁺/K⁺ ATPase (NKA) is mainly responsible for the release of Na⁺ from astrocytes [60]. Astrocytic NKA has a lower affinity to K⁺ than in neurons because it contains α2 subunit instead of α1 and α3 subunits in neurons [61, 62]. Therefore, astrocytic NKA is important for sensing and maintaining the K⁺ balance. Also, by buffering K⁺ during neuronal activity, NKA plays an important role in the production of lactate production in astrocytes [63–65]. Additionally, astrocytes express all three subtypes of Na⁺/Ca²⁺ exchanger (NCX) which is another important player in regulating Na⁺ [60, 66, 67]. Astrocytic NCX is sensitive to changes in the cytosolic concentration of Na⁺ and Ca²⁺ which suitable for their role in preserving the astrocyte ionic homeostasis [68, 69]. In particular, astrocytes mitochondria express a unique version of the exchanger called NCLX in which it can exchange Li⁺ instead of Na⁺ and contribute significantly to maintaining mitochondrial function in astrocytes [70–72].

Given the diverse physiological functions of astrocytes and their vast connections to other astrocytes as well as other types of cells, when astrocytes become reactive after an insult and how their interactions with other brain cells change states the importance of these cells in health and disease. Indeed, their gene expression and transcriptional changes show spatial location-, injury type-, disease-, sex-, and age-specific changes [73], collectively known as “astrocyte reactivity” (Fig. 2). Astrocyte reactivity is the sum of a spectrum of changes in molecular expression, function, hypertrophy, and proliferation in response to a CNS injury. Reactive astrocytes are morphologically distinct from resting astrocytes, characterized by larger longer and ramified branches [74–77]. The changes in astrocyte function can be loss-of-function or gain-of-function, which may be beneficial or detrimental for brain tissue, although these changes are not an all-or-none response [78, 79]. Increased reactive oxygen species (ROS) and chemokine production [80–83], together with impaired Ca²⁺ [84], glutamate [85–87], and synaptic homeostasis [88–90], lead to a variety of clinical consequences, ranging from neurodegenerative diseases, epilepsy, stroke, and cognitive decline [91]. Beneficial astrocytes release a wealth of factors to adjust synapses developmentally, repairing and rewiring them after injury, assist microglia to clear cellular debris, regulate glycogen stores to sustain memory formation, and assist neuronal survival during aglycemia [92–95]. A more controversial role of astrocytes emerges from their role in scar formation, where they limit the injury borders and shield the penumbra region [96–98] as protective cells, while they may become deleterious by forming a barrier and preventing axonal regrowth across the injury area as well [99–102].

Fig. 2.

Reactive astrocytes (green) have distinct morphology compared to resting astrocytes after an ischemic episode in the brain. (A) After a middle cerebral artery occlusion (MCAO, 60 min), neuronal death causes abundant loss of MAP2 labeling (red) in the ischemic core area. (B) The penumbra region is demarcated (white dashed line) with MAP2 (+) neurons and GFAP (+) reactive astrocytes (green) that brightly express green fluorescence. Note that astrocytes are more resilient to ischemic insult than neurons, as there are surviving astrocytes (red asterisk) in the core area. (C) Cell nuclei labeled with DAPI show numerous small bright pyknotic nuclei (yellow arrowhead) in the core area, while the penumbra region is recognized by having healthy nuclei (white asterisk). (D) Merged image shows reactive astrocytes in the gray matter penumbra region and in white matter underlying the core area. A capillary located in subcortical white matter is intensely wrapped by astrocyte end-feet (red arrowhead). Collapsed confocal images (z-stack of 30 images taken at 1 μm thickness) were obtained from a coronal mouse brain slice. Scale bars = 50 μm

Because the response of astrocytes to injury intimately regulates brain function, shapes the extent of injury, and promotes or hinders repair, the energy to sustain astrocytes is crucial for their performance. Astrocytes rely heavily on glycolysis to derive energy; however, they consume ~20% of the brain’s oxygen during oxidative phosphorylation to produce adenosine triphosphate (ATP) [103] within astrocytic mitochondria. Although mitochondria in neurons have been extensively studied, the specific roles of astrocyte mitochondria in astroglial function and response to injury are just starting to be investigated. Part of this delay in appreciating the role of astrocyte mitochondria is attributed to a misperception that astrocyte processes are too small to harbor mitochondria. Astrocytes possess almost as many mitochondria as neurons [104] and recent studies have provided accumulating in vivo and in vitro evidence that mitochondria are even found in distal astrocyte processes, triggering further interest in astrocyte mitochondrial function [105–111]. Another reason for this increased interest is based on preliminary studies that have shown astrocyte mitochondria may play unique roles in response to ischemia. Astrocytes have been consistently shown to be resilient to ischemia and the distribution and specific tasks of astrocyte mitochondria may underlie this adaptability to an environment with no oxygen or glucose. This review will first introduce the diverse structure and function of astrocytes to describe the bioenergetic versatility required of astrocytes that are situated at different locations. Secondly, we will describe the intercellular domain of astrocyte mitochondria to define their roles in supporting and regulating astrocyte-neuron and astrocyte-cerebral vasculature interactions and survival against ischemia. Finally, we will review literature documenting the heterogeneity of astrocyte mitochondria and how subpopulations of astrocyte mitochondria may adapt to interact with other glia and regulate axon function in white matter. Overall, this review will assess the value of astrocyte mitochondria as a therapeutic target to mitigate acute and chronic injury in the CNS.

Astrocyte Mitochondrial Dynamics

The extent of diverse structural and functional aspects of astrocytes is expected to affect the location, size, and number of mitochondria in astrocytes. Therefore, mitochondria are strategically dispersed within astrocytes to sense energy consumption and Ca²⁺ signaling. For instance, increased synaptic activity guides mitochondria to the terminal finer branch areas [112] and immobilizes them to shape Ca²⁺ waves and regulate Ca²⁺ fluctuations [105, 113] to regulate glio-transmission [72, 110]. Evidence suggest that astrocytic mitochondria become immobilized near glutamate transporters and synapses in response to glutamate uptake [108] and primarily due to increased intracellular Ca²⁺ via reversal of the Na⁺/Ca²⁺ exchanger [114–116]. Docking of mitochondria near glutamate transporters is proposed to facilitate glutamate metabolism and ATP generation to meet increased energetics while buffering glutamate-uptake-mediated ionic changes [117]. Expectedly, a mismatch of mitochondrial distribution within astrocytes interrupts neuron-astrocyte synchronization and metabolism, threatening neuronal vitality [118].

It is now well-established that astrocyte numbers stay stable after ischemia; therefore, a mismatch in distribution and dynamics of astrocyte mitochondria at the astrocyte-neuron junction seems to cause the irreversible injury. The prevailing idea has been that the collapse of mitochondrial membrane potential leads to astrocyte death [119]; however, astrocytes demonstrate resilience to ischemia despite profound mitochondrial membrane depolarization. For instance, astrocytes maintained their mitochondrial membrane potential for 2 h after application of fluorocitrate (FC), a partially competitive mitochondrial inhibitor [120, 121]. Even after prolonged applications of FC, which eventually depleted mitochondrial potential, there was minimal astrocyte injury or death [120]. Similarly, in vitro experiments using oxygen-glucose deprivation (OGD) depolarized astrocyte mitochondrial membrane potential without subsequent cell death [122]. In vivo models using middle cerebral artery occlusion (MCAO) further confirmed that astrocyte energy metabolism is impaired after ischemia, but did not result in astrocyte death or correlate with infarct area [123] (Baltan unpublished data Fig. 2). On the other hand, neurons display massive cell death when exposed to OGD (60 min) [124, 125] and when neurons are cocultured with astrocytes treated with FC, an enhanced bi-directional injury is triggered due to astrocyte mitochondrial dysfunction, the subsequent reversal of glutamate transporters, and the resultant excitotoxicity [126, 127]. Similar widespread neuronal death is observed when the astrocyte electron transport chain is specifically targeted [128]. This contrasts with conventional observations that when astrocytes (untreated, control conditions) are co-cultured with neurons, neurons become resilient because astrocyte mitochondria shift from aerobic metabolism to glycolysis to initiate the astrocyte-neuron lactate shuttle and deliver lactate to neurons to attenuate neuronal loss [49, 93, 122, 129]. This support system is limited by astrocyte glycogen storage content and can be depleted if glucose is not supplied in a timely fashion [9, 10, 93, 129]. Together, these observations indicate that first, that the metabolic coupling of the tripartite synapse is heavily dependent upon the performance of astrocyte mitochondria. Second, they suggest that astrocyte resilience to ischemia is in part supported by energy supply derived by mitochondria. Note that some in vitro findings are difficult to extrapolate to in vivo conditions because astrocytes and their mitochondria in culture may show disparate functions due to the lack of a multi-layered network without tripartite synapses and/or gene alterations and receptor expression.

Recent studies have demonstrated that mitochondria are not confined to astrocyte cell bodies but are also present in the finest branches and end-feet, which were traditionally considered to be too small caliber to accommodate these organelles (Fig. 3). Mitochondria have the capacity to form a complex interconnected network of aggregated assemblies to dissipate into a single individual structure [105, 109, 130, 131]. Using a variety of mito-fluorescent mice (CFP, GFP) with an inducible reporter, a heterogeneous population of interconnected mitochondrial meshwork has been shown to occupy a substantial portion of specialized end-feet structures (Figs. 1, 3). While a dense meshwork of elongated mitochondria is typical for cell bodies, thinner and shorter mitochondria ranging from 0.2 to 0.6 μm in length populate the distal branches and end-feet [106–108, 110, 132]. Because mitochondria are highly dynamic organelles with an ability to rapidly change their dynamics in response to metabolic demands, they go in and out of their meshwork and change location between the cell body and branches within astrocytes. Indeed, the relative rates of fission and fusion dictate the shape, size, and distribution of mitochondria. Mitochondrial shaping proteins, such as Mitofusion-1, (MFN-1), mitofusion-2 (MFN-2), and optic atrophy-1 (OPA-1) merge outer and inner membranes of mitochondria during fusion, while dynamin reactive protein-1 (Drp-1) and fission protein-1 (Fis1) mediate fission. Fusion and fission enable the exchange of mitochondrial components such as mitochondrial DNA (mtDNA), lipids, and proteins. Fission architecturally allows entry to constrained sites of high activity such as distal fine branches of astrocytes. In fact, smaller mitochondria with higher surface-to-volume ratios have been shown to be more efficient and to generate more ATP as a response to increased activity [133]. Consistent with this concept, neuronal activity regulates mitochondrial fission to enhance the formation of smaller mitochondria, which can be directed to neuronal spines and filopodia. Subsequently, depletion of Drp-1 reduces small mitochondrial distribution in synaptic terminals, highlighting the importance of the balance between fission and fusion for mitochondria to assume the proper structure for the location and function. Naturally, when this precise balance is perturbed, lack of mitochondrial presence at sites of need may underlie pathological outcomes. Furthermore, increased Drp-1 activity leading to extensive fission leads to mitochondrial membrane depolarization, cytochrome c release, and increased free radical production causing mitochondrial dysfunction.

Fig. 3.

Networks of mitochondria (red) are present in GFAP (+) astrocyte (green) cell bodies, branches, and their end-feet accommodating the energy demand in both grey matter (A) and white matter (B). (a–h) Enlarged orthogonal-view images showing the co-localization (white arrow) of mitochondria and GFAP (+) cell bodies (c, d, e, g, cyan arrowhead) and end-feet (a, b, f, h, magenta arrowhead). Astrocyte end-feet with mitochondria wrap around blood vessels (cyan arrows). Note that in white matter (B), mitochondria can also be seen in other cell types such as oligodendrocytes with Sytox (+) nuclei (white) characterized by a “pearls on a string” appearance along the axon (yellow stars). Collapsed confocal images of the somatosensory cortex (A) and corpus callosum (B) were obtained from a transgenic mito-fluorescent mouse (RiKEN, R26R-mito-eGFP). Scale bars = 10 μm

Mitochondrial trafficking is another aspect of mitochondrial dynamics that is mainly supported by Miro1/2 and TRAKs that reversibly attach mitochondria to kinesin and dynein to facilitate anterograde and retrograde motility, respectively. Similar to what has been reported in neurons, about 15–30% of astrocyte mitochondria are motile, while the remainder are stagnant. Blocking neuronal activity with tetrodotoxin increases mitochondrial activity, while electrical or glutamate application arrests mitochondrial movement in astrocyte processes enriched in glutamate transporters and branches that are part of the tri-partite synapse. Hence, astrocytes sense and respond to neuronal activity by modifying their mitochondrial dynamics. It is therefore unsurprising that astrocyte mitochondrial dynamics undergo major modifications and play crucial roles in various pathological processes (see below).

Astrocyte Mitochondrial Response to Injury

Stimuli that initiate astrocyte reactivity such as aging, injury, and diseases similarly affect astrocyte mitochondria, leading to dysfunction. The hallmarks of this dysfunction consist of loss of Ca²⁺ regulation [134, 135], excessive production of ROS, initiation of cell death cascades, and opening of mitochondrial permeability transition pores (mtPTP; [136]). Elevated cytosolic Ca²⁺ increases and Ca²⁺ transients lead to more frequent Ca²⁺ oscillations, and excess Ca²⁺ accumulation in mitochondria depletes mitochondrial membrane potential, leading to impaired ATP production and opening of mtPTP. Opening of mtPTP results in cytochrome c release as well as other cytokines and ROS [137]. Considering the astrocytic network and the extent of communication among astrocytes, mitochondrial dysfunction, recruiting more astrocytes, and expanding the area of their network can have far-reaching effects. As a result, for instance, β-amyloid-induced oxidative stress in astrocytes causes extensive neuronal damage away from amyloid deposition areas [138, 139]. Neuronal injury can be further potentiated by the release of cytokines, ROS, and inflammatory factors [140–142]. Experiments adding FC to astrocyte cultures, which dissipates astrocyte mitochondrial membrane potentials, reduce glutamate uptake leading to increased neurotoxicity [120]. Likewise, targeting the astrocyte electron transport chain caused diffuse neuronal death [128]. These pieces of evidence imply that astrocytic mitochondria are crucial to support neurons and their function. Therefore, mitochondrial dysfunction impedes the protective roles of astrocytes, suggesting that targeting astrocyte mitochondria may provide therapeutic approaches to surrounding neurons.

Mitochondria at the astrocyte end-feet are associated with cerebrovascular structures and demonstrate high metabolic activity and dynamic Ca²⁺ signaling [143]. However, how astrocyte mitochondrial dynamics contribute to NVU, BBB, and/or cerebrovascular pathologies such as vascular dementia remain underexplored.

Mitochondrial Trafficking Between Astrocytes and Neurons

Astrocyte mitochondria can also be central to the protective and beneficial roles of astrocytes. Responding to the energy demands of astrocytes, mitochondria may regulate the release of growth factors, support synaptic function, and/or form a glial scar. Consistent with this, following a brain injury astrocyte cannot initiate a protective proliferative response if their mitochondria are dysfunctional [128]. Indeed, astrocytes dispose of their dysfunctional mitochondria via mitophagy [144], presumably to minimize the deleterious effects. Interestingly, astrocytes can take up or traffic mitochondria to and from other cells. For instance, mitochondria from retinal ganglion cells have been shown to be taken up by astrocytes, while mitochondria from astrocytes can be transferred to neurons, suggesting bidirectional trafficking of mitochondria between astrocytes and neurons. Current evidence suggests that the purpose of this activity is to deliver damaged mitochondria from neurons to astrocytes to undergo mitophagy, while healthy mitochondria moving from astrocytes to neurons support neurons in distress. Mitochondria are released in vesicles in a Ca²⁺-dependent CD38-cADPR signaling pathway [145]. Subsequently, upregulation of CD38 significantly augments the release of mitochondria-containing vesicles under in vitro and in vivo conditions [145]. Mitochondria originating from astrocytes fuse with the neuronal mitochondria located in the penumbra region and enhance the survival of neurons. Expectedly, downregulation of CD38 in an ischemia model negatively impacted the outcome measures, supporting the observation that astrocyte-mediated mitochondrial release is mediated by CD38 signaling in the brain.

There is significant interest in whether a similar survival mechanism, or lack thereof, contributes to neurodegenerative diseases. Using primary neural cells and human pluripotent stem cell-derived neural cells (hPSCs), a dynamic transfer of mitochondria from astrocytes and from neural cells into astrocytes was shown via CD38/cADPR signaling and involving Miro1 and Miro2. Introducing Alexander disease (AxD) -associated hot spot mutations into the GFAP gene of hPSCs impaired mitochondrial transfer between neural cells and astrocytes and revealed that AxD-associated mutations in the GFAP gene disrupted astrocytic mitochondria transfer, providing a potential pathogenic mechanism in AxD [146]. Interestingly, Miro1, as well as Miro2, played a role in mitochondrial transfer, which seems plausible considering the involvement of the two GTPases in intracellular mitochondrial transport and trafficking [147, 148]. The potential of Miro1 to regulate mitochondrial transfer from mesenchymal stem cells into airway epithelial cells and cardiomyocytes has been previously shown [149, 150]. Interestingly, both Miro1 and Miro2 have two EF-hand Ca²⁺-binding domains [147, 151]; whether they cross-talk with CD38/cADPR signaling remains to be explored. Determining whether mitochondrial transfer between astrocytes and from neuronal cells into astrocytes also occurs in vivo, and revealing more detailed cellular functions of the transfer, require further investigation.

Astrocyte Mitochondria as a Therapeutic Target for Neuronal Diseases

Therapeutic approaches targeting mitochondria for various brain injuries and neurodegenerative diseases using antioxidants, mtPTP inhibitors, uncouplers, and alternative fuels have been a long-standing interest. Conventionally, the majority of these investigations have been performed with the intent of rescuing neuronal mitochondria. However, astrocytic mitochondrial dysfunction may have more widespread effects based on the role, function, and location of astrocytes in various brain structures, causing extensive Ca²⁺ dysregulation, inflammatory responses, and glutamate dysregulation due to energy deprivation. Treating mitochondria in astrocytes can directly benefit neuronal survival because of neuron-astrocyte interdependence. To develop methods in a cell- and organelle-specific manner is a challenge. Several approaches such as recombinant viral vectors, nanoparticles, or specialized peptides [88, 152–156] have been developed. Using these approaches, evidence to conserve neurons via targeting astrocyte mitochondria have been collected in several disease models such as acute cortical lesions [157] β-amyloid-expressing Alzheimer’s disease [87, 88, 158, 159], chronic pain [160], and spinal cord injury [161] and CNS injury [89, 162]. Based upon these encouraging results, further studies consisting of a cocktail of antioxidants, mtPTP modulators, and alternative energy substrates together with CD38 activators can be tested to establish the therapeutic value of conserving astrocyte mitochondria in various brain injury models.

Isolation of functional and healthy mitochondria from endogenous or exogenous sources to transplant to the site of injury has become another topic of interest. The original inspiring data come from the cardiac field, in that pediatric patients with congenital myocardial disease upon transplantation of pectoral muscle mitochondria have shown instant improvement in clinical trials [163]. Similar studies investigating CNS-related diseases have yet to be conducted. However, encouraging observations from several animal studies [145, 164–167] provide compelling rationale, such the study showing that injecting isolated mitochondria into gliomas of living mice triggers a metabolic switch from glycolysis to aerobic respiration, which correlated with reduced tumor growth [168]. In a mouse model of stroke, transplanted mitochondria were taken up by neuronal cells, and astrocytes in the penumbra region were shown to deliver mitochondria to neurons. As a result, neuronal death was reduced, which was associated with improved motor and neurological function [145]. Mitochondrial delivery supplemented with PEP1, a peptide carrier to the medial forebrain bundle of a Parkinson’s rat model, spared neurons in the substantia nigra and improved locomotor function [169]. In contrast, delivery of viable mitochondria to spinal cord injured-animals maintained energetics but failed to spare tissue or improve functional recovery [166]. These results provide proof of principle that mitochondria are organelles that can be transferred from one organ to another and assume the role of performance required by the recipient tissue. Moreover, it can be exchanged and transferred between different cells to support function and survival. What the signals are that differentiate between the donor and recipient are currently not known.

In summary, there are several important reasons why it is important to investigate astrocyte mitochondria as a therapeutic target. First, in the face of an acute injury such as a stroke, neurons are the first to die; hence the window of opportunity to protect neuronal mitochondria is very narrow from the clinical perspective. Second, astrocytes survive acute injury; hence, maintaining their beneficial nature provides sustained protection in addition to reduced injury onset. Third, because astrocytes integrate into many functional units and interact with most cellular and anatomical structures, the protective measures will act on a multitude of elements in addition to neurons. Finally, different protective measures specific to the course of disease pathology can be implemented as the need for prevention and protection progresses to the regeneration and recovery phases.

Role of Astrocyte Mitochondria in Glial Cell Interactions and White-Matter Function

The role that dysfunctional mitochondria have in glial cell function, and its implications for neuronal homeostasis and white-matter function, have been largely understudied. One underlying reason is the misperception that because the white matter and glial cells are more resilient to injury compared to neurons, they do not die or sustain an injury. Indeed, oligodendrocytes, astrocytes, and microglia do not degenerate upon impairment of mitochondrial function, as they rely primarily on glycolysis to produce energy and have a higher antioxidant capacity than neurons. However, oligodendrocytes, axons, and myelin sustain a long-lasting injury due to changes in Ca²⁺ signaling, inflammation, and oxidative stress, resulting in impaired white-matter function. Moreover, recent evidence highlights the role of mitochondrial metabolism and signaling in glial cell function and nearby neuronal support. Table 1 provides an overview of key findings for the role of astrocyte mitochondria in physiological, pathophysiological state, or aging.

Table 1.

Overview of key findings for astrocyte mitochondrial role and structure in the white matter function and injury

Authors, year	Age/Exp. model	Injury Model	Key Points
Reyes & Parpura, 2008	Astrocytes culture	Physiology	Mitochondria modulate cytoplasmic Ca2+ dynamics in astrocytes and play a role in Ca2+-dependent glutamate release from astrocytes
Li, Parpura, Ding et al., 2014	Cell culture	Physiology	Novel approach to imaging mitochondrial Ca2+ dynamics in astrocytes
Hamilton, Verkhratsky, Butt et al., 2008	P15-P30	Physiology	ATP and glutamate evoke Ca2+ signals in white matter astrocytes of the mouse optic nerve. Action potentials trigger axonal release of ATP, which evokes further release of ATP from astrocytes, and this acts by amplifying the initiating signal and by transmitting an intercellular Ca2+ wave to neighboring glia
Parnis, Parpura, Nolte et al., 2013	Cell culture	Physiology & injury	NCLX plays a major role in mitochondrial Ca2+ extrusion, ATP-induced cytosolic Ca2+ release from organelle, astrocytic wound closure and proliferation
Ioannou et al., 2019	Primary neurons & astrocytes culture	Fatty acid toxicity	Fatty Acid metabolism is coupled in neurons and astrocytes to protects neurons from FA toxicity during periods of enhanced activity
Brown, Schousboe, Ransom et al., 2005	Adult Swiss Webster mice	Glucose deprivation	Astrocyte glycogen metabolism is required to maintain axon function (CAP) after glucose deprivation in white matter
Joshi, Mochly-Rosen et al., 2019	Cell culture	Neurodegenerative diseases	Damaged mitochondria released by microglia in a neuro-inflammatory context, trigger A1 astrocytic response and propagate inflammatory neurodegeneration
Oksanen et al., 2017	iPSC-Derived Model	Alzheimer disease	AD astrocytes manifest hallmark of AD pathology, and due to altered metabolism show increased oxidative stress, and compromised neuronal supportive function.
Kawamata et al., 2014	Primary astrocytes culture	ALS	Oxidative stress causes excess calcium release and altered astrocytes response through SNARE-dependent exocytosis via enhanced stored operated calcium entry and ER calcium overload.
Barodia et al., 2019	Astrocyte culture	Parkinson disease	Astrocytes contribute significantly to PD pathology due to high level of PINK-dependent ubiquitin phosphorylation.
Polyzos et al., 2019	12–16 weeks; 80 weeks	Huntington disease	Region-specific astrocytes response to altered glucose level and utilizing alternative fuel source play important role in the susceptibility of neuronal death in HD animal model.
Ishii et al., 2017	4–8-months-; 10–14-months-old	Aging	New model of aging brain: chronic oxidative stress causes astrocyte defects in mice with impaired mitochondrial electron transport chain functionality
Clarke, Barres et al., 2018	P7; P30; 10-weeks-; 9.5months-; 2-years-old	Aging	Astrocytes display phenotype of neuro-inflammatory Al-like reactive astrocytes, which would contribute to vulnerability of the aged brain to injury
Hayakawa, Lo et al., 2016	12–14 weeks	Focal cerebral ischemia	Astrocytes release functional mitochondria to neurons; mediated by CD38 and cyclic ADP ribose signaling to amplify cell survival signals. Neuroprotection, neuro-recovery mechanisms after stroke
Quintana, Simpkins et al., 2019	Primary astrocytes culture	Hypoxia	Hypoxia increases mitochondria fission and fragmentation corresponded to Drpl dephosphorylation at Ser 637. Reoxygenation marks the initiation of elevated mitophagic activity at the perinuclear region where a large number of the smallest mitochondria occurred.
Fiebig, Beckervordersandforth et al., 2019	4 months	Physiology & stroke	Under physiological conditions, astrocytes survive to dysfunctional mitochondrial ETC and oxPhos, but under injury require functional machinery for proliferation and neuroprotection
O’Donnell, Jackson and Robinson, 2016	Hippocampal slices culture from P6-P8 rats	Oxygen Glucose deprivation & Excitotoxicity	Transient OGD causes delayed excitoxic death of CA1 pyramidal neurons, delayed loss of mitochondria and increased Ca2+ signaling at astrocytic processes, and increased colocalization of phagosome marker LC3B
Ren, Zhang et al., 2020	8 months	Traumatic brain injury	In Astrocytes, RvDl induces higher mitophagy of damaged mitochondria, elimination of extra mitochondria-derived ROS, protects mitochondria morphology and astrocyte membrane potential, allowing an enhanced survival of neurons
Constantinou, Fern, 2009	P10 “throughout”	Overactivation of neurotransmitter receptors	Overactivation of neurotransmitter receptors induces swollen and expended mitochondria, vacuolization, membrane breakdown in white matter astrocyte of developing CNS
Shu et al., 2019	Primary hippocampal astrocyte culture	Depression; Chronic mild stress	Antidepressant fluoxetine protects astrocyte in CMS model mice by increasing clearance of damaged mitochondria

Astrocytes and oligodendrocytes originate from the embryonic ectoderm, while microglia originate from the mesoderm and enter the vertebrate brain during embryogenesis. Advances in counting techniques have demonstrated that while the overall ratio of neurons to glial cells varies between different regions in the brain, a ratio of ~1:1 glia to neuron exists in the entire human brain [170]. Oligodendrocytes are responsible for axon myelination, providing axons with an “insulating coat” that enhances nerve impulse conduction interrupted at regular intervals by internodal segments of myelin separated by gaps (nodes of Ranvier) [171]. Oligodendrocytes are found in both gray matter and white matter, but they are a major fraction of all the cells in white matter. Microglial cells are resident macrophages distributed throughout the central nervous system (CNS) [172]. As innate immune cells, microglia are activated by infection, tissue injury, or xenobiotics. Upon activation, microglia retract their cytoplasmic extensions and migrate to the site of injury, where they proliferate and become antigen-presenting cells. In astrocytes, IFNγ stimulation increased the expression of MHCII, while endocytosis is inhibited to prolong surface retention of antigens [173]. Microglia phagocytose degenerating cells and act as sources of immunoregulatory and neuromodulatory factors such as cytokines, chemokines, and neurotrophic factors. Microglia can be activated by cell-surface receptors for endotoxins, cytokines, chemokines, misfolded proteins, serum factors, and ATP. While mild activation is a key adaptive immune response, continuous activation or overactivation of microglia is thought to contribute to neurodegeneration [174–176]. Despite reports of apoptosis in astrocytes and microglia under different experimental conditions, there is little information regarding the loss or degeneration of these glial cells with respect to human disorders. Conversely, oligodendrocytes are known to degenerate in demyelinating disorders such as multiple sclerosis, and to be affected directly or indirectly by most known disorders in the CNS including ischemia, trauma, and neurodegeneration. Glutamate/Ca²⁺ excitotoxicity, inflammation (cytokines), and oxidative stress are common triggers for oligodendrocyte injury in these pathological situations. The high lipid and iron content of oligodendrocytes also makes them susceptible to oxidative damage induced by cytokines [177]. Importantly, mitochondrial respiration/metabolism seems to be primarily involved in oligodendrocyte differentiation, while glycolysis appears to be sufficient to maintain post-myelinated (differentiated) oligodendrocytes [178]. Accordingly, demyelination disorders linked to mitochondrial dysfunction seem to be primarily linked to increased oxidative damage and changes in Free Fatty Acid (FFA) metabolism, but not energy failure [179–181].

Dysfunction of glial mitochondria can be detrimental to white-matter function and can trigger and participate in various neurodegenerative diseases. The mitochondrial Ca²⁺ signaling pattern, and the capacity to initiate and contribute to inflammation and oxidative stress together encompass major injury mechanisms contributing to various disease pathologies. However, very little is known about the impact of mitochondrial Ca²⁺ homeostasis on glial signaling. As in other cell types, functional mitochondria in astrocytes and oligodendrocytes regulate Ca²⁺ waves generated by the activation of inositol 1,4,5-triphosphate (IP3) receptors (IP3R) and the release of Ca²⁺ from the endoplasmic reticulum (ER) [182–184]. Mitochondrial Ca²⁺ has also been shown to regulate vesicular glutamate release from astrocytes, which modulates synaptic communication and excitability [185]. Ca²⁺ accumulation in mitochondria also modulates oxidative phosphorylation and energy production. Ca²⁺ release from the ER stimulates mitochondria-dependent energy production in astrocytes [186]. A recent report demonstrated that Ca²⁺ release via NCX is coupled to store-operated Ca²⁺ entry (triggered by Ca²⁺ depletion from ER stores) and regulates astrocyte proliferation and excitotoxic glutamate release [72, 187, 188]. Therefore, not only do mitochondria regulate Ca²⁺ accumulation and dynamics, but also its release. Ultrastructural analysis has revealed that mitochondria of white-matter astrocytes are more elongated than those in gray-matter astrocytes [189], but how this contributes to cell-to-cell interaction and function is yet to be explored. Note that in addition to its local impact, astrocyte networking can aggravate and disseminate mitochondrial Ca²⁺ signaling away from the center of injury, recruiting more cells and contributing to the progression of neurodegenerative diseases in white matter.

It has been established that mitochondrial dysfunction triggers inflammatory responses mainly due to changes in microglial mitochondrial metabolism following activation. Consequently, classic activation of microglia (M1-like phenotype) was recently reported to be paralleled by a metabolic switch from mitochondrial OXPHOS to glycolysis that enhances carbon flux to the Pentose Phosphate Pathway (PPP) [190–192]. Interestingly, inhibition of complex I activity activates microglial cells [193–195], while impairment of mitochondrial fission reduces the production of pro-inflammatory signals [196]. Induction of the M2-like phenotype results in no observable changes in mitochondrial oxygen consumption or lactate production [191]. However, mitochondrial toxins such as 3-nitropropionic acid and rotenone impair the transition to the M2-like phenotype induced by IL-4 [197]. These results suggest that mitochondrial dysfunction in microglia can exacerbate the pro-inflammatory M1 phenotype and result in the release of neurotoxic pro-inflammatory cytokines, and enhanced ROS/RNS formation [198]. Pro-inflammatory cytokines released from microglia also “activate” astrocytes, which might also produce TNFα to potentiate microglia activation. Inflammation is a key contributor to most neurological disorders. As a result, co-cultures of microglia and astrocytes produce more neurotoxic factors than either activated cell type alone [199]. Whether astrocytes can be activated in the absence of microglia is still unclear since most studies using primary cultures of astrocytes also contain at least 5% of microglia that significantly contribute to astrocyte activation [200, 201].

The enhanced resistance to oxidative damage in astrocytes is observed despite the fact that astrocyte mitochondria have deficient mitochondrial respiration and increased ROS formation when compared to neurons [202]. Interestingly, a comparative study demonstrated that astrocytes are more resistant to oxidative damage than microglia or oligodendrocytes [203]. Astrocytes contain higher levels of endogenous antioxidants and antioxidant systems that include NADPH and G6PD (glucose-6-phosphate dehydrogenase). The importance of astrocytes for neuronal redox homeostasis was established by a recent study demonstrating that conditional depletion of astrocytes promotes neuronal injury by oxidative stress [204]. This study raises the question of what is the role of mitochondria in redox homeostasis in astrocytes and neurons? The loss of GSH by its export to neurons or due to the detoxification of electrophiles is expected to prompt astrocytes to replenish GSH precursors. Interestingly, GSH depletion upregulates mitochondrial activity in astrocytes [205], but the exact mechanisms that regulate this phenomenon are still unclear.

This brief overview of the role of mitochondria in glial cell function that includes metabolism, redox homeostasis, Ca²⁺ signaling, inflammation, and cell death, clearly indicates the importance of mitochondrial health in glial cells and its relevance to neuronal function. Yet, this review also highlights our limited understanding of mitochondria function in glial cells and the need for further investigations in this rapidly-expanding area. Many questions remain to be answered regarding the role of mitochondria in neurological disorders, suggesting that it is time to think about mitochondrial health and dysfunction in a more inclusive context outside of neuronal cells. It has become clear that mitochondrial mtPTP plays a key role in a wide variety of human diseases whose common pathology may be based on mitochondrial dysfunction triggered by Ca²⁺ and potentiated by oxidative stress [206]. Reactive oxygen species cause axonal degeneration and a reduction in axonal transport, leading to axonal dystrophies and neurodegeneration including Alzheimer’s disease [207], amyotrophic lateral sclerosis [208], Parkinson’s disease [209], and Huntington’s disease [210]. Given the importance of axonal transport for the preservation of axonal integrity, surprisingly little is known about how oxidative stress affects axonal transport, and whether this contributes to the damaging effects of elevated levels of ROS.

Recent advances investigating small-molecule inhibitors for mtPTP represent compounds of high therapeutic value since mtPTP activation and opening forms a shared target for numerous diseases [206]. Consequently, the search for targeted small-molecule therapeutics for some of the most wide-spread and therapeutically challenging human diseases, such as multiple sclerosis, amyotrophic lateral sclerosis, Alzheimer’s disease, Parkinson’s disease, Huntington’s Disease, and stroke is advancing. For instance, ischemia-reperfusion injury is a key disorder in which mtPTP opening plays a prominent role in ischemic damage of any tissue, which is most pronounced in ischemic damage of the heart and brain. Excitotoxicity, which is a major pathway of ischemia-reperfusion injury, is characterized by the excessive entry of Ca²⁺ into neurons that can be primarily triggered by glutamate and NMDA receptor activation inducing mtPTP opening. Alzheimer’s disease is the most common form of mental disability in the elderly and the merging of various mechanisms leading to Ca²⁺ dyshomeostasis activates mtPTP, initiating the apoptosis of neurons and nearby cells. Dopaminergic neurons are distinctively reliant on voltage-dependent L-type Ca²⁺ channels for independent pacemaking activity and tonic dopamine release [211]. Consequently, these cells are particularly vulnerable to perturbations in the Ca²⁺ buffering capacity of mitochondria, which lead to mtPTP opening in Parkinson’s patients. Huntington’s disease, a progressive genetic disorder that results in motor, cognitive, and psychiatric disturbances caused by mutations in the gene encoding huntingtin (Htt) that ultimately leads to death in adulthood, is another example that appears to involve mtPTP-dependent mitochondrial defects in its pathogenesis. In amyotrophic lateral sclerosis, affected motor neurons show mitochondrial swelling and fragmentation, and mitochondrial Ca²⁺ induces abnormal membrane depolarizations that lead to mtPTP opening. Multiple sclerosis is the most common disabling disease of young and middle-aged adults, and axonal degeneration is a critical part of the pathogenesis of MS and a major determinant of permanent disability. Axoplasmic Ca²⁺ overload driven by ionic imbalances and ROS is postulated to lead to mitochondrial dysfunction and result in pathologic opening of the PTP, which ultimately may be critical to axonal degeneration in MS [211]. Therefore, there is a long-list of human pathologies in which such mtPTP inhibitors may play a crucial role. Note that contributing mitochondrial dysfunction has conventionally been attributed to neurons and axons. The importance of glia-axon interactions in terms of metabolism, signaling, and function is now recognized, but questions as to what the role of astrocyte mitochondria is in the pathogenesis of these diseases currently remain unanswered. Future research addressing these questions will reveal a better understanding of neurodegenerative diseases and identify novel therapeutic targets.