Programming axonal mitochondrial maintenance and bioenergetics in neurodegeneration and regeneration

Abstract

Mitochondria generate ATP essential for neuronal growth, function, and regeneration. Due to their polarized structures, neurons face exceptional challenges to deliver mitochondria to and maintain energy homeostasis throughout long axons and terminal branches where energy is in high demand. Chronic mitochondrial dysfunction accompanied with bioenergetic failure is a pathological hallmark of major neurodegenerative diseases. Brain injury triggers acute mitochondrial damage and a local energy crisis that accelerates neuron death. Thus, mitochondrial maintenance defects and axonal energy deficits emerge as central problems in neurodegenerative disorders and brain injury. Recent studies have started to uncover the intrinsic mechanisms that neurons adopt to maintain (or reprogram) axonal mitochondrial density and integrity, and their bioenergetic capacity, upon sensing energy stress. In this review, we discuss recent advances on how neurons maintain a healthy pool of axonal mitochondria, as well as potential therapeutic strategies that target bioenergetic restoration to power neuronal survival, function, and regeneration.

Introduction

Neurons are among the most specialized cells with unique morphologic, functional, and bioenergetic properties. First, they have a polarized geometry with three distinctive compartments: the soma, complex dendritic arbors, and an extremely long axon with extensive branches and terminals. A single nigrostriatal dopaminergic (DA) neuron in the rat brain with widely spread and highly dense axonal arborizations can extend to approximately a half meter long if all axon fibers are summed (Matsuda et al., 2009) (Figure 1A). Similarly, in the mouse peripheral nervous system (PNS), a cutaneous sensory neuron can contact ~200 hair follicles through its axon arbors containing several thousand branch points with a total length of up to one meter (Wu et al., 2012) (Figure 1B). Such an extensive architecture is even more intricate in human brains. Second, neurons are postmitotic and differentiated cells that can live through the entire life of an organism, and thus must maintain axonal energy homeostasis during their lifetime.

Figure 1. Mitochondrial Biology Essential for Axonal Mitochondria Maintenance.

(A, B) Axonal arborizations of a rat nigrostriatal DA neuron in the striatum (A), where red and blue lines indicate the axon fibers located in the striosome and matrix compartments, respectively; and a mouse cutaneous sensory neuron (B), where the outer arbor of the cutaneous sensory afferents with large areas and free endings at a depth of ~10 micron from the skin surface connecting > 50 vertical branches. Adapted with permission from Matsuda et al (2009) and Wu et al (2012), respectively.

(C) Mitochondrial biogenesis contributes to axon mitochondrial maintenance. Reduced energy supply and/or increased energy consumption leads to an elevated AMP/ATP ratio, which activates AMPK. AMPK promotes the activity or expression of PGC-1α and its downstream signaling pathways either directly or via potentiating the activity of SIRT1 and its subsequent deacetylation of PGC-1α. The increased level of phosphorylated PGC-1α in the nucleus stimulates mitochondrial biogenesis and respiration by activating transcription factors, including nuclear respiratory factor-1/2 (NRF1/2), and by forming heteromeric complexes with nuclear hormone receptors such as PPARγ. NRF1/2 promote the expression of nuclear-encoded mitochondrial transcription factor A (TFAM), which upregulates expression of mtDNA-encoded proteins. Axonal mitochondrial mRNA translation is also upregulated via the PGC-1α pathway. Axonal late endosomes (LE) carry RNA granules and ribosomes and dock on mitochondria, serving as sites for de novo mitochondrial protein synthesis.

(D) Mitochondrial fission and fusion help maintain axonal mitochondrial distribution and integrity. DRP1 is recruited from the cytosol to the OMM, where it assembles with adaptors FIS1, MFF, MiD49, and MiD51 to constrict the mitochondrial tubule and execute fission. ER tubules play a role in aiding DRP1-dependent fission. Fission is necessary to release a healthy mitochondrion from the mitochondrial network in the soma before it undergoes anterograde transport into axon terminals. Fission is also required for mitochondrial biogenesis by generating a new healthy mitochondrion from a preexisting one, as well as for mitophagy by segregating damaged subdomains for local autophagosome engulfment and degradation, or for retrograde transport toward the soma. MFN1/2 and OPA1 are essential to mediate the sequential fusion of the outer and inner membranes of adjacent mitochondria. Fusion helps ease the burden of stressed mitochondria by mixing their content to maximize ATP production and protect stressed mitochondria from mitophagy.

(E) Three pathways of mitochondrial quality control. Left: Misfolded proteins from the intermembrane space (IMS) or matrix or MIM can be degraded by two mitochondrial AAA+ protease complexes, i-AAA and m-AAA, with catalytic sites facing opposite sides of the MIM. Middle: a small portion of damaged proteins can be pinched off through vesicle budding forming mitochondrial-derived vesicles (MDVs), and subsequently transferred to lysosomes for degradation. Right: an entire damaged mitochondrion or a segregated subdomain can be sequestrated within an autophagosome, followed by fusion with lysosomes (Ly) to form an autolysosome for degradation, a process known as mitophagy.

(F) Mitochondrial trafficking and anchoring play a critical role in axon mitochondrial maintenance. Long-distance mitochondrial transport is driven by motors moving along MTs. While kinesin-1 motors drive anterograde transport from the soma toward distal axons, cytoplasmic dynein motors mediate retrograde transport toward the soma. Anchoring is essential for positioning mitochondria at high energy consumption sites to provide stable energy sources. Regulation of mitochondrial trafficking and anchoring help distribute healthy mitochondria in axons and remove damaged mitochondria from axons, thus maintaining axonal energy homeostasis essential for neuronal development and regeneration.

Third, neurons consume a huge amount of energy to power ion channels and pumps that are enriched in axons to maintain resting potentials and fire action potentials (APs) (Harris et al., 2012). These exceptional properties render neurons the most energy-demanding cell type in humans and axons highly vulnerable to bioenergetic failure. It is not surprising that the human brain, which accounts for only 2% of our body mass, consumes nearly 20% of the total cellular energy in the form of adenosine triphosphate (ATP). By analytical calculations, it was predicted that approximately 55% of the total ATP in neurons is consumed within axonal terminals, where approximately 1 × 10⁶ ATP molecules are required to restore ionic gradients and Ca²⁺ levels, and more than 20,000 ATP molecules are consumed for each glutamate synaptic vesicle (SV) recycling (Harris et al., 2012; Rangaraju et al., 2014; Schotten et al., 2015). ATP levels within axonal terminals vary, given dramatic changes in synaptic activity. At resting state, ATP levels within axonal terminals were estimated at 0.6–2.6 mM measured by a luciferase-based Syn-ATP reporter (Rangaraju et al., 2014) and 2–4 mM measured by the fluorescence resonance energy transfer (FRET)-based ATP sensor ATeam (Pathak et al., 2015).

ATP is generated through glycolysis and mitochondrial oxidative phosphorylation (OxPhos). Glycolysis generates two net ATP by converting one glucose molecule into two pyruvate molecules that fuel OxPhos to produce 30–36 ATP molecules through electron transfer coupled with ATP synthase activity in the respiratory chain (Yellen, 2018). Thus, changes in glucose uptake and glycolysis affect OxPhos. In contrast to proliferating cells that mainly utilize glycolysis to produce ATP, neurons primarily rely on mitochondrial OxPhos to supply ATP. Neurons are therefore particularly vulnerable to mitochondrial dysfunction and impaired mitochondrial trafficking to sites with high energy demand, including axonal branches, growth cones, nodes of Ranvier, and presynaptic terminals (Ruthel and Hollenbeck, 2003; Zhang et al., 2010; Tao et al., 2014). Although ATP readily diffuses between adjacent terminals (Pathak et al., 2015), ATP diffusion capacity is quite limited throughout extremely long axons (Hubley et al., 1996; Sun et al., 2013), thus preventing rapid adaptation of ATP supply in response to locally changing energy demand in distal axons. To maintain axonal bioenergetics, neurons deploy complex mechanisms transporting healthy mitochondria into terminals as local energy sources and removing damaged mitochondria from distal areas. This is an emerging theme relevant to a range of neurological disorders and injuries associated with axonal mitochondrial dysfunction and bioenergetic failure.

In this review, we focus on the challenges that neurons face and the signaling pathways they manage to maintain axonal mitochondrial density and integrity in order to preserve axonal energy homeostasis in health, neurodegeneration, and regeneration after injury. We provide an overview of neuronal mitochondrial biology, including biogenesis, fission and fusion, quality control, and trafficking and anchoring, which all work coordinately to maintain a healthy pool of axonal mitochondria. Given space limitations, we next limit our discussion to recent findings and new insights into axonal mitochondrial maintenance with a special focus on trafficking and anchoring under metabolic stress, energy deficits, or pathological conditions. We highlight how impaired maintenance of axonal mitochondria and local bioenergetic failure act as prominent and convergent contributors to the pathogenesis of major neurodegenerative diseases, as well as regeneration failure after brain injuries. At the end of the article, we provide perspectives on the potential therapeutic strategies targeting axonal bioenergetic restoration.

Mitochondrial Biology Essential for Axonal Mitochondria Maintenance

Mitochondrial biogenesis

Mitochondrial biogenesis is a process in which mitochondria grow in number and size. Since mitochondria cannot be generated de novo, mitochondrial biogenesis mainly depends on the incorporation of newly synthesized proteins and lipids into pre-existing mitochondria or through mitochondrial fission. In vertebrates, mitochondria are composed of more than 1000 proteins (Calvo et al., 2016), most of which are encoded in the nucleus and only 13 are encoded by mitochondrial DNA (mtDNA) (Alston et al., 2017). The classic model is that neuronal mitochondrial biogenesis primarily occurs in the soma, where major machineries for DNA replication, RNA translation, and protein synthesis are enriched; these newly generated mitochondria are then transported into axonal arborizations (Davis and Clayton, 1996). Based on mitochondrial mean velocity and low motility in mature neurons (Sheng, 2017), a newly generated mitochondrion in the soma would take days to reach axonal terminals of a long motor neuron. This soma-restricted biogenesis concept has been challenged by emerging lines of evidence showing local mtDNA replication and mRNA translation of mitochondrial proteins in distal axons of the central and peripheral nervous systems (Amiri and Hollenbeck, et al., 2008; Gale et al., 2018; Van Laar et al., 2018). These observations indicate that mitochondrial biogenesis can take place in distal axons to meet local bioenergetic demand for powering axonal growth and branching and synaptic transmission. In retinal ganglion cells (RGCs), axonal late endosomes can carry both RNA granules and ribosomes and dock on mitochondria, serving as sites for de novo mitochondrial protein synthesis (Figure 1C) (Cioni et al., 2019). Except for short-lived proteins with high turnover rate, long-lived mitochondrial proteins, most of which are members of OxPhos complexes, can last for >4 months in mouse brain tissues, contributing to the structural stability of mitochondria (Borrnba-Warczak et al., 2021). However, the relative contribution of somatic vs axonal mitochondrial biogenesis to axonal mitochondrial maintenance is less clear.

PGC-1α and NRF2 are master transcription coactivators that control expression of nuclear-encoded mitochondrial genes (Figure 1C) (LeBleu et al., 2014) as well as increase axonal mitochondrial density and local bioenergetic capacity in a manner dependent on mitochondrial fission (Lin et al., 2021). Energy stress-induced activation of AMP-activated protein kinase (AMPK) can directly phosphorylate PGC-1 and upregulate its own gene expression and mitochondrial RNA expression (Jäger et al., 2007), which can be transported to axons for distal protein translation. Local upregulation of mitochondrial mRNA translation and somatic activation of PGC-1α and NRF2 can also be triggered by mitochondrial stress in an experimental setting that mimics Parkinson’s disease (PD) (Van Laar et al., 2018). Axonal accumulation of phospho-TDP-43, an Amyotrophic Lateral Sclerosis (ALS)-linked RNA/DNA binding protein, decreases distal translation of mitochondrial proteins by sequestering their mRNA, ultimately impairing axonal mitochondrial integrity (Altman et al., 2021). Thus, local mitochondrial biogenesis is regarded as an early response to local energetic stress and benefits axonal mitochondrial maintenance. It would be intriguing to determine whether reprograming mitochondrial biogenesis could contribute to axonal bioenergetic recovery in a broader range of neurodegenerative diseases.

Mitochondrial fusion and fission

Mitochondrial morphology and size are controlled by fission and fusion that maintain their integrity, distribution, and bioenergetic metabolism (Chan, 2020; Wai and Langer, 2016). Fission is the division of a mitochondrion into two smaller ones, which is mediated by the cytosolic GTPase dynamin-related protein 1 (DRP1) (Pitts et al., 1999) through its interaction with mitochondrial adaptors (Loson et al., 2013) and with the aid of actin filaments and endoplasmic reticulum (ER) (Figure 1D) (Kraus et al., 2021). The four known adaptors include fission protein 1 (FIS1) (Mozdy et al., 2000), mitochondrial fission factor (MFF) (Otera et al., 2010), and mitochondrial dynamics proteins of 49 and 51 kDa (MiD49 and MiD51) (Otera et al., 2016). Mitochondria form an interconnected network in the soma, whereas in axons they are vesicular or short tubular structures distributed in a linearly interspersed pattern. Therefore, mitochondrial fission in the soma facilitates their transport into axon terminals (Berthet et al., 2014). Generally, a mitochondrion of larger size has higher bioenergetic and Ca²⁺ buffering capacity when compared to a smaller one. However, mitochondria of smaller size display higher motility along axons and are more readily located at axonal branch sites and synapses (Kang et al., 2008). Activation of MFF is required to maintain the smaller size of mitochondria for delivery into axon branches (Lewis et al., 2018). DRP1-dependent fission is also involved in axonal mitochondrial biogenesis (Lin et al., 2021) and mitophagy by dividing mitochondria into fragments amenable to autophagosome engulfment or by segregating damaged subdomains for degradation (Twig et al., 2008; Burman et al., 2016). Interestingly, these two distinct fates of mitochondrial fission — biogenesis vs mitophagy — can be predetermined by the site of fission. Division at the midzone more likely leads to biogenesis, while division at the periphery is more prone to mitophagy (Kleele et al., 2021). Although DRP1 is engaged in both types of fission, MFF is involved in the former event while FIS1 is required for the latter.

Fusion is the merging of two mitochondria into one, typically through end-to-end collision, and is coordinated by three large GTPases: optic atrophy 1 (OPA1) for inner membrane fusion and mitofusin 1 and 2 (MFN1 and 2) for outer membrane fusion (Figure 1D) (Alexander et al., 2000; Delettre et al., 2000; Song et al., 2009). Fusion helps ease the burden of stressed mitochondria by mixing their content to maximize ATP production and protect stressed mitochondria from mitophagy (Chan, 2020; Rambold et al., 2011). Thus, fission and fusion dynamics serve as an additional layer of regulation for axonal mitochondrial maintenance. Disrupted fission/fusion dynamics impair mitochondrial transport, distribution, and integrity in axons; these detrimental effects are mostly observed in axonal terminals, given their high energy demand. For example, Drp1 DATcre KO mice exhibit depleted mitochondria in DA neuron axons (Berthet et al., 2014). Similarly, loss of Drp1 in hippocampal CA1 neurons in Drp1 CamKIIcre KO mice disrupts axonal mitochondria distribution, leaving neurons unable to maintain ATP homeostasis at axonal terminals when energy consumption is increased by synaptic activity (Shields et al., 2015). Inducible Drp1 deletion in adult mouse forebrain neurons reduces mitochondrial content and ATP production at presynaptic terminals, impairing short-term plasticity and synaptic transmission under high frequency stimulation (Oettinghaus et al., 2016). Similarly, Drp1 mutant Drosophila neuromuscular junctions (NMJs) show depleted mitochondria and reduced local ATP (Verstreken et al., 2005). Human iPSC-derived DA neurons expressing a PD-linked OPA1 mutant display a significant loss of axonal mitochondria with trafficking defects, leading to a prominent reduction in the number of synapses (Iannielli et al., 2019). Expressing Charcot-Marie-Tooth type 2A syndrome (CMT2A)-associated MFN2 mutants impairs axonal mitochondrial transport, resulting in the clustering of fragmented mitochondria in proximal axons (Baloh et al., 2007). Altogether, these studies suggest a crucial role of mitochondrial fission/fusion dynamics in axonal mitochondrial maintenance, energy homeostasis, and synaptic transmission.

Mitochondrial quality control

Chronic mitochondrial dysfunction is a central problem associated with major neurodegenerative diseases, where progressive pathological stresses lead to energy deficits and axon degeneration. Mitochondria are the major sources of reactive oxygen species (ROS), posing mitochondrial proteins at greater risk of ROS-induced damage (Shadel and Horvath, 2015). Thus, recovery of stressed mitochondria or turnover of damaged ones constitutes a critical step in the maintenance of axonal bioenergetics. There are three major pathways for maintaining mitochondrial integrity based on the severity of their damage. Misfolded mitochondrial proteins can be degraded by mitochondrial AAA+ proteases in the mitochondrial inner membrane (MIM) and matrix (Song et al., 2021) (Figure 1E, left). Removing mitochondrial-derived vesicles (MDVs) that contain oxidized and/or damaged proteins and lipids is a second mechanism for mitochondrial quality control and for the steady-state turnover of assembled mitochondrial protein complexes (Sugiura et al., 2014) (Figure 1E, middle). After budding off from mitochondria, a process mediated by Mitochondrial Rho GTPase (MIRO) 1/2-driven membrane protrusions along microtubules (MTs) and DRP1-dependent scission, MDVs are targeted to lysosomes for degradation (König et al., 2021).

The third quality control pathway is through mitophagy, a specialized autophagy targeting an entire or fragmented mitochondrion for degradation (Pickles and Youle, 2015) (Figure 1E, right). One key mitophagy mechanism is activated by PTEN-induced kinase 1 (PINK1; PARK6) and the cytosolic E3 ubiquitin ligase Parkin (PARK2) (Clark et al., 2006; Park et al., 2006; Poole et al., 2008), both of which are genetically associated with familial PD (Kitada et al., 1998; Valente et al., 2004). Upon mitochondrial depolarization, PINK1 is stabilized on the outer mitochondrial membrane (OMM), where it triggers Parkin recruitment and a series of phosphorylation and ubiquitination events of OMM proteins, subsequently recruiting the autophagosomal membrane and associated proteins (Narendra et al., 2008; Kondapalli et al., 2012; Sarraf et al., 2013; Lazarou et al., 2015). Most of our knowledge regarding mitophagy has been gained through studying proliferating cell lines that generally use glycolysis as their primary ATP source, thereby global activation of mitophagy would unlikely cause an acute energy crisis. In contrast, ATP supply via glycolysis is rather limited in neurons (Hall et al., 2012; Yellen, 2018); this bioenergetic feature necessitates neurons to preserve energy homeostasis by maintaining or recovering mitochondrial integrity, at least in the early stages of mitochondrial stress. Indeed, when treated with a similar dose of depolarizing agents used in proliferating cell types, onset of PINK1/Parkin-mediated mitophagy in primary neurons is significantly delayed and spatially restricted to the soma (Cai et al., 2012; Van Laar et al., 2011). A recent study revealed that not all mitochondria with Parkin recruitment are destined for degradation — a subset retains mitochondrial function with polarized membrane potential (Li et al., 2021). These findings argue for an early protective mechanism that helps neuronal mitochondria resist insults and/or recover from depolarization. This may be an interpretation of why Parkin KO DA neurons remain viable and Parkin/PINK1/DJ-1 triple-KO mice lack DA degeneration (Goldberg et al., 2003; Kitada et al., 2009; Perez and Palmiter, 2005).

When neurons are under chronic and mild stress — a condition relevant to aging-associated progressive neurodegeneration, they may adopt an early protective mechanism for recovering stressed mitochondria in axons that are highly vulnerable to bioenergetic failure. It is presumable that axonal mitophagy is the last resort for quality control when mitochondria are severely damaged and recovery mechanisms have failed. A study using a neuronal ischemic injury model showed that axonal mitochondrial dysfunction triggers their retrograde transport to the soma for mitophagy (Zheng et al., 2019). Consistently, applying mild oxidative stress in rat hippocampal neurons preferentially induces somatic mitophagy but rarely axonal mitophagy (Evans and Holzbaur, 2020). Blocking mitophagy in Parkin or Pink1 mutant Drosophila motor and sensory neurons does not increase axonal mitochondrial density, but instead results in the accumulation of damaged mitochondria in the soma (Devireddy et al., 2015; Lin et al., 2021; Sung et al., 2016), suggesting that axons are not the primary site for mitochondrial turnover via mitophagy.

These studies suggest an alternative mechanism as the first line of surveillance for axonal mitochondrial integrity. One attractive explanation for this axon-specific feature is that both axonal functions and PINK1/Parkin-mediated mitophagy rely on sustainable ATP supply. Activation of axonal mitophagy would quickly deplete local mitochondria and reduce ATP supply. Thus, neurons likely adopt a bioenergetic restriction on axonal mitophagy activation, as they cannot tolerate acute removal of large pools of axonal mitochondria in a short time-window (Van Laar et al., 2011). This assumption is supported by a recent study in OPA1 mutant nematode GABAergic motor neurons and mouse RGCs, where excess mitophagy reduces axonal mitochondrial density and triggers neurodegeneration (Zaninello et al., 2021). In contrast, inhibiting autophagic activity restores axonal mitochondrial load and protectes neurons from mitochondrial dysfunction. It is worth noting that when mitochondria are depolarized focally within a small area of axons, and thus energy homeostasis would not be globally disrupted, neurons may bypass the bioenergetic restriction to activate axonal mitophagy (Ashrafi et al., 2014) (Figure 1F).

In mature neuron axons where the majority of mitochondria remains stationary, mildly stressed mitochondria can be remobilized by releasing the axonal mitochondrial anchor syntaphilin (SNPH, discuss further in next sections) (Figure 1F), thus facilitating their retrograde transport for recovery either by fusing with healthy mitochondria, or degradation in the soma where mature lysosomes are enriched (Lin et al., 2017). Stress-enhanced Mito-ER contacts may also help maintain mitochondrial integrity (Puri et al., 2019). These mechanisms ensure that mitophagy is restrained in axons to preserve local energy supply in the early stages of disease.

Mitochondrial trafficking and anchoring

Mitochondrial trafficking and anchoring add yet another critical layer of regulation for axonal mitochondrial maintenance. Neurons deploy specialized mechanisms to overcome the complex challenges of maintaining bioenergetic homeostasis in distal axons. Specifically, these mechanisms enable neurons to distribute mitochondria at growth cones and axonal branches during developmental stages, and recruit and position mitochondria at active synapses and nodes of Ranvier in mature neurons (Li et al., 2020; Li and Sheng, 2022) (Figure 1F). In mature or adult neurons, mitochondrial remobilization is essential for removing injury-damaged mitochondria from distal axons and replenishing healthy ones to these regions, thus recovering neurons from a local energy crisis (Han et al., 2020; Huang et al., 2021; Zhou et al., 2016).

Axonal mitochondria in cultured developing neurons display complex motility patterns with highly variable mean velocities ranging from 0.2 to 2.0 μm/sec that may be attributed to a “saltatory motility” mode: mitochondria move bidirectionally, pause and change direction (Sheng, 2014). This saltatory movement is achieved by a set of motors, adaptors, and anchors that work coordinately (Figure 2A) (Text Box 1). As neurons mature both in vitro and in vivo, axonal mitochondrial motility dramatically declines under physiological conditions, and a large pool of axonal mitochondria enters a steady phase in regions of high energy consumption, such as branch points, presynaptic terminals, and nodes of Ranvier (Lewis et al., 2016; Ohno et al., 2014; Smit-Rigter et al., 2016; Vagnoni and Bullock, 2016).

Figure 2. Ca²⁺ and AMPK Signaling Pathways Regulating Axon Energy Maintenance.

(A) Axonal mitochondria display complex motility patterns: they move bidirectionally, pause and change direction frequently, and can be re-mobilized and re-distributed. This motility behavior is achieved by a set of motors, adaptors, and an anchor that work coordinately in response to changes in neuronal activity, as well as metabolic and energy status. Thus, motors, adaptors, and the anchor are central targets of various signaling pathways.

(B) AP firing and Ca²⁺ signaling arrest axonal mitochondria. Neurons adopt a Ca²⁺-sensitive mechanism positioning mitochondria at Ca²⁺ entry sites. Ca²⁺ binding to the two EF-hand motifs of MIRO1 disrupts KIF5-adaptor coupling or KIF5-MT engagement, thus temporarily arresting mitochondria. SNPH stably anchors axonal mitochondria by inhibiting motor ATPase activity. The ‘MIRO1-Ca²⁺-SNPH’ model illustrates the molecular interplay between motor-adaptor and anchor which ensures stable arrest of mitochondria in axons.

(C) Neurons adopt an energy-sensitive AMPK signaling mechanism to recruit and retain presynaptic mitochondria. Sustained synaptic activity induces presynaptic ATP depletion, which activates the master energy stress sensor AMPK-PAK signaling axis. This signaling axis facilitates myosin-6-driven mitochondrial recruitment and SNPH-mediated anchoring on presynaptic F-actin, thus maintaining ATP supply during intensive synaptic activity. This signaling process represents a crosstalk between energy sensing (AMPK-PAK5) and mitochondrial anchoring (Myosin-6-SNPH).

Text Box 1. Axonal Mitochondrial Trafficking Machineries.

Long-distance axonal mitochondrial transport is driven by molecular motors traveling along microtubules (MTs), which display a uniform polarity within axons: the plus-ends are directed distally and minus-ends toward the soma (Baas et al., 1988). Thus, kinesin-1 family (KIF5) motors are the primary motors driving mitochondrial anterograde transport toward distal axons, while cytoplasmic dynein motors mediate retrograde mitochondrial transport toward the soma (Pilling et al., 2006; Tanaka et al., 1998). KIF5 motors are connected to mitochondria through adaptors mainly including the MIRO-TRAK complex. MIRO1/2 (Mitochondrial Rho GTPase) are OMM proteins with two GTPase domains and two Ca²⁺-binding EF-hand motifs and connect KIF5 via trafficking kinesin-binding proteins (TRAKs) or its Drosophila ortholog Milton (see reviews by Devine and Kittler, 2018; Misgeld and Schwarz, 2017). Loss of Miro1 in mice causes depletion of mitochondria from CST axons and progressive neurological deficits (Nguyen et al., 2014). TRAK1/2 or Milton link KIF5 motors at one end and interact with MIRO at the other end, thus loading KIF5 motors on mitochondria (Brickley et al., 2011). Interestingly, TRAK1 is mainly localized in axons and binds both kinesin and dynein motors, whereas TRAK2 is localized to dendrites and preferentially binds dynein motors (van Spronsen et al., 2013). Consistently, depleting TRAK1 inhibits axonal outgrowth, while disrupting TRAK2 impairs dendrite morphology. A recent study further showed that KIF5-driven axonal transport is dependent on its interaction with TRAK1 (Henrichs et al., 2020). In Drosophila motor and sensory neurons, milton mutants display mitochondrial depletion in axons and at synapses (Stowers et al., 2002; Glater et al., 2006). Thus, the MIRO-TRAK-KIF5 complex acts as the key transport machinery that drives mitochondrial anterograde transport in axons (Figure 2A). Syntabulin is an alternative adaptor for KIF5 motors in driving mitochondrial transport into axons (Cai et al., 2005). Syntabulin targets to OMM through its C-terminal transmembrane domain and interacts with KIF5 via its kinesin-binding domain; knockdown of syntabulin impairs mitochondrial transport from the soma into axons. A recent genetic screen in C. elegans identified two OMM proteins named Metaxin1/2 that contribute to mitochondrial transport by forming a complex with MIRO and KIF5 (Zhao et al., 2021). The presence of multiple motor-adaptor complexes may reflect the complex regulation of mitochondrial transport under various conditions or in different nerve systems.

Both KIF5 and dynein motors likely share the same set of MIRO-TRAK adaptor complexes, thus coordinating opposing motor activity to drive bidirectional mitochondria transport (Figure 2A). Initial insights into this paradigm came from the observations that disrupting either KIF5, dynein, or dynactin was sufficient to impair bidirectional mitochondrial transport (Martin et al., 1999; Piling et al., 2006). This assumption was further supported by two studies in Drosophila where loss of dmiro impairs both anterograde and retrograde mitochondrial transport in axons (Guo et al., 2005), while overexpressing dMiro enhances bidirectional transport (Russo et al., 2009). In C. elegans, the MIRO1-TRAK complex is indispensable for dynein-driven mitochondrial transport (Zhao et al., 2021). A direct interaction of TRAKs with the dynactin p150^Glued subunit was also revealed (van Spronsen et al., 2013). Both KIF5 and dynein-dynactin were recently reported to functionally interact with TRAK2 (Fenton et al., 2021); TRAKs coordinate the recruitment and activation of dynein and KIF5 to drive bidirectional transport (Canty et al., 2021). Thus, by assembling a super motor complex, KIF5 and dynein work coordinately to drive bidirectional mitochondria transport.

Mitochondrial anchoring is essential for arresting mobile mitochondria and positioning them at high energy consumption sites to provide stable energy sources. One well-established static anchor specific for axonal mitochondria is SNPH (Figure 2A) (Kang et al., 2008; Li et al., 2020). SNPH targets axonal mitochondria through its OMM-insertion domain and axon-sorting sequence, and anchors mitochondria on MTs through its MT-binding domain. Overexpressing SNPH in mouse hippocampal and cortical neurons abolishes axonal mitochondrial transport, whereas deleting Snph in mice robustly enhances axonal mitochondrial motility by up to 78% in mature neurons in vitro and 71% in adult brains (Chen and Sheng, 2013; Zhou et al., 2016). Interestingly, SNPH expression is developmentally regulated: it is undetectable at embryonic stages but becomes readily detectable at postnatal day 7 (P7) and reaches its peak at P14. Its expression pattern is inversely associated with the progressive decline in the motility of axonal mitochondria with neuron maturation. An optogenetic study using a light-induced heterodimerization system (LOV-PDZ) demonstrated that the balance between mobile and stationary pools of axonal mitochondria can be shifted by selective recruitment of KIF5 or SNPH onto axonal mitochondria (van Bergeijk et al., 2015). Consistently, a recent reconstitution study further showed that adding SNPH onto mitochondria is sufficient to oppose kinesin-driven forces, immobilizing motor-loaded mitochondria (Canty et al., 2021). These studies support the notion that axonal mitochondrial maintenance is intricately regulated by coordination between trafficking and anchoring mechanisms (Figure 2A), where their motile vs stationary status could change as SNPH expression is elevated with neuron maturation (Zhou et al., 2016), or its anchoring is switched off in mature neurons upon injuries (Huang et al., 2021).

Signaling Pathways Regulating Axonal Mitochondrial Maintenance

Mitochondrial positioning ensures an adequate and stable supply of ATP in metabolically active compartments. Trafficking vs anchoring of axonal mitochondria is regulated in order to sense and respond to changes in metabolic demand or alterations in bioenergetic status during neuronal growth, regeneration, synaptic activity, or under mitochondrial stress (see reviews by Devine and Kittler, 2018; Giorgi et al., 2018; Misgeld and Schwarz, 2017; Sheng, 2017). Thus, it is not surprising that motor adaptors and anchors are central targets of the signaling pathways that regulate axonal mitochondrial maintenance.

Growth signaling during neuronal development

Association between enhanced axonal energy demand and mitochondrial delivery is most apparent in developing neurons, where proteins and organelles are robustly synthesized and delivered to support axon outgrowth, branching, and synaptogenesis, all of which are highly energy-demanding processes (Cheng and Sheng, 2021; Spillane et al., 2013). We and others observed higher axonal mitochondrial motility (~20–45%) at early developing stages (Morris and Hollenbeck, 1993; Zhou et al., 2016). A correlation between polarized mitochondrial transport and axonal/dendritic growth has been established. Depleting TRAK1 inhibits axonal outgrowth, while disrupting TRAK2 impairs dendrite morphology (van Spronsen et al., 2013). Deleting Miro1 during mouse development disrupts neuronal morphogenesis (Lopez-Domenech et al., 2016). Overexpressing SNPH restricts mitochondria to proximal axons, lowering ATP/ADP ratios in the most distal axon segments and reducing the size of growth cones (Zhou et al., 2016). This study also showed that overexpressing MIRO1 increases both mitochondrial density in distal axons and growth cone size, suggesting that proper mitochondrial delivery is essential to maintain axon growth capacity. The mitochondrial maintenance in growing axons of sensory neurons may be triggered by nerve growth factor upon binding to its receptor TrkA that induces mitochondrial anchoring on F-actin (Chada and Hollenbeck, 2004).

Calcium signaling in synaptic activity

AP firing triggers synaptic transmission, which requires high energy and induces a transient elevation of intracellular Ca²⁺. Accordingly, neurons adopt a Ca²⁺-sensitive mechanism to arrest mobile mitochondria (Chang et al., 2006) upon MIRO1’s binding to Ca²⁺ through its two EF-hand motifs (Saotome et al., 2008). Ca²⁺ binding disrupts the MIRO1-TRAK-KIF5 coupling or KIF5-MT engagement (Macaskill et al., 2009; Wang and Schwarz, 2009). Thus, MIRO1-Ca²⁺ signaling regulates mitochondrial positioning and homeostatic plasticity in an activity-dependent manner. It is worth noting that two mouse genetic studies independently revealed that MIRO1 is dispensable for activity-dependent inhibition of mitochondrial motility (Nguyen et al., 2014; Lopez-Domenech et al., 2016). Furthermore, in Miro1/2 dKO mouse embryonic fibroblasts (MEFs), mitochondrial movement is reduced but not abolished, and TRAK1/2 can be recruited to mitochondria to mediate anterograde transport in a Miro-independent manner (Lopez-Domenech et al., 2018). AP firing (100 Hz for 1 sec) induces transient Ca²⁺ entry that only temporally arrests mitochondria for ~150 sec (MacAskill et al., 2009), raising the question as how mitochondria can be stably anchored. Interestingly, SNPH can compete with TRAK to bind KIF5 and inhibit motor ATPase activity and thus stably capture mitochondria. Deleting mouse Snph gene abolishes axonal mitochondrial arrest in response to AP firing (Chen and Sheng, 2013), suggesting that MIRO-1Ca²⁺ signaling alone is not sufficient to arrest mitochondria in the absence of SNPH. A revised ‘MIRO1-Ca²⁺-SNPH’ model is therefore proposed to illustrate the molecular interplay between motor and anchor that ensures stable capture of mitochondria in axons (Figure 2B). Consistently, by attaching or detaching Kif1a motor to or from mitochondria with an optical dimerizer zapalog system, a recent study revealed two populations of stationary mitochondria in axons as ‘movable’ and ‘immovable’ that may represent transiently arrested and stably captured mitochondria, respectively (Gutnick et al., 2019).

During nerve impulses, evoked Ca²⁺ elevation and Na⁺ pump activation synergistically arrests mitochondrial motility at nodes of Ranvier. Bathing nerve explants in Ca²⁺ free solutions or blocking the Na/K-ATPase pump abolishes the activity-triggered arrest of axonal mitochondria (Zhang et al., 2010). A second study also showed a dramatic reduction of mitochondrial motility at nodes following stimulation (Ohno et al., 2011). Thus, AP firing and Ca²⁺ transients serve as critical signaling pathways that position axonal mitochondria to support energy-demanding processes in axons and at synapses.

AMPK signaling under energy stress

Neurons adopt special mechanisms enhancing mitochondrial delivery to axons and synapses upon sensing energy stress. Presynaptic mitochondria play an essential role in synaptic transmission by generating ATP and sequestering presynaptic Ca²⁺. A brief interruption in local ATP synthesis or depletion of presynaptic mitochondria impairs synaptic transmission (see review by Li and Sheng, 2022). ADP gradients established in proximity to synapses arrest mitochondria (Mironov, 2009). One attractive energy stress sensor is AMPK, a master regulator of cellular energy homeostasis (Herzig and Shao, 2018). Only ~33% of presynaptic terminals in cultured hippocampal neurons and in hippocampi in vivo retain mitochondria at steady state (Sun et al., 2013; Smith et al., 2016). Therefore, neurons must adopt a mechanism to recruit and capture mitochondria at presynaptic terminals. Synaptic activity activates AMPK signaling to adapt to the rate of glycolysis and mitochondrial respiration (Marinangeli et al., 2018) and facilitate mitochondrial recruitment to presynaptic terminals (Tao et al., 2014) to sustain long-term potentiation (Yu et al., 2016). A recent study revealed a crosstalk between energy sensing AMPK-p21-activated kinase (PAK) and mitochondrial anchoring (Myosin 6-SNPH) (Li et al., 2020). Through this energy signaling axis, presynaptic ATP depletion, induced by sustained synaptic activity, activates AMPK-PAK signaling to facilitate myosin 6-driven mitochondrial recruitment and SNPH-mediated anchoring on presynaptic F-actin, thus maintaining an enhanced local ATP supply during intensive synaptic activity (Figure 2C).

Spatiotemporal axon branching is crucial for developing neurons to establish synaptic connections. Activity-induced ATP deficits also activate AMPK, which increases the anterograde transport of mitochondria to axon branching sites (Tao et al., 2014). Remarkably, SNPH-mediated anchoring underlies mitochondrial accumulation at axon branching sites. LKB1 (liver kinase B1) and NUAK1, an AMPK-like kinase, are required for cortical axon branching in vivo. While the LKB1-NUAK1 signaling pathway replenishes ATP supply at branching sites by recruiting/retaining mitochondria, depleting SNPH abolished this effect (Courchet et al., 2013). Likewise, local energetic stress in distal axons and synapses can trigger selective inhibition of mitochondrial retrograde transport via AMPK signaling, contributing to an increase in mitochondrial density at distal regions where energy is depleted (Watters et al., 2020). Thus, axonal mitochondrial trafficking vs anchoring is tightly regulated during neurite outgrowth, axonal branching, and synaptic activity by sensing local ATP levels. In addition, AMPK signaling activates mitochondrial biogenesis through the AMPK–PGC1α–NRF1 axis to enhance ATP supply in axonal terminals (Vaarmann et al., 2016), further boosting local energy supply.

ROS signaling under oxidative stress

The mitochondrial electron transport chain is one of the major sources of ROS, which are generated from the incomplete reduction of oxygen, and include superoxide anion (O₂⁻), hydrogen peroxide (H₂O₂), and hydroxyl radicals (HO•). While physiological levels of ROS serve as second messengers that facilitate various signaling pathways, excess ROS cause oxidative stress by damaging macromolecules and organelles, a hallmark of aging-associated neurodegeneration (See review by Court and Coleman, 2012; Lopez-Otin et al., 2013). Studies with cultured neurons showed that the external addition of H₂O₂ or mutation of Cu/Zu superoxide dismutase 1 (SOD1), the major scavenger of ROS, is sufficient to diminish mitochondrial motility (Fang et al., 2012). Using Drosophila neuronal culture and the third instar larval nervous system, a study revealed that excess ROS elevates intracellular Ca²⁺, thus arresting mitochondrial motility, likely through MIRO-Ca²⁺ signaling (Liao et al., 2017). A second study using rat hippocampal neurons revealed that ROS decreases mitochondrial motility by activating p38 MAPK kinase and the downstream phosphorylation of mitochondrial adaptors (Debattisti et al., 2017).

Glycosylation under high glucose availability

Neuronal energy metabolism relies on a continuous supply of glucose. Physiological levels of glucose are of spatial heterogeneity along axons and fall within a limited range, rendering mitochondrial anchoring a plausible regulatory mechanism for spatial enhancement of metabolic flux in certain regions (Agrawal et al., 2018). O-GlcNAc transferase (OGT) catalyzes protein OGlcNAcylation, a posttranslational modification that adds O-GlcNAc to Ser/Thr residues. Levels of O-GlcNAcylation are regulated by glucose availability (Mueller et al., 2021). It was reported that TRAK (Milton) serves as a substrate of OGT (Pekkurnaz et al., 2014). In the presence of high local glucose, glycosylation of TRAK (Milton) leads to its increased association with FHL2 (four and a half LIM domains protein 2), which arrests mitochondria in regions with high glucose availability through F-actin accumulation around mitochondria, thus facilitating local mitochondrial ATP production (Basu et al., 2021).

While these signaling pathways separately enable neurons to adapt axonal mitochondrial trafficking and positioning to changes in local metabolic demand and/or bioenergetic status, we should also bear in mind that they can act interdependently. For instance, while transient Ca²⁺ signaling arrests mobile mitochondria, mitochondrial Ca²⁺ accumulation serves as a signaling pathway to enhance mitochondrial energetic capacity by stimulating ATP synthesis, activating OxPhos, increasing respiration, or activating Ca²⁺-dependent mitochondrial enzymes (Ashrafi et al., 2020; Jouaville et al., 1999; Llorente-Folch et al., 2015). Similarly, there is a crosstalk between Ca²⁺ and redox signaling; mitochondrial ROS serve as both a mediator and regulator of Ca²⁺ signaling and thus control mitochondrial function (Gorlach et al., 2015).

Chronic Bioenergetic Failure in Neurodegeneration

Axonal mitochondrial dysfunction is the central problem associated with neurodegenerative diseases, where various chronic pathological stresses induce progressive accumulation of dysfunctional mitochondria in distal axons, leading to chronic energy deficits (Cunnane et al., 2020; Pathak et al., 2013). Thus, early removal or clearance of dysfunctional mitochondria, accompanying by replenishment with healthy ones, constitutes a critical step in axonal bioenergetic restoration (Figure 1F).

Aging-linked changes in axonal mitochondrial maintenance capacity were observed in C. elegans, where mitochondrial density, motility, and stress resistance all decline progressively with age; genetic manipulations that extend lifespan slow this adverse trend (Morsci et al., 2016). In major aging-associated neurodegenerative diseases, including PD, ALS, Alzheimer’s disease (AD), and many others, disease-specific gene mutations alter mitochondrial trafficking and anchoring, disrupt fission and fusion, impair energetic metabolism, or reduce mitophagy and quality control (see reviews by Chan, 2020; Devine and Kittler, 2018; Pickrell and Youle, 2015; Sheng and Cai, 2012). All these changes directly or indirectly reduce axonal mitochondrial maintenance capacity, leading to local bioenergetic deficits. In addition, neurological disorders that broadly affect axon transport also develop phenotypes with progressive bioenergetic failure and ultimately axon degeneration (De Vos et al., 2008). The gain- or loss-of-function mutations of many disease-linked genes induce protein aggregates (such as APP, α-synuclein) that also interfere with mitochondrial integrity, trafficking, or quality control (Bourdenx et al., 2017; Millecamps and Julien, 2013; Norat et al., 2020). Maternally inherited mtDNA mutations cause severe neurological and multisystem disorders by impairing mitochondrial bioenergetics (Carelli and Chan, 2014). Most neurodegenerative diseases are sporadic with numerous interdependent etiologies that may also contribute to axonal energy failure by disrupting mitochondria maintenance or impairing bioenergetic function (Schon and Przedborski, 2011).

It is thus challenging to determine whether defective maintenance of axonal mitochondria is just an epiphenomenon that occurs in a short time window or plays a genuine pathogenic role in facilitating disease progression, or both. Here, we summarize a range of major neurodegenerative diseases with their associated gene mutations, many displaying impaired axonal mitochondrial maintenance and bioenergetic failure (Table 1). Given space limitations, we restrict our discussion to the interconnected adaptive and maladaptive behavior of mitochondrial motility and mitophagy and focus on axonal energy maintenance along the progression of chronic neurodegeneration. We refer readers to other reviews on mitochondrial dysfunction and bioenergetic alterations following programmed axon death — Wallerian degeneration (Merlini et al., 2022) and neuroinflammatory lesions of multiple sclerosis (Witte et al., 2014).

Table1.

Major mitochondria- and disease-associated genes impair axonal mitochondrial maintenance

Gene and associated disease	Known major functions	Disease risk mutations and associated models	Impaired maintenance of axonal mitochondria	References
Mitochondrial fusion and fission
mitofusin 2 (MFN2) Charcot-Marie-Tooth disease Type 2A	Mediate OMM fusion and ER-Mito tethering	Mouse expressing hMFN2T105M in motoneurons	Aggregated mito in axons with uneven distribution.	Bannerman et al., 2016; Detmer et al., 2008
		Mouse expressing hMFN2R94 in neurons	Smaller and fewer mito in axons of distal sciatic nerve, aberrant mito accumulation at terminals.	Cartoni et al., 2010; Strickland et al., 2014
		cKO MFN2 Tg-mouse in hippocampus and cortex	Mito fragmentation and ultrastructural damage, reduced density.	Jiang et al., 2018
		Drosophila expressing hMFN2R94Q/T105M/R364W in neurons	Impaired fusion, mito aggregations, enlarged mito, depletion of mito at NMJ; increased MVBs.	El Fissi et al., 2018
		Sural nerve biopsies from patients	Degenerative changes of mitochondria (smaller size, swelling, aggregation etc.)	Verhoeven et al., 2006
Optic atrophy 1 (OPA1) Autosomal Dominant optic atrophy, Parkinson’s disease (PD)	Mediate IMM fusion, maintain mito crista junctions, limit cyto C mobilization, preserve mtDNA stability.	PD-OPA1A495V iPSC-derived DA neurons	Reduced mito density with aberrant clustering, mito fragmentation, impaired motility, reduced membrane potential.	Iannielli et al., 2019
		RGCs from heterozygous mouse expressing OPA1Q285STOP	Reduced mito length, increased mito motility and velocity, decreased OCR.	Sun et al., 2020
		Rat RGCs and CGCs depletion of OPA1 by siRNA	Mito aggregation.	Kamei et al., 2005
		RGCs from OPA1enu/+ Tg-mouse	Mito with disorganized cristae.	Alavi et al., 2007
		Mouse RGCs expressing with OPA1K301A/R905	Decreased mito density with fragmentation.	Zaninello et al., 2020
		cKO OPA1 Tg-mouse in RGCs (Opa1fl/fl::Grik4-Cre)	Uneven mito distribution.	Zaninello et al., 2020
		Drosophila with pan-neuron OPA1 knockdown	Reduced mito density along axon and at NMJ, reduced motility.	Trevisan et al., 2018
		C. elegans expressing OPA1K301A in motoneurons	Reduced mito contents.	Zaninello et al., 2020
dynamin-related protein 1 (DNM1L/DRP1) Encephalopathy, Refractory Epilepsy, Hypotonia, Dominant optic atrophy.	Mediate mito fission and mito quality control	Drosophila with pan-neuron DRP1 knockdown	Mito elongation, reduced mito at NMJ.	Trevisan et al., 2018
		DA neurons from DAT-Cre DRP1−/− cKO mouse	Depletion mito from axon, disrupted mito motility, elongated mito.	Berthet et al., 2014
		Primary hippocampal neurons from CamKII-Cre DRP1−/− cKO mouse	Disrupted distribution with increased size and intra-mito space, reduced ATP production.	Shields et al., 2015
		Optic nerve of En1-DRP1+/− cKO mouse	Elongated mito.	Gerber et al., 2017
Mitochondrial Quality Control
PTEN-induced kinase 1-PARK6 (PINK1) PD-PARK6	Mito serine - threonine protein kinase, mitochondrial quality control.	Drosophila with PINK1B9 null mutant	Diminished mito membrane potential and altered mito transport directionality.	Liu et al., 2012; Devireddy et al., 2015
		Drosophila overexpressing dPINK1	Disturbed mito motility and distribution, reduced number of mito at NMJ	Liu et al., 2012
Parkin PD-PARK2	Cytosolic E3 ubiquitin ligase, mitochondrial quality control	Drosophila with Park25 null mutant	Reduced mito flux.	Sung et al., 2016
Mitochondrial Trafficking and Anchoring
Trafficking kinesin protein 1 (TRAK1) Fatal encephalopathy, Hyperekplexia, Refractory status epilepticus.	Mitochondrial motor adaptor	Mouse hippocampal/cortical neurons with TRAK1 knockdown	Decreased mito motility	Brickley and Stephenson, 2011; van Spronsen et al., 2013
		Drosophila with milton KO	Mito depletion	Stowers et al., 2002; Glater et al., 2006
Other neurodegenerative disease-associated genes that impair axonal mitochondrial maintenance
APP Familial AD (FAD)	Multifunctional protein, produce Aβ.	Neurons from double hAPP/PS1 Tg-mouse	Impaired mito motility and altered mito distribution.	Trushina et al., 2012
		Primary neurons from hAPP mouse (Tg2576 line)	Reduced mito motility	Calkins et al., 2011
		Neurons from mutant hAPPswe/ind Tg (J20) mice (C57BL/6J)	Impaired mito motility and increased mito ROS level	Han et al., 2020; Ye et al., 2015; Lin et al., 2018
		Primary neurons from 5XFAD mouse	Disrupted mito anterograde transport and increased ROS level.	Wang et al., 2019
		Mushroom body neurons from Drosophila expressing hAβ42	Mito fragmentation with reduced size, reduced volume.	Wang and Davis, 2021
		DLM motor neurons from adult pan-neuron Aβarc flies	Reduced number of mito, increased mito size, impaired mito transport.	Zhao et al., 2010.
Tau (MAPT) FAD Frontotemporal dementia (FTD)	MT-binding protein, modulate MT assembly and stability, stabilize interaction between motors and MT.	Rat hippocampal neurons expressing cleaved tau	Impaired mito transport, reduced mito membrane potential and ATP production.	Quintanilla et al., 2020
		rTg4510 mouse expressing hTauP301L	Disturbed mito distribution.	Kopeikina et al., 2011
		Tibial nerve explants from hTauP301L knock-in mouse	Disturbed mito motility and reduced density.	Gilley et al., 2012
		Sciatic nerve from mouse expressing TauK369I	Reduced mito density and impaired anterograde transport.	Ittner et al., 2008
		hESC-derived neurons expressing pseudo-hyperphosphorylated tau	Reduced mito motility and ATP level.	Mertens et al., 2013
Presenilin (PSEN) AD	Catalytic subunit of gamma-secretase for substrates, include APP.	Neurons from hPS1M146L and double APP/PS1 Tg mice	Impaired motility and reduced velocity, altered distribution.	Trushina et al., 2012
		Hippocampal neurons from PS1M146V knock-in mouse	Reduced mito density.	Pigino et al., 2003
LRRK2 PD-PARK8	Autophagy, trafficking, mitochondrial localization.	LRRK2G2019S iPSC-derived neurons	Delayed mito arrest and mitophagy upon stress.	Hsieh et al., 2016
α-synuclein (SNCA) PD-PARK1 AND PARK4 Dementia with Lewy bodies	Presynaptic signaling and membrane trafficking.	Larval zebrafish expressing hαSyn in peripheral sensory neurons	Increased mito density, increased spherical mito with swelling, decreased mito motility with increased percentage of retrograde transport in the motile pool.	O’Donnell et al., 2014
TDP-43 (TARDBP) ALS, Frontotemporal lobar degeneration with ubiquitinated inclusions (FTLD-U).	Multifunctional, mRNA splicing, transcription regulation.	hTDP-43prp Tg-mouse	Altered distribution and morphology of mito.	Xu et al., 2010
		Sciatic nerve from Thy1.2-mitoDendra-hTDP-43A315T Tg-mouse	Reduced mito motility and reduced size, aberrant clustering of mito, mito accumulation at NMJ.	Magrane’ et al., 2013
		Rat motor neuron expressing hTDP-43Q331K/V	Altered mito morphology, reduced mito length and density, reduced mito membrane potential, increased ROS level.	Wang et al., 2013
		Mouse motoneuron with TDP-43 knockdown	Reduced mito density and membrane potential.	Briese et al., 2020
SOD1 fALS	Mitochondrial reductase converting superoxide anions to hydrogen peroxide.	Spinal cord/motor neurons/sciatic nerves/DRG neurons from hSOD1G93A Tg-mouse	Impaired mito motility and reduced mito density, mito fragmentation, increased MDVs.	Xie et al., 2015; Lin et al., 2018; Bilsland et al., 2010
		Rat cortical neurons expressing SOD1A4V/G37R/G85R	Increased inter-mito distance, reduced mito density, increased mito retrograde while reduced anterograde transport.	De Vos et al., 2007
huntingtin (HTT) Huntington’s disease (HD)	Multifunctional protein mediating long-and short-range vesicle and organelle transport.	Cortical neurons from HdhCAG140/+Tg-mouse	Mito fragmentation, increased motility, reduced mito membrane potential.	Virlogeux et al., 2018
		Primary neurons from HD72 Tg-mouse and htt KO Tg-mouse	Impaired mito motility and velocity.	Trushina et al., 2004
		Striatal neurons from HdhCAG150 knock-in mouse	Reduced mito motility and velocity.	Orr et al., 2008
		Primary neurons from BACHD Tg-mouse expressing full-length hHtt with 97 CAA and CAG (mixed repeats).	Mito fragmentation, reduced mito anterograde motility.	Shirendeb et al., 2011

Abbreviations: APP: amyloid precursor protein; CGCs: cervical ganglion cells; cyto C: cytochrome c; DJ1: protein deglycase DJ-1; DLM: dorsal longitudinal muscle; DRG: dorsal root ganglion; ESC: embryonic stem cell; h: human; IMM: inner mitochondrial membrane; iPSC: induced pluripotent stem cell; OCR: oxygen consumption rate; OMM: outer mitochondrial membrane; LRRK2: leucine-rich repeat serine/threonine-protein kinase 2; MAPT: microtubule-associated protein tau; MDVs: mitochondrial derived vesicles; Mito: mitochondria; MVBs: multivesicular bodies; NMJ: neuromuscular junction; RGCs: retina ganglion cells; SOD1:Cu/Zu superoxide dismutase-1; TDP-43: TAR DNA-binding protein-43; Tg-mouse: transgenic mouse; TRAK1: trafficking kinesin protein 1.

Somatic vs distal mitophagy

Defective PINK1/Parkin-mediated mitophagy constitutes a prominent pathology in major neurodegenerative diseases (Khalil et al., 2015; Ye et al., 2015), despite mutations in PINK1 and Parkin being initially identified in PD patients. To facilitate mitophagy in distal axons, a model for arresting damaged axonal mitochondria was proposed: upon acute mitochondrial depolarization and damage, MIRO1 is rapidly degraded via the Parkin-ubiquitination-proteasome pathway, thus motile mitochondria are arrested for local mitophagy (Wang et al., 2011; Ashrafi et al., 2014; Liu et al., 2012). Accordingly, PD-linked human iPSCs display slowed MIRO1 degradation, leading to a failure in arresting mitochondria for mitophagy (Hsieh et al., 2016). Thus, arresting motility of damaged mitochondria precedes Pink1/Parkin-mediated mitophagy within axons (Figure 3A).

Figure 3. Quality Control Models for Damaged vs Stressed Axonal Mitochondria.

(A) Arresting damaged mitochondria for mitophagy in distal axons. Upon acute mitochondrial damage, MIRO1 is rapidly degraded via the Parkin-mediated ubiquitination-proteasome pathway, thus facilitating local mitophagy by arresting damaged mobile mitochondria within axons.

(B) Remobilizing chronically stressed mitochondria from distal axons. Mature neurons, where the majority of axonal mitochondria are stationary, adopt an intrinsic mechanism releasing SNPH vesicles from chronically (mildly) stressed mitochondria in order to remobilize these anchored mitochondria from distal axons and retrogradely transport them for repair by fusion with healthy mitochondria along their trafficking route or for elimination by somatic mitophagy.

However, in mature neurons, majority of axonal mitochondria remain in the stationary status due to elevated expression of mitochondrial anchor SNPH. Therefore, a second model for axonal mitochondrial quality control has also been proposed. Under chronic and mild stress conditions, mature neurons adopt an intrinsic mechanism releasing SNPH cargo vesicles, an axon-specific type of MDVs, from stressed mitochondria to remobilize these anchored mitochondria for retrograde transport (Lin et al., 2017), either for repair by fusing with healthy mitochondria along their trafficking route or for elimination in the soma through mitophagy (Figure 3B). Interestingly, this SNPH-releasing mechanism is robustly activated in axons of spinal ventral root motor neurons during presymptomatic stages of fALS-linked SOD1^G93A mice and in the early pathological stages of AD-linked human APP (hAPP) mutant cortical neurons (Lin et al., 2017), reflecting axonal mitochondrial surveillance at early disease stages. Progressive mitochondrial damage depletes SNPH in the late stages in fALS- and AD-linked mouse brains and postmortem human AD brains. This model is supported by a study in cultured rat hippocampal neurons, where mild oxidative stress or mild depolarization primarily induced somatic but not axonal/dendritic mitophagy (Evans and Holzbaur, 2020). Consistently, in vivo imaging analysis showed that the majority of mitochondrial turnover via acidified mito-lysosomes occurs in the Purkinje neuronal somata in mito-QC mice, a transgenic mouse line expressing a pH-sensitive fluorescent mitochondrial signal (McWilliams et al., 2016). Thus, these two pathways — distal vs somatic mitophagy — may reflect different severities of mitochondrial impairment (mild or chronic stress vs acute or severe damage, respectively) or different disease progression (early vs late stages, respectively).

Acute Bioenergetic Crisis in Neural Injury

While developing neurons possess robust neurite growth, mature CNS neurons typically fail to regrow after injury due to an intrinsic/extrinsic decline in permissive conditions or an increase in nonpermissive conditions, leading to permanent neurological impairment (He and Jin, 2016). Regeneration is a complicated growth program, where injured axons reform growth cones, reorganize cytoskeleton, reassemble synapses, and rewire circuits. All these events require a high level of energy consumption mostly supplied by local mitochondria within injured axons (Smith and Gallo, 2018). Regeneration is particularly problematic after brain injury and ischemia that acutely damage mitochondria, leading to a local energy crisis and release of toxic and apoptotic factors (Sheng, 2017; Vosler et al., 2009). Recent studies have started to address two fundamental questions of whether injury-induced mitochondrial damage contributes to the energetic restriction that accounts for regeneration failure in the CNS, and whether neurons could restore axonal bioenergetics by recruiting healthy mitochondria into injured axons.

CNS injuries trigger acute energy crisis

In contrast to chronic mitochondrial stress or dysfunction during progressive neurodegeneration, brain injuries are acute insults that trigger mitochondrial damage leading to a local energy crisis (Cavallucci et al., 2014; O’Donnell et al., 2013). Both traumatic brain injury (TBI) and spinal cord injury (SCI) are initiated by primary mechanical injury that triggers secondary mitochondrial damage. In an adult mouse TBI model, mitochondrial respiratory chain dysfunction was observed as early as 30 minutes after injury and persisted up to 72 hours in the cortex and hippocampus (Singh et al., 2006). TBI damages mitochondrial structures, including swelling and crista disruption (Wiley et al., 2016), and reduces NAD⁺ levels that further inhibit glycolysis, thus accelerating local energy deficits (Jalloh et al., 2015). Similarly, in a rat SCI model, mitochondrial bioenergetics declines starting at 12 hours post-injury and persisted for at least 24 hours (Jin et al., 2004; Sullivan et al., 2007). Traumatic axonal injury, a subgroup of TBI with dominant axonal injury (Hill et al., 2016), associates with reduced mitochondrial transport, cristae disruption, and swelling or fragmentation (Court and Coleman, 2012). Axotomy in cultured superior cervical ganglion neurons induces decline of cytosolic NAD, and thus inhibits glycolysis and mitochondrial respiration; Wld^S protects axons from Wallerian degeneration by partially rescuing this energetic decline (Godzik and Coleman, 2015).

Ischemic stroke occurs due to occlusion of an artery in the brain and typically leads to cell death within the affected brain region. While impaired delivery of glucose and oxygen directly affects bioenergetics, enhanced ROS production and impaired fission/fusion or mitophagy during ischemia-reperfusion further disturb mitochondrial function (Yang et al., 2018). ATP levels remarkably fall during the first 5 min of arterial occlusion, and this energy crisis stabilizes at ~15–30% of the ATP content in the non-ischemic region for the first 2 hours of focal ischemia (Folbergrova et al., 1992; Welsh et al., 1991). Due to the limited oxygen supply, glucose within the ischemic tissue is metabolized mainly via glycolysis, resulting in a 10-fold higher lactate production than that in non-ischemic tissue. After reperfusion, ATP levels in ischemic tissue are partially recovered to 50–70% of that in control tissue within the first hour (Folbergrova et al., 1995; Welsh et al., 1991). Consistently, in mature cultured neurons, axonal mitochondrial integrity is impaired and ATP levels are declined after oxygen and glucose deprivation and reperfusion (OGD-R), leading to axonal degeneration (Huang et al., 2021). Thus, mitochondrial dysfunction and energy crisis are hallmarks of ischemic insult that induces neuronal death.

Promoting CNS regeneration by turning off mitochondrial anchoring

In mature neurons and adult mouse brains, SNPH is highly expressed (Zhou et al., 2016), which contributes to the most stationary axonal mitochondria. Thus, SNPH becomes an attractive target for remobilizing and replacing injury-damaged mitochondria with healthy ones and thus recovering axonal energy metabolism. By applying cultured cortical neurons in microfluidic devices, we recently demonstrated that axotomy rapidly depolarizes local mitochondria, leading to an energy crisis in the vicinity of injured axons (Zhou et al., 2016; Huang et al., 2021) (Figure 4A, left). In mature neurons, deleting Snph remobilizes axonal mitochondria (Figure 4A, right), restores local energy supply (Figure 4B), and facilitates regenerative capacity after injury (Figure 4C). Enhancing axonal mitochondrial transport in adult Snph^−/− mice also boosts regeneration of dorsal root ganglion (DRG) neurons after sciatic nerve crush injury (Zhou et al., 2016). A recent SCI study examined the long-projecting corticospinal tract (CST) axons following a 5^th cervical (C5) dorsal hemisection (DH) in adult Snph^−/− mice (Han et al., 2020). While WT mice showed little CST axon regrowth beyond the lesion, Snph^−/− mice exhibited robust CST axon regeneration past the lesion, extending into the caudal spinal cord (Figure 4D), where regenerated CST axons establish cortico-spinal motor connections and conduct synaptic transmission. Most promisingly, Snph^−/− mice facilitate delivery of healthy mitochondria from the motor cortex into the regenerating CST axons and display notable forelimb dexterous improvement after the C5 SCI (Han et al., 2020). Similarly, ischemic stress triggers the retrograde transport of damaged mitochondria from distal axons towards the soma for turnover through mitophagy (Zheng et al., 2019). Blocking mitochondrial transport by overexpressing SNPH impairs the removal of damaged mitochondria from axons and aggravates ischemia-induced axon injury. In contrast, reinforced mitochondrial retrograde transport promotes somatic mitophagy and thus attenuates axonal mitochondrial dysfunction from ischemic injury (Zheng et al., 2019). A recent study reported that injury-ischemic stress activates an AKT-PAK5 energy signaling axis through local axonal PAK5 translation, which is normally suppressed after postnatal development. Activated mitochondria-targeted PAK5 induces intra-axonal phosphorylation of SNPH, an event required for turning off SNPH-anchoring switch, thus remobilizing and replacing damaged mitochondria with healthy ones, leading to the recovery from injury-induced axonal energy crisis (Huang et al., 2021) (Figure 4E, left). Such a local PAK5 signaling pathway enhances axonal mitochondrial transport by 2.13-fold as early as 10 minutes post injury and recovers mitochondrial membrane potential within injured axons 4 hours after ischemia-reperfusion. This study suggests that mature CNS neurons adopt an “energy repair program” by replacing damaged mitochondria with healthy ones. Thus, reprogramming the AKT-PAK5-SNPH signaling axis represents a potential therapeutic target to support neuron survival and regeneration after injury-ischemia.

Figure 4. Promoting CNS Regeneration by Enhancing Mitochondrial Transport.

(A-C) Regeneration requires a high level of energy demand that is mainly supplied by local mitochondria within injured axons. In mature neurons, the majority of axonal mitochondria are anchored by SNPH. Brain injuries damage mitochondria, leading to a local energy crisis that contributes to CNS regeneration failure. Remobilizing mitochondria by deleting SNPH helps remove damaged mitochondria from injured axons and replenishment with healthy ones to restore local energy supply (A, B), thus robustly facilitating axon regeneration (C). WT and Snph KO neurons (B) were infected with lentivirus encoding ATP probe GO-ATeam2, followed by imaging of axon tips 6 hours after axon injury. Heatmap index (C) indicates ATP levels within axon tips. WT and Snph KO neurons in microfluidic chamber devices (C) were axotomized at DIV12, followed by βIII-tubulin labeling at DIV18. Scale bars, 10 μm (B) and 100 μm (C).

(D) Snph^−/− mice display enhanced axonal regeneration after spinal cord injury (SCI). Each color represents a trace of BDA-labelled CST axons caudal to the lesion in a single section. Reconstruction of a stack of all 40 color-coded images shows CST axon regeneration across and beyond the lesion (labelled by a red arrow) in Snph^−/− mice at 8 weeks post SCI. Scale bars, 500 μm. Adapted with permission from Han et al. (2020).

(E) Signaling pathways promote CNS regeneration by targeting mitochondrial trafficking and anchoring. Left: Targeting SNPH. Injury-ischemia activates local axonal AKT-PAK5 signaling that remobilizes damaged mitochondria by turning off the anchoring switch through phosphorylation of SNPH, thus replacing damaged mitochondria with healthy ones, and accelerating axonal energy recovery to support CNS regeneration (Huang et al., 2021). Right: Targeting motor-adaptor. The cAMP-PKA-kinesin-1 axis enhances mitochondrial flux into axons and suppresses an aging-dependent decline in mitochondrial motility by upregulating kinesin-1 (Vagnoni and Bullock, 2018) and enhances axon regrowth (Xu et al., 2017). Mitochondrial protein ARMCX1 promotes axonal mitochondrial transport by regulating MIRO1. ARMCX1-enhanced axonal mitochondrial transport is critical to support RGC survival and axon regeneration after injury (Cartoni et al., 2016). Injury-responsive DLK-1 signaling helps recruit mitochondria to injured axons of GABA motor neurons in C. elegans to support axon regeneration (Han et al., 2016).

Promoting CNS regeneration by targeting mitochondrial transport motor-adaptor

Signaling pathways targeting motor adaptors have been reported to facilitate mitochondrial transport and axonal regeneration. MIRO1 is another target in promoting axon regeneration. Overexpression of MIRO1 in mature cortical neurons enhances axon regrowth, while depleting MIRO1 abolishes axonal regeneration (Huang et al., 2021; Zhou et al., 2016). Mitochondrial protein ARMCX1 promotes axonal mitochondrial transport by regulating MIRO1 in RGCs of Socs3 and Pten double KO mice, where ARMCX1 expression is elevated, and regenerative capacity is enhanced. Interestingly, ARMCX1-enhanced axonal mitochondrial transport supports RGC survival and facilitates axon regeneration after injury (Cartoni et al., 2016) (Figure 4E, right). Consistently, histone deacetylase 6 (HDAC6) decreases mitochondrial transport in mouse DRG neurons by deacetylation of MIRO1; this modification inhibits DRG neuron axon growth by blocking mitochondrial transport (Kalinski et al., 2019). In addition, the mitochondrial protein HUMMR is induced by hypoxia-inducible factor 1a (HIF-1a) and enhances mitochondrial transport into axons in response to hypoxia by targeting MIRO1/2 (Li et al., 2009). Several lines of evidence support the notion that accelerated delivery of mitochondria to injured axons helps meet the increased energy demand needed for axon survival. In the C. elegans ric-7 mutant where mitochondrial transport is impaired, delivering mitochondria into axons helps reduce degeneration of injured axons (Rawson et al., 2014). Examining the proximal segments of transected intercostal nerves revealed injury-induced mitochondrial re-distribution to growth cones (Misgeld et al., 2007). Maintaining energy supply is critical for axonal protection after injury through Wld^S (Shen et al., 2013); inducing mitochondrial flux into injured axons protects axons from Wallerian degeneration (Avery et al., 2012). The cAMP-PKA-kinesin-1 axis enhances mitochondrial flux into axons in aging Drosophila wing neurons where PKA and kinesin-1 decline with aging; activating cAMP/PKA signaling suppresses the aging-related decline in mitochondrial motility by upregulating kinesin-1 (Vagnoni and Bullock, 2018) (Figure 4E, middle). Consistently, activating cAMP enhances mitochondrial trafficking and axon regrowth in the zebrafish CNS (Xu et al., 2017). Mitochondria play critical roles in axon regeneration of mechanosensory neurons of C. elegans (Knowlton et al., 2017). Axon injury stress in C. elegans activates DLK-1 signaling, which helps recruit mitochondria to injured axons of GABA motor neurons to facilitate regeneration (Han et al., 2016) (Figure 4E, right). Altogether, these studies support the emerging concept that facilitating CNS regeneration requires reprograming of mitochondrial transport (Sheng 2017). A high-content screen identified compounds to enhance axonal mitochondrial motility (Shlevkov et al., 2019), providing therapeutic potential for “reprograming” axonal mitochondrial transport in injured nerve systems.

Perspectives: Restoration of Axonal Bioenergetics

Since axonal energy deficits occur during chronic neurodegenerative diseases and after acute injuries and ischemia, energy restoring programs have been considered emerging strategies for alleviating disease progression and supporting neuron survival and regeneration. As we described above, axonal mitochondrial trafficking/anchoring, biogenesis, and quality control are critical for maintaining a healthy pool of mitochondria within axons, thus serving as potentially important therapeutic targets for restoration of axonal bioenergetics. Here, we also provide perspectives on recent developments of the following energy restoring strategies.

Mitochondrial transfer and transplantation

Cell-to-cell mitochondrial transfer has surfaced as a potential avenue for therapeutic development in brain injury and stroke. Intercellular mitochondria transfer is achieved via extracellular vesicles or tunneling nanotubes (TNTs), thus conferring healthy mitochondria to recipient cells with energy deficits (Figure 5A). It was reported in the mouse brain following transient focal cerebral ischemia that healthy mitochondria were released from astrocytes into the extracellular space and mitochondrial particles were then transferred into energetically stressed neurons to support neuronal viability (Hayakawa et al., 2016). Increased extracellular mitochondrial particles were also found in the cerebrospinal fluid after subarachnoid hemorrhage in rat models and human patients (Chou et al., 2017). Neural stem cells rescue mitochondrial dysfunction in recipient cells by transferring mitochondria (Peruzzotti-Jametti et al., 2021). TNTs facilitate mitochondrial transfer between microglia both in vitro and in vivo (Scheiblich et al., 2021).

Figure 5. Energy Repair Strategies to Restore Axon Bioenergetics.

(A) Mitochondrial transfer via extracellular vesicles or TNTs confers healthy mitochondria from donor cells to recipient neurons (or axons) with energy deficits. Mitochondrial transfer has surfaced as a potential avenue for therapeutic development in brain injury and stroke.

(B) Mitochondrial transplantation is the delivery of isolated healthy mitochondria via intracerebral or intraarterial injection to rescue neuronal mitochondrial dysfunction and has been applied in pre-clinical studies of acute CNS injuries and neurodegenerative disorders. Mechanisms underlying the internalization of isolated (naked) mitochondria by recipient neurons are largely unknown.

(C) Boosting axon bioenergetics via glia-neuron crosstalk or bioenergetic compounds. Left: Astrocytes export lactate to axons where it is converted to pyruvate for fueling mitochondrial bioenergetics. Middle: Oligodendrocytes release exosomes that shuttle SIRT2 into axons, where it boosts energy metabolism by deacetylation of mitochondrial proteins ANT1/2. Right: Creatine (Cr) is enriched in the brain and converted into phosphocreatine (PCr) by creatine kinase for rapid converting ADP to ATP independent of the mitochondrial electron transport chain. Creatine administration has been used in clinical trials on TBI, SCI, and stroke.

Mitochondrial transplantation is the delivery of isolated mitochondria via intracerebral or intraarterial injection (Norat et al., 2020) (Figure 5B) and has been shown to rescue mitochondrial respiratory activity in ischemic rat brains (Huang et al., 2016), as well as to recover mitochondrial integrity in injured hippocampal neurons and increase neurite regeneration in an in vitro TBI model (Chien et al., 2018). Similarly, transplantation of mitochondria into injured spinal cords preserves bioenergetics and improves locomotor functional recovery after SCI (Gollihue et al., 2018; Li et al., 2019). While macropinocytosis has been suggested in the engulfment of exogenous mitochondria on the cell surface (Kesner et al., 2016.), mechanisms regulating mitochondrial entry into recipient cells remain largely unknown. How mitochondria behave in recipient cells and their long-term bioenergetic benefits need to be thoroughly evaluated. It is also challenging to retain bioenergetic activity of isolated (naked) mitochondria before injection, trace mitochondria after injection, and improve efficacy and safety of mitochondrial transplantation (Lightowlers et al., 2020).

Glia-axon bioenergetic crosstalk

Considering the intricate networks in the human brain where billions of neurons and glial cells wire together (Allen and Lyons, 2018), a comprehensive maintenance of axonal bioenergetic status must include contributions from glial cells. For example, an astrocyte-neuron lactate shuttle (ANLS) hypothesis proposes that astrocytes export lactate to neurons where it is converted to pyruvate to fuel mitochondrial OxPhos (Pellerin and Magistretti, 1994) (Figure 5C, left). Oligodendrocytes serve as myelinating cells surrounding axons of the CNS; this unique structure ideally positions oligodendrocytes to support axonal energy metabolism by delivering energy substrates to axons of neurons when they are deprived of such substrates (Funfschilling et al., 2012; Lee et al., 2012; Meyer et al., 2018; Saab et al., 2016). We recently revealed a transcellular signaling pathway through which oligodendrocyte-derived exosomes containing NAD-dependent deacetylase sirtuin 2 (SIRT2) boosts axonal energy metabolism by deacetylation of mitochondrial adenine nucleotide translocases 1 and 2 (ANT1/2) (Chamberlain et al., 2021) (Figure 5C, middle). SIRT2 is undetectable in neurons but highly expressed in mature oligodendrocytes where they are released via exosomes and internalized into axons. Deletion of Sirt2 in mice abolishes the oligodendrocyte-to-axon energetic crosstalk, suggesting this transcellular signaling as an attractive therapeutic target for reversing axonal energy deficits in major neurological disorders.

Bioenergetic compounds

Using blood-brain-barrier permeable bioenergetic compounds represents an alternative therapeutic direction. Creatine is one such therapeutic metabolite used in CNS injuries in rodent models (Tarnopolsky and Beal, 2001). Creatine is converted into phosphocreatine (PCr) by creatine kinase for rapid converting ADP to ATP, independent of mitochondria (Wyss and Kaddurah-Daouk, 2000) (Figure 5C, right). In a model of hypoxia/ischemia, application of creatine alleviates axonal ATP loss and reduces axon damage (Shen and Goldberg, 2012). Consistently, after focal ischemia by middle cerebral artery occlusion, both ATP and PCr concentrations decline following decreased blood flow (Obrenovitch et al., 1988). Feeding creatine increases ATP and PCr levels in the mouse brainstem 30 min after anoxia treatment (Wilken et al., 2000). Creatine administration ameliorates TBI-induced cortical damage in mice and rats by maintaining bioenergetics (Sullivan et al., 2000) and promotes locomotor capacity in rat SCI models (Hausmann et al., 2002). Consistently, creatine treatment promotes axon regeneration in a mouse SCI model (Han et al., 2020). Creatine administration improves the upper extremity work capacity in patients with cervical-level SCI (Jacobs et al., 2002), and reduces traumatic headache, dizziness, and fatigue in TBI patients (Sakellaris et al., 2008). Future development of more effective bioenergetic compounds is a promising therapeutic direction to reverse energy crisis.

Abbreviations List

AD

Alzheimer’s disease

ALS

Amyotrophic Lateral Sclerosis

AMPK

AMP-activated protein kinase

ANT1/2

adenine nucleotide translocases 1 and 2

APs

action potentials

ATP

adenosine triphosphate

CMT2A

Charcot-Marie-Tooth type 2A syndrome

CNS

central nervous system

CST

corticospinal tract

DA

dopaminergic

DH

dorsal hemisection

DRG

dorsal root ganglion

DRP1

dynamin-related protein 1

ER

endoplasmic reticulum

FIS1

fission protein 1

FRET

fluorescence resonance energy transfer

HDAC6

histone deacetylase 6

KIF5

kinesin-1 family

MEFs

mouse embryonic fibroblasts

MFF

mitochondrial fission factor

MFN1/2

mitofusin 1 and 2

MiD49 and MiD51

mitochondrial dynamics proteins of 49 and 51 kDa

MIM

mitochondrial inner membrane

MIRO1/2

Mitochondrial Rho GTPase 1/2

Mito

mitochondria

MT

microtubule

NMJ

neuromuscular junction

PD

Parkinson’s disease

PNS

peripheral nervous system

OGD-R

oxygen and glucose deprivation and reperfusion

OMM

outer mitochondrial membrane

OPA1

optic atrophy 1

PAK

p21-activated kinase

PCr

phosphocreatine

RGCs

retinal ganglion cells

ROS

reactive oxygen species

SCI

spinal cord injury

SIRT2

sirtuin 2

SV

synaptic vesicle

TBI

traumatic brain injury

TNTs

tunneling nanotubes