Novel insights into the mechanism of reactive oxygen species-mediated neurodegeneration

Shuji Wakatsuki, Toshiyuki Araki

Abstract Neurite degeneration, a major component of many neurodegenerative diseases, such as Parkinson’s disease, Alzheimer’s disease, and amyotrophic lateral sclerosis, is not part of the typical apoptosis signaling mechanism, but rather it appears that a self-destructive process is in action.Oxidative stress is a well-known inducer of neurodegenerative pathways: neuronal cell death and neurite degeneration.Although oxidative stress exerts cytotoxic effects leading to neuronal loss, the pathogenic mechanisms and precise signaling pathways by which oxidative stress causes neurite degeneration have remained entirely unknown.We previously reported that reactive oxygen species generated by NADPH oxidases induce activation of the E3 ubiquitin ligase ZNRF1 in neurons, which promotes neurite degeneration.In this process, the phosphorylation of an NADPH oxidase subunit p47-phox at the 345th serine residue serves as an important checkpoint to initiate the ZNRF1-dependent neurite degeneration.Evidence provides new insights into the mechanism of reactive oxygen species-mediated neurodegeneration.In this review, we focus specifically on reactive oxygen species-induced neurite degeneration by highlighting a phosphorylation-dependent regulation of the molecular interaction between ZNRF1 and the NADPH oxidase complex.

Key Words: neurite degeneration; oxidative stress; phosphorylation; reactive oxygen species;ubiquitin ligase

Introduction

Neurons, the main component of our brain, have long processes called axons and dendrites.By communicating with other cells through these neurites,they form a neural network that is the basis of all brain function.In many neurologic disorders, including Parkinson’s disease, Alzheimer’s disease, and amyotrophic lateral sclerosis, neurites are gradually lost and neurons die causing the brain and spinal cord to malfunction (Gan et al., 2018; Soto and Pritzkow, 2018).A study has shown that cell death often requires activation of an intracellular reaction that actively kills cells (Moujalled et al., 2021).On the other hand, it has been known that neurite degeneration which precedes neuronal cell death is controlled by a mechanism other than cell death, but the details remain unknown (DiAntonio, 2019; Coleman and Hӧke, 2020).

Neurite degeneration observed in the area of neurites distal to the site of physical injury is known as Wallerian degeneration, which is a prototypical neurite degeneration (Waller, 1851).It has been demonstrated that Wallerian degeneration is not a passive phenomenon caused by the disruption of transport from the cell body due to damaged neurites, resulting in depletion of nutrients and materials; but rather an active one that requires enzymatic reactions within the neurites (Coleman and Hӧke, 2020; Krauss et al., 2020).For example, in neurons lacking the NAD-degrading enzyme sterile-α and armadillo-motif-containing protein 1 (SARM1), Wallerian degradation is markedly delayed in a variety of animal models, including mammals and insects (Figley and DiAntonio, 2020; Krauss et al., 2020).Thus, it has become clear that Wallerian degeneration is robustly regulated by self-destructive reactions, whereas it is still unknown how physical damage to neurites translates into destructive reactions.

The human brain consumes about 20% of the body’s energy (Watts et al.,2018).Oxygen metabolism produces reactive oxygen species (ROS) as a byproduct.When environmental stressors exacerbate ROS generation or detoxification mechanisms fail to remove excess ROS, ROS serves as oxidative stress and damages cells (Shields et al., 2021).Neurons have high-energetic activities that pose significant challenges in detoxifying ROS, particularly in highly specialized cell compartments such as axons and dendrites.Oxidative stress has long been associated with neurologic disorders.For example,mutations in known Parkinson’s disease-related genes such as PINK1, PARK2,SNCA, and LRRK2 are thought to impair mitochondrial function, resulting in increased ROS levels and vulnerability to oxidative stress (Dias et al., 2013;Kolodkin et al., 2020).However, the pathogenic mechanisms and precise signaling pathways by which oxidative stress causes neurite degeneration are unknown.

In this review, we focus specifically on ROS-induced neurite degeneration,highlighting a phosphorylation-dependent regulation of molecular interaction between the E3 ubiquitin ligase zinc and ring finger 1 (ZNRF1) and nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (Wakatsuki et al., 2015, 2022; Figure 1).We then discuss the putative target of the ROS-induced neurite degeneration as a therapeutic strategy for treating neurologic diseases.

Search Strategy and Selection Criteria

The references cited in this review have been obtained from the following databases for studies regarding the roles of ROS in neurodegeneration:PubMed, Google Scholar, and Science Direct.No restriction on publication dates was applied for search strategy.

Neuritic Self-Destructive Pathways

During development, excess or unnecessary axons, dendrites, and synapses are removed to establish a mature neural network.This process of selective removal of such structures, called neurite pruning, occurs without the death of the neuron that owns them (Riccomagno and Kolodkin, 2015).Genetic and biochemical studies in several models provide evidence that caspase activity is required for this developmental pruning (Unsain and Barker, 2015; Dehkordi et al., 2022; Figure 2).The involvement of death receptor 6 has also been shown in mammalian neurite pruning (Nikolaev et al., 2009).

In a series of genetic screens in flies and mice, mitogen activated protein kinase (MAPK) and its upstream kinase, dual leucine zipper kinase (DLK),are identified to be required for axonal degeneration (Miller et al., 2009).Mutant mice lacking DLK show a significant delay in degeneration of the distal axon of the severed sciatic nerve compared with wild-type animals.DLK acts through another kinase c-Jun N-terminus kinase (JNK).Inhibition of this pathway delays axonal fragmentation induced by a microtubule-destabilizing drug vincristine.These results suggest that the DLK/JNK pathway functions in a variety of injury-induced axonal self-destruction.In addition to the DLK/JNK pathway, several kinases that promote axonal degeneration have recently been identified (Yang et al., 2015; Figure 2).

Figure 1｜Model of the current hypothesis.

Figure 2 ｜ Signaling pathways of neuritic self-destruction.

Accumulating evidence suggests the significant role of SARM1 in promotion of axonal degeneration.SARM1 was identified by loss-of-function gene screening for axonal protection after injury.SARM1 is a highly conserved protein consisting of an N-terminal Armadillo/HEAT motif (ARM) domain, two sterile α motif (SAM) domains, and a C-terminal Toll/interleukin-1 receptor domain, each of which is involved in autoinhibition, oligomerization, and NAD hydrolysis, respectively (Funakoshi and Araki, 2021; Sambashivan and Freeman, 2021).In steady state neurons, SARM1 is kept inactive by the autoinhibitory ARM domain, but when this autoinhibition is abolished by injury, the C-terminal Toll/interleukin-1 receptor domain self-associates and hydrolyzes NAD.The finding that SARM1 is an autoregulatory NADase strongly suggests that disruption of NAD-related metabolism in axons is closely associated with degeneration (Figure 2), but the mechanism by which axonal damage activates SARM1 is not yet fully understood.

Recently, mechanistic links between the essential molecules of the axonal degeneration program have been emerging (Summers et al., 2020), and these components may converge into a single pathway (Figure 2).However,the relationship between this program and oxidative stress remains elusive.Before we discuss this in detail, we first summarize the involvement of oxidative stress in neurodegeneration as well as in neurite degeneration.

Reactive Oxygen Species-Mediated Regulation of Neurite Degeneration

High levels of oxidative stress are commonly observed in the brains of patients with neurodegenerative conditions (Zuo et al., 2015).Many experimental studies have been conducted to elucidate the significance of oxidative stress in the pathophysiology of neurodegenerative diseases (Yeung et al., 2021).ROS is endogenously generated mostly by mitochondria and NADPH oxidase (Shields et al., 2021).ROS may not be sufficient to induce neurodegenerative diseases, but it exacerbates disease progression through oxidative damage to proteins by causing molecular cross-linking that results in misfolding and destructuring.Interestingly, neurons have been identified as vulnerable to oxidative damage due to their high oxygen consumption,high poly-unsaturated fatty acid content in membranes, and weak antioxidant defenses (Watts et al., 2018; Kolodkin et al., 2020).Accumulation of misfolded proteins is part of the pathogenesis of several neurodegenerative diseases.Aggregation of these misfolded or modified proteins, on the other hand,triggers an inflammatory response in the brain, leading to mitochondrial dysfunction with concomitant abnormal ROS generation, which is strongly associated with this pathophysiological mechanism (Soto and Pritzkow, 2018).For example, accumulation of mutant huntingtin in Huntington’s disease interacts directly with mitochondria, causing a decrease in energy supply and increased ROS generation, ultimately leading to neuronal cell death (Ross and Tabrizi, 2011).Using an

in vivo

model, Musgrove et al.(2019) showed that overexpression of α-synuclein induced oxidative stress in cholinergic neurons forming the dorsal motor nucleus of the vagus nerve.In this process, modified forms of α-synuclein, such as α-synuclein oligomers, bind with high affinity to the TOM20 receptor in the outer mitochondrial membrane and inhibit protein uptake into the mitochondria, thereby impairing mitochondrial respiration.This triggers the accumulation of ROS, leading to neurodegeneration, i.e.,neuronal cell death and axonal degeneration.Thus, it seems obvious that ROS generated from axonal mitochondria induces axonal degeneration.

There are not many reports in which the causal relationship between ROS generation and axonal degeneration has been rigorously examined.Geden et al.(2021) solved this problem by using a microfluidic chamber in which axons and neuronal bodies can be spatially and fluidly separated and cultured.Persistent nuclear DNA damage by chemotherapeutic reagents activates p53 dependent apoptosis.These reagents also damage mitochondrial DNA(mtDNA).They showed that selective exposure of a DNA-damaging reagent cisplatin only to the axonal compartment of the microfluidic chamber increases axonal ROS production and axonal degeneration without causing apoptosis by damaging nuclear DNA.Interestingly, this mtDNA damageinduced axonal degeneration could not be prevented by loss of p53 or by suppression of any of the axonal degeneration pathways previously described.These results suggest that mtDNA damage induces axonal degeneration through a response different from the one by nuclear DNA damage and that ROS may play a key role in the progression of degeneration (Geden et al.,2021).However, the details of how ROS generated in axons promotes axonal degeneration remain unclear.We previously reported that ROS generated by NADPH oxidases induces activation of the E3 ubiquitin ligase ZNRF1 in neurons, which promotes neurite degeneration.We first summarize the mechanism of ZNRF1-dependent neurite degeneration.

ZNRF1: A Key Molecule in the Regulation of Neurite Degeneration

We have previously developed an “

in vitro

Wallerian degeneration model”,in which radially extended neurites from primary cultured murine dorsal root ganglia neurons are injured to analyze the progression of injury-induced degeneration (Araki et al., 2004; Wakatsuki et al., 2011).Not only is this model simple and reproducible, but the morphological features of neurites observed during degeneration are very similar to those caused by toxic stimuli or in neurodegenerative diseases.Using this model, we analyzed the intracellular signaling mechanisms that regulate neurite degeneration in a variety of pathophysiological situations and found that the E3 ubiquitin ligase ZNRF1 promotes Wallerian degradation by targeting AKT, which is degraded through the ubiquitin proteasome system (Wakatsuki et al., 2011).AKT loss leads to glycogen synthetase kinase 3B activation, whereby it phosphorylates and inactivates the microtubule-binding protein collapsin response mediator protein 2 (CRMP2).After this series of reactions, CRMP2 is degraded and thereby cytoskeletal integrity is lost, which promotes neurite degeneration.Thus, ZNRF1 is a key molecule in the regulation of Wallerian degeneration.

Activation of ZNRF1 by Oxidative Stress

Phosphorylated CRMP2 is often observed in the neurons of animal models and patients with brain ischemia, as well as in other neurodegenerative diseases (Ryan and Pimplikar, 2005; Cole et al., 2007; Hou et al., 2009;Williamson et al., 2011).Our results showing ZNRF1-dependent AKT degradation leading to CRMP2 phosphorylation suggest that CRMP2 phosphorylation might be an indicator for the activation of ZNRF1-mediated signaling in neurons.We employed a focal cerebral ischemia model to examine whether the ZNRF1-mediated signaling might be activated in neurons under oxidative stress.Focal ischemia is known to cause different types of cell death.Neurons in the ischemic core undergo necrotic cell death,whereas neurons in the ischemic penumbra surrounding the ischemic core mostly show delayed neuronal apoptosis, with oxidative stress being strongly implicated in the latter (Ueda and Fujita, 2004; Broughton et al., 2009).Using this model, we found increased immunoreactivity of phosphorylated CRMP2 in neurons within the ischemic penumbra (Wakatsuki et al., 2015).Thus, theZNRF1-dependent signaling pathway seems to be involved in the regulation of oxidative stress-induced neurodegeneration.

ZNRF1, which is constitutively expressed but inert in native neurites, becomes activated to initiate degeneration in response to neurite injury.In our

in vitro

Wallerian degeneration model, a significant increase in ROS generation was detected in neurites 3 hours after injury.From these, we hypothesized that ROS might modify ZNRF1 function.Based on this hypothesis, we found that ROS generated by NADPH oxidase in response to an injury activates ZNRF1,thereby inducing intracellular signaling that ultimately leads to neurite degeneration.Through this process, ROS activated the ubiquitin ligase activity of ZNRF1 by inducing phosphorylation of its tyrosine residue at position 103 (Y103) via epidermal growth factor receptor (EGFR), leading AKT to proteasomal degradation (Wakatsuki et al., 2015; Wakatsuki and Araki, 2016).This series of reactions is also observed in

in vivo

neurodegenerative models,including cerebral ischemia and optic nerve injury.Thus, the NADPH oxidasemediated activation of ZNRF1 is a pathophysiologically important event in the early stages of neurodegeneration.Although the detailed molecular mechanisms underlying this process remain unclear, we have recently shown that a component of NADPH oxidase complex p47-phox/neutrophil cytosolic factor 1 may interact with and regulate ZNRF1 activity (Wakatsuki et al.,2022).Before we discuss this in detail, we will review the mechanism of NADPH oxidase activation.

Regulation of NADPH Oxidase Activity

The enzymatic activity of NADPH oxidase is known to be initiated by interactions with its components (Bedard and Krause, 2007; Lassègue and Griendling, 2010).In the classic NADPH oxidase reaction, a phagocytic oxidase catalytic core NOX2, which exists as a heterodimer with p22-phox, binds to cytosolic regulatory subunits including p47-phox, p67-phox, p40-phox, and Rac1 GTPase in response to an activation stimulus.Without such stimulation,these subunits are in different intracellular compartments: NOX2 is in the membrane, while the other subunits are in the cytosol, and NADPH oxidase remains leaving NADPH oxidase disjointed and inactive.Phosphorylation of p47-phox (hereinafter referred to as p47) is essential for coupling cytoplasmic subunits to assemble and activate NADPH oxidase.p47 contains a tandem Src-homology 3 domain that is required for binding to p22.In resting cells,these Src-homology 3 domains of p47, which are fully bound to a selfrepressing region, prevent binding to p22.Phosphorylation of several key amino acid residues from serine 303 (S303) to S379 in the carboxy-terminal region of p47 by protein kinases such as MAPK and protein kinase C inhibit this intramolecular interaction, thereby relocating p47 to the membrane to activate oxidase (Rastogi et al., 2016).Thus, phosphorylation of p47 is critical in regulating NADPH oxidase activity.

Regulation of Molecular Interaction between ZNRF1 and p47 through Phosphorylation by p38 Mitogen Activated Protein Kinase

It is also known that NADPH oxidase interacts with a variety of intracellular signaling molecules, thereby mutually regulating enzyme activity (Bedard and Krause, 2007; Lassègue and Griendling, 2010).NADPH oxidases are expressed in the peripheral nervous system, and their overexpression is associated with detrimental effects on nerve function and contributes to peripheral neuropathies (Eid et al., 2022).To investigate the detailed regulation mechanism of ZNRF1 activity by NADPH oxidase, we searched for the catalytic subunit of NADPH oxidase (NOX) that binds to ZNRF1 in injured neurites using an

in vitro

Wallerian degeneration model (Wakatsuki et al.,2022).Six genes were identified in rodents: NOX1 through 4 and dual oxidase 1 and 2.We previously reported that down-regulation of NOX2, 3, 4 and dual oxidase 2 along with decreased ZNRF1 phosphorylation levels significantly inhibit neurite degeneration (Wakatsuki et al., 2015; Wakatsuki and Araki,2016).With pull-down experiments, we examined the relationship between these NOX proteins and ZNRF1 in neurites before and after injury revealing that ZNRF1 may interact with NOX2, NOX3, and p47.As mentioned above,serine phosphorylation from S303 to S379 in the carboxy-terminal region of p47 is an essential part of NADPH oxidase activation.We found that p38 MAPK phosphorylates S345 in a nerve injury-dependent manner and that the phosphorylated p47 interacts with ZNRF1.Interestingly, phosphorylated p47 at S345 (p47 pS345) could not associate with the phosphorylated p47 at Y103(p47 pY103) but could associate with the unphosphorylated form of ZNRF1 Y103F (Figure 1).

Next, we determined the region of ZNRF1 required for molecular interaction with p47 pS345 and found that ZNRF1 binds directly to p47 via an amino acid region from 11 to 144.Since Y103 is located in this region of ZNRF1, and p47 pS345 can bind to ZNRF1 only before but not after Y103 phosphorylation,we hypothesized that phosphorylation at Y103 might induce its dissociation from p47.As hypothesized, phospho-mimetic p47 S345D can interact with ZNRF1 WT in naive neurites, and this interaction was not maintained by Y103 phosphorylation.Furthermore, p47 S345D can interact with ZNRF1 Y103F in naive and even in injured nerites.Taken together, these findings suggest that p47 phosphorylation at S345 in response to nerve injury might initiate the direct interaction with ZNRF1, and that ZNRF1 phosphorylation at Y103 might lead to their dissociation (Figure 1; Wakatsuki et al., 2022).

Phosphorylation of p47 at S345: an Initial Checkpoint for Reactive Oxygen Species-Induced Neurite Degeneration

Does the transient interaction of p47 pS345 with ZNRF1 contribute to the induction of ROS generation and the activation of ZNRF1? To answer this,we examined whether overexpression of p47 and its mutants in cultured dorsal root ganglia neurons might affect ROS generation and neurite integrity.Overexpression of p47 WT or S345D did not affect ROS levels in neurites and neurite integrity.Overexpression of a constitutively active form of p47(p47 S303, 304, 328D, or p47 3SD) (Roepstorff et al., 2008; De Virgiliis et al.,2020) weakly increased ROS levels (the level was much weaker than that seen in injured neurites) but did not induce neurite degeneration.On the other hand, overexpression of a dominant negative form of p47 (p47 S303, 304,and 328A, or p47 3SA) (Roepstorff et al., 2008) decreased ROS generation and suppressed neurite degeneration.Interestingly, overexpression of p47 S345A significantly suppressed ZNRF1 phosphorylation by EGFR and AKT ubiquitination, thereby protecting neurites from injury-induced degeneration.Detailed examination with pull-down experiments revealed that in injured neurites, p47 S345A can associate with NOX2, but not with ZNRF1; whereas p47 3SA can bind to ZNRF1, but not with NOX2.Thus, phosphorylation of S345 is considered critical for p47 to bind and activate ZNRF1, whereas phosphorylation of S303, 304, and 328 is considered necessary for its association with NOX2 to activate NADPH oxidase.What about the association with EGFR? In injured neurites, EGFR can bind p47 S345A, but not p47 3SA.These results suggested that p47 S345A can associate with the NADPH oxidase complex and bind to EGFR; whereas the NADPH oxidase complex with p47 S345A cannot associate with and activate ZNRF1 (Figure 2).Thus, p47 S345A could behave like a “dominant-negative” that inhibits ZNRF1 activation.Our findings imply that p47 phosphorylation at S345 may serve as an essential checkpoint to initiate ROS-mediated neurite degeneration (Wakatsuki et al.,2022).

Conclusion and Future Directions

There are two phosphorylation-dependent regulations in p47 activation:one through protein kinase C phosphorylation at S303, 304, 328 in ROS generation, and the other through MAPK phosphorylation at S345 in ZNRF1 activation.With the finding that p38 MAPK-dependent phosphorylation of p47 at S345 is required to initiate the ZNRF1-dependent neurite degeneration signal, it is conceivable that pS345 is a priming step to initiate this pathway.Neurite injury induces: (1) protein kinase C-dependent phosphorylation of p47 to form a functional NADPH oxidase complex to produce ROS, and (2)p38 MAPK-dependent phosphorylation of p47 to associate with and activate ZNRF1.Activation of ZNRF1 by ROS is involved in both neuronal cell death and neurite degeneration in

in vivo

neurodegenerative models.From these, we consider that pharmacological or genetic interventions that target each step of ZNRF1 activation, i.e.phosphorylation by EGFR and/or its interaction with p47 pS345, may lead to improvement of disease symptoms and suppression of disease progression.

Dang et al.(2006) reported that in neutrophils, proinflammatory cytokines such as granulocyte-macrophage colony-stimulating factor and tumor necrosis factor induce phosphorylation of p47 at S345 and regulate oxidative bursts in the “priming” process.Extracellular signal-regulated kinases (ERKs) are the protein kinases involved in granulocyte-macrophage colony-stimulating factorinduced phosphorylation of S345, while p38 MAPK controls tumor necrosis factor-α-induced phosphorylation of the same site.Neutrophils engulf invading foreign microorganisms, generate ROS, and degranulate to release various enzymes.This process is the front line of defense against microbial invasion, and the ability to generate ROS is maximized through prior exposure to microbes or endogenous substances (Vogt et al., 2019).Thus, S345 is a point of convergence used by different MAPK activities to induce priming of ROS generation.The molecular mechanism to activate p38 MAPK by nerve injury is still unknown.Detailed analysis of how a variety of degenerating stimuli could induce S345 phosphorylation may lead to a potential answer to this question.

Yang et al.(2015) showed that the mitogen-activated protein kinase kinase 4 pathway plays an important role in intracellular signaling downstream of SARM1.The activation of mitogen-activated protein kinase kinase 4 is reportedly observed within 1-5 minutes after nerve injury (Yang et al., 2015).Thus, nerve injury might elicit acute signals to activate mitogen-activated protein kinase kinase 4.The pathways activated over a similar time frame are also required to control activation of p38 MAPK.The activation of ZNRF1 by increased oxidative stress in neurites is initiated within a few hours of injury and lasts for more than one day (Wakatsuki et al., 2011; Araki and Wakatsuki,2019).The SARM1-MAPK pathway elicits energy deficits in injured neurites,whereas signaling initiated by the activation of ZNRF1 destroys neurites by destabilizing the cytoskeletal structure.Multiple signaling pathways are required to regulate the progression of neurite degeneration, presumably because multiple different cellular events need to be targeted within differenttime frames for efficient progression of neurite degeneration.

Under normal conditions, ROS regulates redox homeostasis and functions as an important messenger in cellular signaling (Lennicke and Cochemé, 2021).In contrast to our findings, beneficial functions of ROS in the process of neurite regeneration have been reported.Hervera et al.(2018) showed that hydrogen peroxide might induce expression of regeneration process-relatedgenes after injury and promote regeneration of injured axons.Loss of function of phosphatase and tensin homolog deleted from chromosome 10 (PTEN) is known to promote axonal regeneration in the mammalian central nervous system.ROS generated by axonal NOX2 oxidizes PTEN in an injury-dependent manner, which activates the phosphatidylinositol-3 kinase (PI3K) pathway.This indicates that there might be a positive feedback loop between the NOX complex and phosphatidylinositol-3 kinase, and continuous activation of NOX2 in injured neurites may allow the regeneration program to be maintained overtime.Thus, the molecular processes under ROS-mediated control in neuronal cells, especially in neurite degeneration and/or regeneration, are diverse and plenty of unresolved issues remain.It is also worthwhile to discuss that mitochondria have crucial roles in neurite degeneration and/or regeneration.The exciting fields of ROS-mediated signaling and neuropathic pathology will continue to advance, revealing more remarkable and unexpected regulatory targets.

Author contributions:

SW drafted the manuscript, conducted literature search and prepared figures.TA provided constructive suggestions and edited the manuscript.Both authors approved the final version of the manuscript.

Conflicts of interest:

The authors declare no conflicts of interest.

Open access statement:

This is an open access journal, and articles are distributed under the terms of the Creative Commons AttributionNonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms.