Ferroptosis Is Regulated by Mitochondria in Neurodegenerative Diseases

Neurodegenerative diseases are characterized by a gradual decline in motor and/or cognitive function and are caused by the selective degeneration and loss of neurons in the central nervous system [1]. The most common neurodegenerative diseases are Alzheimer’s disease (AD), Parkinson’s disease (PD), and Huntington’s disease (HD). In recent years, research has increasingly focused on iron neurochemistry and its impact on neurodegenerative diseases [2].

Ferroptosis is defined as an iron-dependent form of regulated cell death, which occurs through the lethal accumulation of lipid-based reactive oxygen species (ROS) when glutathione (GSH)-dependent lipid peroxide repair systems are compromised [3, 4]. The ultrastructural feature of ferroptosis is distinctly altered mitochondrial morphology, which distinguishes it from classic morphological alterations associated with apoptosis, necrosis, or autophagy [5]. The earliest discovery of ferroptosis was as the mechanism by which the synthetic lethal compounds erastin and RAS-selective lethal 3 (RSL3) selectively kill oncogenic RAS mutant cells. Iron chelators (deferoxamine and deferiprone) and lipid-soluble antioxidants (ferrostatin-1, liproxstatin-1, and vitamin E) were reported to be effective in suppressing ferroptosis [4, 5], suggesting the existence of iron dyshomeostasis and lipid peroxide accumulation during ferroptosis. Several salient and established features of neurodegenerative diseases (including lipid peroxidation and iron dyshomeostasis) are consistent with ferroptosis, which means that ferroptosis may be involved in the progression of neurodegenerative diseases [6, 7].

Whether mitochondria have an impact on the occurrence and development of ferroptosis remains a popular research topic, and recent studies of the nervous system have revealed that the process of ferroptosis in neurodegenerative diseases may be largely regulated by mitochondria [8]. Mitochondria are involved in the regulation of iron homeostasis in the nervous system, and studies also confirm that mitochondria play a key role in boosting the death signaling imposed by increased lipid peroxidation in neuronal cells [8, 9]. Therefore, mitochondria may regulate ferroptosis from multiple links, thereby affecting the progression of neurodegenerative diseases [10, 11]. Mitochondria provide the specific lipid precursors necessary for ferroptosis via fatty acid metabolism and glutaminolysis [4, 12]. Furthermore, lipid peroxides produced in vitro by mitochondria can cause mitochondrial lipid peroxidation and mitochondrial damage through the spread of oxidative stress, thus damaging the regulation of iron homeostasis by mitochondria, and eventually leading to the development of ferroptosis [4, 13].

This review focuses on the major neurodegenerative diseases (AD, PD, and HD) and summarizes the mechanisms of ferroptosis and its impact on the neurodegenerative process. In addition, we focus on the relationship between ferroptosis and mitochondria in neurodegenerative diseases and detail the role of mitochondria in regulating the development and occurrence of ferroptosis. This review indicates possible pathways for future research into ferroptosis and provides a reference for the treatment of neurodegenerative diseases.

Circulating iron exists in the form of Fe³⁺ by binding to transferrin. Transferrin-Fe³⁺ is recognized by the high-affinity membrane protein transferrin receptor 1 (TFR1), and the complex is then internalized via clathrin-dependent endocytosis to form endosomes [14]. Early endosomal acidification triggers conformational changes in transferrin and its receptors, thereby promoting the release of Fe³⁺. This free Fe³⁺ is subsequently reduced to Fe²⁺ by the metal reductase Steap and is then transported into the cytoplasm by divalent metal transporter 1 (DMT1) [15]. This iron then becomes part of the labile iron pool (LIP) in the mitochondria or cytoplasm, and the excess iron is stored in ferritin [16]. The export of intracellular iron depends on ferroportin (FPN), which is currently the only known iron exporter [17] (Fig. 1).

The availability of iron is one of the characteristics of ferroptosis [18]. The initial study found that the iron chelator desferrioxamine could be used to eliminate ferroptosis caused by erastin, which revealed that iron participates in ferroptosis from the beginning [5]. Furthermore, some studies have shown that transferrin is crucial for inducing ferroptosis, which further confirms the importance of iron in ferroptosis [19].

As a redox-active metal, iron can produce hydroxyl radicals by catalyzing the decomposition of H₂O₂, called the Fenton reaction, which can partially explain the accumulation of ROS in ferroptosis [20]. Moreover, enzymes involved in lipid peroxidation in ferroptosis (e.g., lipoxygenases, LOXs) have also been found to be iron-dependent [21]. Iron is essential for normal neuronal function, but excessive labile iron can increase radical-induced cell death [22]. Iron dyshomeostasis is closely associated with neurodegenerative diseases and is reflected in ferroptosis [23].

The regulation of cellular iron content at the translational level mainly depends on the iron regulatory protein (IRP 1/2)/iron regulatory element (IRE) system [24]. IRP is an RNA binding protein that controls the translation of a group of mRNAs involved in iron homeostasis by binding to IRE, and IRP2 serves as the main sensor of labile iron in neurons [24]. When the intracellular iron content is insufficient, IRP binds to IRE located in the 5′ untranslated regions (5′ UTR) of mRNA of iron-responsive proteins, such as FPN, ferritin, β-amyloid precursor protein (APP), and α-synuclein, thus inhibiting the translation of these iron-responsive proteins to reduce iron export and free iron storage. The binding of IRPs to IRE in the 3′ UTR of TFR1 and DMT1 mRNA promotes the translation of TFR1 and DMT1, which in turn promotes iron uptake [23, 25, 26]. When intracellular iron levels rise, the IRP and IRE dissociate to produce the opposite effect, reducing iron absorption and increasing free iron storage and iron export [24, 27]. IRP2 is essential in regulating iron in the brain, and IRP2 dysregulation is closely related to iron accumulation in neurodegeneration [24, 28]. In the study of aging-related auditory cortical neurodegeneration, it has been found that the upregulation of IRP2 increases intracellular iron levels and causes ferroptosis [29].

Ferritinophagy has been found to promote ferroptosis [30]. Intracellular iron storage relies mainly on ferritin, which includes ferritin light chain polypeptide 1 (FTL1) and ferritin heavy chain polypeptide 1 (FTH1). Ferritin is the main intracellular iron storage protein complex [31]. Iron is released from ferritin mainly by selective autophagy mediated by nuclear receptor coactivator 4 (NCOA4), which is known as ferritinophagy [32]. Ferritin in combination with NCOA4 is transported to lysosomes for degradation, after which iron is released for cellular physiological activities [33]. Given that NCOA4-mediated ferritinophagy regulates intracellular iron levels, it has been shown that the sensitivity of ferroptosis is influenced by NCOA4 levels [30, 32]. Inhibition of NCOA4 repressed ferritin degradation and suppressed ferroptosis, while its overexpression increases the LIP to promote the accumulation of ROS and the occurrence of ferroptosis [32]. Dysregulation of NCOA4-mediated ferritinophagy may be a potential mechanism for neurodegenerative diseases as it may have an important effect on brain iron levels [34] (Fig. 1).

Lipids are responsible for maintaining the integrity of biological membranes, and the extensive peroxidation of lipids causes changes in the assembly, composition, structure, and dynamics of lipid membranes [35], leading to cellular dysfunction and ferroptosis [36]. Polyunsaturated fatty acids (PUFAs) such as arachidonic acid (AA) and adrenic acid (AdA) are important drivers of ferroptosis [37]. Acyl-CoA synthetase long-chain family member 4 (ACSL4) catalyzes the ligation of an AA or AdA to produce AA or AdA acyl Co-A derivatives. These are then esterified into phosphatidylethanolamines (AA-PE and ADA-PE) by lysophosphatidylcholine acyltransferase 3 (LPCTA) [18]. Studies show that ACSL4 is an important driver of ferroptosis, and oxidized AA-PE and AdA-PE have been confirmed as a sign of ferroptosis [38, 39].

AA/AdA-PE peroxidation can be divided into the enzymatic and nonenzymatic pathways [40]. The enzymatic pathway of ferroptosis mainly depends on LOXs [41]. LOXs are the biggest contributor to the synthesis of lipid hydroperoxide, and LOXs are named and classified based on the position specificity of oxidation of AA [35]. Among them, 12/15-LOX may be the main participant of ferroptosis, which forms lethal species of double-and triple-oxygenated (12/15-hydroperoxy)-diacylated PE peroxides by oxidizing AA/AdA-PE, and has been extensively studied in neuronal damage and oxidative stress [42, 43]. The nonenzymatic pathway mainly relies on Fe²⁺ to catalyze the decomposition of H₂O₂ and lipid hydroperoxide (ROOH) through the Fenton reaction to produce hydroxyl radicals (OH•) and lipid alkoxy groups (RO•), thereby promoting intracellular ROS accumulation [20]. The ROS produced by the Fenton reaction in turn causes lipid peroxidation and furthers the propagation of ROS by oxidizing the polyunsaturated acyl chains in AA/AdA-PE, leading to lipid membrane integrity damage, and ultimately ferroptosis [27, 35] (Fig. 1).

The GSH redox cycle is an important cellular enzymatic defense system, protecting against damage from hydrogen peroxide and lipid hydroperoxide [44]. Glutathione peroxidase 4 (GPX4) is an intracellular antioxidant enzyme; it uses GSH as a cofactor to catalyze the reduction of lipid peroxides and protects cells and membranes against peroxidation [45], so GSH depletion or GPX4 inactivation can cause lipid hydroperoxide accumulation and subsequent ferroptosis [46]. The maintenance of GSH levels is regulated by a concentration-driven Na⁺-independent cystine/glutamate antiporter (the x_c^– system) [47]. The x_c^– system is a heterodimeric amino acid transporter comprising a light chain xCT (SLC7A11) and a heavy chain 4F2hc (SLC3A2), which provides the raw materials needed for GSH synthesis by transporting glutamate extracellularly and cystine intracellularly [47, 48]. Inhibition of the x_c^– system is the main cause of GSH depletion, as was reported in the original study of ferroptosis, where erastin induced ferroptosis by inhibiting the x_c^– system [49].

In addition, the excessive accumulation of extracellular glutamate is an important factor for the inhibition of the x_c^– system [50, 51]. A high concentration of extracellular glutamate can induce neuronal cell death, which is called oxidative glutamate toxicity [52]. The concentration of extracellular glutamate will reach a high level through a variety of pathways when the nervous system is under pathological conditions (e.g., ischemia and trauma). These pathways include the conversion of extracellular glutamine into glutamate by glutaminase (GLS) after cell lysis, and the closure of high-affinity glutamate transporters responsible for scavenging extracellular glutamate [53, 54]. The high concentration of extracellular glutamate inhibits intracellular transport of cystine (synthetic material of GSH) by the x_c^– system, leading to GSH depletion, which causes lipid peroxidation and massive ROS production. This form of cell death was originally called “oxytosis” [52]. Recent studies tend to regard oxytosis and ferroptosis as two names for the same cell death pathway because they are highly similar in molecular pathways [55, 56].

Furthermore, depletion of GSH, which is a cofactor of GPX4, causes the inhibition of GPX4 activity and activates LOX to induce lipid peroxidation [9, 57]. Numerous studies on oxidative stress and neuronal injury have found that depletion of GSH induces the activation of 12/15-LOX, and signal transducers and activators of transcription may be involved in upregulation of 12/15-LOX levels [42, 58]. Additionally, it was found that activated 12/15-LOX amplifies oxidative stress by attacking mitochondria, which causes mitochondrial injury and massive ROS production [59]. In addition to 12/15-LOX, the use of 5-LOX inhibitor zileuton in HT22 cells resisted ferroptosis induced by erastin, which suggests that 5-LOX may also affect ferroptosis, but further research is needed [60]. The other mechanism that induces ferroptosis is the direct binding of RAS-selective lethal small molecule 3 (RSL3) to GPX4, and the suppressed GPX4 fails to clear the accumulated lipid hydroperoxides, thereby inducing ferroptosis [41, 49]. Therefore, inactivation of GPX4 is also a key factor that induces ferroptosis development [61] (Fig. 1).

AD is a chronic, progressive and irreversible neurodegenerative disease characterized by memory loss, language disorders, severe behavioral abnormalities, and learning deficits [62]. Previous studies have shown that the aggregation and accumulation of amyloid-β (Aβ) in AD leads to neuronal cell death [63]. However, the treatment of Aβ alone does not prevent the progression of AD, suggesting the presence of other factors [64]. Recent research has found evidence of ferroptosis in AD, including iron dyshomeostasis, increased expression of xCT and lipid peroxidation, coexistent with an augmented excitatory glutamate:inhibitory GABA ratio [65].

Many studies have reported that iron dyshomeostasis is one of the causes of the pathogenesis of AD, alongside amyloid plaques and tau tangles [66, 67]. Iron is essential for normal neuronal function, but the excessive labile iron will cause oxidative stress and cell death [22]. Quantitative susceptibility mapping has shown that there is obvious iron accumulation in multiple areas of the AD brain [68]. Studies on cerebrospinal fluid ferritin also suggested that iron may play a role in promoting disease progression in the prodromal stage of AD [69]. Accumulation of iron also accelerates the degree and speed of Aβ accumulation into β-amyloid plaques and is closely related to the decline of cognitive ability in AD patients [70, 71]. Apolipoprotein Eε4 allele (APOE4) is the strongest known genetic risk factor for sporadic AD [72]. A study on APOE4 carriers found that APOE4 may cause deposition of iron in AD brains by reducing the delivery of high-density lipoprotein (an important lipoprotein in the regulation of intracellular iron levels) [72, 73].

Normally, the iron homeostasis within cells is strictly regulated, as they are chelated by the main cellular iron storage protein ferritin in a bioavailable and nontoxic form [74]. The expression of ferritin FTH and FTL was enhanced in AD brains, suggesting an increase in LIP in AD [65]. The ferritin in AD may be different from the physiological ferritin, and its structural changes and functional disorders would increase toxic brain ferrous ions [75]. In addition, ferritin aggregate formation has been found in studies of neurodegeneration, which leads to functional ferritin defects, thereby exacerbating iron-mediated oxidative stress [76]. An increase in compensatory ceruloplasmin is observed in AD, which is necessary to oxidize Fe²⁺ to Fe³⁺ and promote FPN-mediated iron export [65]. FPN levels are abnormally downregulated in AD, leading to a decrease in iron export, causing an increase in LIPs within the cell and generation of lipid peroxidation (a sign of ferroptosis) [77]. Elevated ACSL4 levels have been found in AD, and ACSL4 may determine the sensitivity of ferroptosis by increasing lipid peroxidation [38]. Indeed, an increase in lipid peroxidation products has been found in AD [78]. Furthermore, enhanced expression of the x_c^–system light chain subunit xCT was observed in AD [65]. Upregulation of xCT suggested that cells are affected by oxidative stress, such as glutamate-induced oxidative toxicity, and the x_c^–system is inhibited [79]. A higher ratio of excitatory glutamate to inhibitory GABA is indeed found in AD [65]. Excessive extracellular glutamate reduces the synthesis of GSH by inhibiting the x_c^–system, and thus disturbs the redox balance of cells and induces ferroptosis [50, 51].

Tau tangles in AD may also be a potential factor leading to iron dyshomeostasis and ferroptosis [80]. APP plays an important role in regulating the iron content of neurons; it promotes iron export by maintaining the stability of FPN [81, 82]. Iron status may directly impact the translation of APP through an iron-responsive mRNA element (IRE) in its 5′-UTR [83]. Tau is a microtubule-associated protein that has the function of transporting APP to the cell membrane and stabilizing the FPN1-APP complex [84]. Therefore, the inactivation of tau affects the transport of APP to the cell membrane, resulting in iron accumulation and oxidative stress-induced cell death in AD, which may be related to ferroptosis [80, 84] (Fig. 2).

PD is caused by the death of midbrain dopaminergic neurons, and the most severe degeneration occurs in a subgroup of melanized neurons located in the substantia nigra (SN), where iron accumulation and enhanced lipid peroxidation have been detected [85]. Recent studies have confirmed that ferroptosis is involved in the death of dopaminergic neurons in PD. For example, a human dopaminergic neuron precursor-derived cell line (from LUHMES cells) has unique sensitivity to ferroptosis and exhibits distinct characteristics of erastin-induced ferroptosis; in addition, ferroptosis in LUHMES cells is signaled by protein kinase Cα (PKCα)-activated MEK, independent of RAS activation, suggesting that there are some metabolic pathways that are specific to dopaminergic cells [86].

Significant iron deposition has been detected in the SN pars compacta in PD, and the accumulation of iron is consistent with the progress of PD [87, 88], suggesting that elevated iron levels from various causes may be important in the induction of ferroptosis in PD [89, 90]. Transferrin can serve as a natural ligand for FPN-mediated iron export to deliver and remove iron from cells, and it is found that transferrin has been depleted in PD SN, which may be an important factor for iron accumulation in PD SN [91]. In addition, a significant reduction in ceruloplasmin activity has been reported in the SN of PD patients [92]. Ceruloplasmin, a multicopper ferroxidase, is an iron-export ferroxidase that is abundant in plasma and is expressed in glial cells [93]. It promotes cellular iron export by oxidizing Fe²⁺ present on the cell surface by FPN and integrates Fe³⁺ incorporation into extracellular transferrin [94]. Reduced ceruloplasmin activity in PD can cause lipid peroxidation and neurodegeneration in the SN by inducing iron accumulation [92]. In addition, a dissociation between iron accumulation and ferritin upregulation was observed in the aged SN previously, and it was found that higher levels of iron were not associated with increased levels of ferritin [95]. There are also studies showing that ferritin levels in the SN of PD patients are decreased, and H-ferritin and L-ferritin have been reported to be 75 and 37% of normal values, thus making neurons susceptible to iron-induced oxidative damage [96]. The use of iron chelator deferiprone and ferroptosis inhibitor ferrostatin-1 is beneficial to PD, further indicating that ferroptosis is involved in the process of PD [86, 97, 98].

Studies have demonstrated that the imbalance in iron homeostasis in PD is affected by NO [99]. NO is recognized as a physiological medium in the brain and plays an important role in neural network long-term potentiation, synaptic plasticity, and activity-dependent modification [100]. NO can regulate cellular iron metabolism and affect intracellular iron homeostasis through its ability to inactivate aconitase and activate IRP-1 [101, 102]. When iron levels are high, IRP-1 assembles a [4Fe–4S] cluster and acquires aconitase activity, while also losing its function as a posttranscriptional regulator [103]. NO directly attacks the Fe-S cluster of aconitase, induces its decomposition, and converts the enzyme to IRP-1 [104]. The overexpression of IRP-1 leads to increased TFR and DMT1 levels and reduced ferritin levels in PD, resulting in intracellular iron accumulation and subsequent ferroptosis [7]. In addition, the elevated NO that occurs in PD also inhibits APP translation by promoting the binding of IRP to IRE in the 5′ UTR, resulting in impaired iron release and iron retention in the SN, which further confirms the involvement of NO in iron imbalance and ferroptosis [99].

Similar to AD, tau deficiency has also been detected in PD, which impairs the function of APP transport to the membrane and inhibits iron release from dopaminergic neurons, thereby causing intracellular iron accumulation and loss of dopaminergic neurons [84, 105]. In addition, it has been found that DJ-1 (a protein closely related to PD) plays a key role in protecting cells against ferroptosis by maintaining cysteine synthesized by the transsulfuration pathway [106]. The transsulfuration pathway is another way to generate cysteine in cells besides the introduction of cystine by the x_c^– system [107]. Studies have shown that DJ-1 maintains the activity of S-adenosylhomocysteine hydrolase (a key enzyme of the transsulfuration pathway), thereby maintaining GSH synthesis and resisting ferroptosis when erastin inhibits the x_c^– system [106]. The absence of DJ-1 is an important factor for the early occurrence of PD, which suggests that ferroptosis may be involved in the occurrence of PD [106, 108].

α-Synuclein accumulation is an important feature of PD [109]. While α-synuclein’s role in ferroptosis has not been clearly explored, some studies have found that it is related to iron [110]. α-Synuclein has been identified as a ferrireductase, combining with Fe³⁺ and reducing it to Fe²⁺, and cells with purified recombinant α-synuclein or that overexpress α-synuclein have significantly increased Fe²⁺ levels [111]. In another experiment, it was reported that toxic oligomers of α-synuclein can interact with metal ions such as iron, resulting in iron-dependent oxidative stress [112]. In addition, IRE has been detected in the 5′ UTR of mRNA that encode α-synuclein [113], which indicates that α-synuclein may affect iron homeostasis, but whether it affects ferroptosis still needs further study.

Studies have shown that nuclear factor erythroid 2-related factor 2 (NRF2) is involved in the regulation of iron levels in PD and resists oxidative stress [114]. Under normal conditions, NRF2 resides in the cytosol bound to its negative regulator Kelch-like ECH-associated protein 1 (Keap1) [115]. However, states of increased oxidative stress facilitate the dissociation of NRF2 from Keap1, and promote the nuclear translocation of NRF2 [115]. NRF2 has been demonstrated to regulate the activities of various proteins related to ferroptosis and lipid peroxidation [116]. Both the light chain and heavy chain of ferritin, the key iron storage protein, as well as FPN, which is responsible for iron efflux out of the cell, are controlled by NRF2 [117]. NRF2 is also involved in the regulation of lipids via the ligand-mediated transcription factor peroxisome proliferator-activated receptor gamma (PPARγ) [118]. PPARγ is a major regulator of lipid metabolism and can be activated by oxidized lipids relevant to the initiation of ferroptosis [119]. In addition, many enzymes related to GSH synthesis and metabolism are regulated by NRF2, including glutathione synthetase, catalytic and modulatory subunits of glutamate-cysteine ligase, as well as xCT [120-123]. GPX4, which is closely related to ferroptosis, is also an established NRF2 transcription target [124]. Despite numerous studies suggesting that NRF2 may be involved in resistance to ferroptosis, direct in vivo evidence for an anti-ferroptotic effect of NRF2 induction is still lacking, and it is currently unknown whether and to what extent the demonstrated neuroprotective efficacy of NRF2 activation in models of neurodegeneration involve attenuation of ferroptosis, so further research is needed [11] (Fig. 2).

HD is an autosomal dominant, fatal neurodegenerative disease caused by abnormal CAG repeats in the gene encoding the ubiquitin-expressing protein Huntingtin [125]. Several recent studies have observed many features of ferroptosis in HD, suggesting that HD pathology is likely to involve ferroptosis [126]. In striatal neurons in the R6/2 HD mouse model, the lipid peroxidation marker 4-hydroxy-2-nonenal (4-HNE) adducts was significantly increased. In addition, 4-HNE colocalized with mutant Huntingtin protein (mHtt) in this mouse model, indicating that enhanced lipid peroxidation occurs in HD [127]. Furthermore, a significant decrease in GSH was observed in the hippocampus of a 3-nitropropionic acid (3-NP)-related HD rat model, suggesting a weakened antioxidant defense capacity in HD [128]. Previous research has revealed that excess iron deposition occurs in certain areas of the HD brain, such as the striatum and globus pallidus, and that elevated iron levels precede changes in brain tissue morphology, suggesting that iron dyshomeostasis is involved in HD neurodegeneration [129]. In addition, an increase in ferritin and FPN and a decrease in IRP and TFR have been detected in the striatum and cortex of the R6/2 HD mouse model, and these changes may be compensatory responses to iron accumulation in HD [130]. A decrease in NRF2 activation response has also been found in HD neural stem cells [131]. These ferroptosis-related characteristics are all found in HD, further suggesting that HD has a close relationship with ferroptosis [126].

Studies have reported that mice expressing mHtt in both neurons and astrocytes have more severe neurological symptoms than mice expressing mHtt in neurons only, suggesting that astrocytes are important in HD [132]. The accumulation of mHtt in astrocytes can reduce glutamate transporter protein 1 (GLT-1), thereby reducing GLT-1-dependent glutamate reuptake and leading to excessive accumulation of extracellular glutamate and, in turn, excitatory toxicity [133]. Extracellular glutamate accumulation can inhibit the x_c^– system and thus the transport of cystine to neurons, leading to insufficient GSH synthesis and a resulting increase in lipid peroxidation and ferroptosis [134].

Glutamic acid accumulation also triggers a unique neuronal iron homeostasis regulatory mechanism: the N-methyl-D-aspartate receptor (NMDAR)-NO-dexras1-peripheral benzodiazepine receptor-associated protein 7 (PAP7)-DMT1 iron uptake signaling cascade [135]. NMDAR is a glutamate receptor and ion channel protein that is found in nerve cells. Glutamate can activate NMDAR, thus allowing Ca²⁺ to enter cells, and Ca²⁺/calmodulin further induces neuronal NO synthase to synthesize NO [136]. NO then causes s-nitrosylation of Dexras at cysteine 11, thus activating this brain-enriched member of the RAS family of small G proteins that is selectively induced by dexamethasone [137]. Activated Dexras induces iron uptake by interacting with the Golgi protein acyl-coenzyme A binding domain containing 3 proteins (ACBD3), and eventually acting on DMT1 [135]. This indicates that mHtt can cause glutamate excitotoxicity and lead to iron accumulation and ferroptosis in HD [126]. Ras homolog enriched in striatum (Rhes) is closely homologous to Dexras1 and is specifically expressed in the striatum [138]. Studies have reported that Rhes induces iron uptake by DMT1 through a mechanism similar to that of Dexras1 affected by glutamic acid toxicity, and that Rhes can combine with mHtt to cause sulfonylation of mHtt, thereby increasing mHtt neurotoxicity [138, 139].

Heat shock proteins (HSPs) are also involved in the regulation of ferroptosis in HD [140]. HSPB1 (also known as mouse HSP25 or human HSP27) is a small HSP, and its expression is regulated by heat shock factors (HSFs), the major transcription factors that induce HSP synthesis during stress [141]. HSPB1 phosphorylation is regulated by PKC, and studies have reported that phosphorylated HSPB1 can provide resistance to erastin-induced ferroptosis by inhibiting cytoskeleton-mediated iron uptake and the production of lipid ROS [140]. Furthermore, HSPB1 is effectively activated during acute stress, but this activation does not occur in the chronic degeneration that is observed in HD, which may be an important cause of iron accumulation and oxidative damage in HD [142]. In addition, HSPB1 has been demonstrated to prevent tau accumulation in hippocampal neurons and promote its degradation or refolding [143] (Fig. 2).

Mitochondria, as the center of energy metabolism in neurons, are the main site of ROS production, and are also very vulnerable to oxidative stress [144]. It remains controversial whether mitochondria play a decisive role in ferroptosis and its development, because some of the initial studies into ferroptosis did not report any effect of mitochondria on ferroptosis. For example, it was initially reported that erastin-induced ferroptosis does not require the mitochondrial electron transport chain [5], that a mitochondria-targeted superoxide probe MitoSOX is not oxidized in erastin-treated cells [5], and that the ferroptosis inhibitor ferrostatin-1 cannot inhibit rotenone-induced MitoSOX oxidation [145]. However, given that these earlier mechanistic probes were indirect and potentially nonspecific reporters of mitochondrial lipid peroxidation [146], and that the only unique morphological feature of erastin-treated cells is that mitochondria become smaller and membrane density increases [5], it may still be that mitochondria play an important role in ferroptosis [127]. Indeed, mitochondria have been shown to play a key role in neuronal oxidative death signaling in glutamate-induced ferroptosis model [9]. Here, we discuss in detail the relationship between mitochondria and ferroptosis in neurodegenerative diseases.

There is increasing evidence that mitochondria are involved in cell metabolism during ferroptosis, including lipid metabolism and amino acid metabolism (involving cysteine and glutamine) [10]. Mitochondria are the main executors of cell metabolism, and previous studies have demonstrated that glutaminolysis in mitochondria is an important inducer of ferroptosis [19]. Glutaminolysis refers to the conversion of glutamine to α-ketoglutarate (α-KG), catalyzed by GLS, which supplements tricarboxylic acid (TCA) cycle intermediate products; this is an anaplerotic reaction [147]. α-KG and its downstream TCA cycle metabolites, including succinic acid, fumaric acid, and malic acid, can produce citrate, an important initiator of fatty acid metabolism throughout the TCA cycle. This suggests that glutaminolysis is involved in the synthesis of specific lipid precursors that are required for ferroptosis [12]. In addition, previous studies have shown that mitochondrial GSL2 (rather than cytosolic GSL1) catalyzes glutaminolysis during ferroptosis, further indicating that mitochondria are involved in the development of this cell death mechanism [19]. In addition to glutaminolysis, mitochondrial fatty acid metabolism may also provide the specific lipid precursors that are required for ferroptosis [4]. Acyl-coenzyme A synthetase 2 (ACSF2) and citrate synthase (CS) are enzymes that are necessary for mitochondrial fatty acid metabolism, and studies have reported that ACSF2 and CS knockdown can inhibit erastin-induced ferroptosis. These results suggest that mitochondrial fatty acid metabolism is involved in ferroptosis [5].

Mitochondria not only provide the lipid precursors that are required for ferroptosis, but also participate in the regulation of iron homeostasis in neurodegenerative diseases [8]. Mitochondrial ferritin (FtMt) is an iron storage protein that is located in the mitochondria, and it has a structure and function similar to that of cytoplasmic ferritin [148]. FtMt expression is limited to mitochondria in testicular cells, the central nervous system, and other high oxygen-consuming tissues [149]. Previous studies have revealed that FtMt regulates cellular iron metabolism, and that high mitochondrial FtMt expression can reduce iron accumulation in the cytoplasm and significantly inhibit erastin-induced ferroptosis in SH-SY5Y cells [27]. These results indicate that FtMt may affect ferroptosis by regulating iron homeostasis in neurodegenerative diseases [8, 150]. Mitoferrin-1 and -2 promote mitochondrial iron uptake, which may be beneficial under physiological conditions but have deleterious effects under conditions of iron dyshomeostasis, such as the increased expression of mitoferrin-1/2 that is observed in PD models [151]. In contrast, the outer mitochondrial membrane protein CDGSH iron sulfur domain 1 (CISD1, also termed mitoNEET) inhibits mitochondrial absorption of Fe²⁺ and its transport to the mitochondrial matrix. The pharmacological or genetic inhibition of CISD1 enhances mitochondrial iron accumulation and subsequent mitochondrial lipid peroxidation, thus increasing erastin-induced ferroptosis [13]. Mitochondrial iron accumulation is also found in ferroptosis related to Friedreich’s ataxia, and the mitochondrial-targeted antioxidant XJB-5-131 is effective in inhibiting ferroptosis [152]. Although mitochondrial iron accumulation has been observed in some studies, fluorescent probes have recently been used to detect the alterations of intracellular iron during the execution of ferroptosis and found the conspicuous elevation of labile iron in the lysosomes and endoplasmic reticulum, but not mitochondria [153]. This may be affected by the time of detection or cell type, but it also suggests that the role of labile iron in mitochondria related to ferroptosis needs further exploration [150, 153].

Mitochondrial lipid peroxidation may be a key process in the development of ferroptosis in neurodegenerative diseases [154]. In an early study of the connection between GPX4 knockout and neurodegeneration, it was found that inhibition of the GSH/GPx4 system converts oxidative stress into a 12/15-LOX-dependent lipid peroxide signal that finally activates apoptosis-inducing factor (AIF) [9]. Relocation of AIF from mitochondria to the nucleus indicates that mitochondria are damaged by lipid peroxidation, further suggesting that mitochondria may play an important role in neuronal cell lipid peroxidation and death signaling [9]. It has been reported that the initiation of lethal lipid signals from GPX4^–/– cells appears to occur outside of the mitochondria, suggesting that ferroptosis is caused by lipid peroxidation outside the mitochondria [4, 155]. Sterol carrier protein 2 (SCP-2) is a mitochondrial transporter that preferentially transports oxidized lipids to the mitochondria in the transmission of oxidative stress between intracellular compartments. The inhibition of SCP-2 can inhibit erastin-induced ferroptosis, suggesting that mitochondrial lipid peroxidation is involved in this cell death mechanism [155]. In addition, studies have used mitochondrial-targeted antioxidants to confirm the effects of mitochondrial lipid peroxidation on ferroptosis. For example, nitroxide XJB-5-131 is a mitochondrial-specific antioxidant that strongly inhibits the formation of oxidized fatty acids from cardiolipin, a mitochondrial-specific phospholipid [146]. Studies have shown that XJB-5-131 can strongly inhibit erastin- or RSL3-induced ferroptosis, suggesting that the protection of mitochondrial lipids may be sufficient to prevent ferroptosis, and further confirming that mitochondrial lipid peroxidation is a key factor in ferroptosis [146]. In ferroptosis caused by cysteine deprivation, it was reported that lipid ROS first has a distribution that significantly colocalizes with mitochondria, and then appears in the plasma membrane. Moreover, mitochondrial ETC may participate in the production of lipid ROS, further confirming that mitochondria play an important role in ferroptosis [12]. In addition, some studies have demonstrated that CISD1 inhibits ferroptosis by preventing mitochondrial lipid peroxidation, thus confirming that mitochondrial lipid peroxidation promotes the process of ferroptosis [13].

Recent studies have suggested that mitochondrial damage may be the final step of ferroptosis-related oxidative cell death in neurodegenerative diseases [156]. In a study of neurodegenerative diseases using PC12 cells, tert-butyl hydroperoxide (t-BHP) was reported to induce ferroptosis, which was accompanied by a decrease in mitochondrial membrane potential and ATP production as well as an increase in mitochondrial ROS [156]. Previous studies have demonstrated that the separation of IRP-1 from IRE is ATP-dependent, so the depletion of ATP leads to enhanced IRP-1 and IRE binding activity, which in turn causes TFR upregulation and a subsequent increase in iron content [157]. In addition, while ferrostatin-1 suppressed ferroptosis, it also reversed mitochondrial dysfunction, suggesting that mitochondrial dysfunction is closely related to ferroptosis [156]. Similar phenomena have been reported in research using another type of neuronal cell, HT-22, indicating that lipid peroxidation and an increase in ROS caused by ferroptosis can induce transactivation of the mitochondrial-dependent proapoptotic protein Bid, which in turn leads to a loss of mitochondrial membrane potential and mitochondrial membrane integrity, an impairment in ATP synthesis, and an increase in ROS production and a release of AIF from mitochondria to the nucleus [56]. These results are in line with earlier studies demonstrating a role for AIF in oxidative cell death by loss of GPX4 and suggest that the protection of mitochondria may be an important treatment measure for neurological diseases with ferroptosis [158] (Fig. 3).

In summarizing the related mechanisms of ferroptosis, we have focused on the role of ferroptosis in neurodegenerative diseases and revealed that mitochondria are an important regulatory unit. Ferroptosis is an important influencing factor for lipid peroxidation and cell degener-ation in neurodegenerative diseases. The process of ferroptosis is closely linked to mitochondria: mitochondria regulate many aspects of ferroptosis, including cell metabolism, iron dyshomeostasis, and lipid peroxidation, eventually leading to neurodegeneration and neurodegenerative diseases. Although many studies have reported the effects of ferroptosis in the nervous system, many of the potential mechanisms remain poorly understood and need to be explored in future research.

We thank Bronwen Gardner, PhD, from Liwen Bianji, Edanz Editing China (www.liwenbianji.cn/ac), for editing the English text of a draft of the manuscript.

The authors have no conflicts of interest to declare.

Funding was supported by the Jilin Provincial Research Foundation for the Development of Science and Technology Projects (Grant No. 20190201152JC); Hospital Youth Foundation of the 1st Hospital of Jilin University (Grant No. JDYY102019018); Foundation of Jilin Provincial Department of Finance (Grant No. JLSCZD 2019-015); Industrial technology research and development project of Jilin development and Reform Commission (Grant No. 2020C034-4).

Juepu Zhou: conceptualization, writing of the original draft. Yao Jin: data curation. Yuhong Lei and Tianyi Liu: visualization. Zheng Wan: software. Hao Meng: conceptualization, writing – review & editing. Honglei Wang: conceptualization, project administration.