The Potential Role of Ferroptosis in Alzheimer’s Disease

INTRODUCTION

Alzheimer’s disease (AD) is a progressive neuro-degenerative disorder, and is the most common cause of dementia, accounting for an estimated 60%–80%of all cases [1]. Data from the World Alzheimer Report 2019 reported that approximately 50 million people have AD worldwide, and with an aging global population, the incidence of AD has increased dramatically. Approximately 152 million people world-wide are estimated to have AD by 2050. Moreover, AD is now the sixth leading cause of death in the United States [1]. Thus, the condition will bring a heavy burden on families and healthcare systems. Patients with AD usually require the assistance of caregivers in their daily living, especially at the late stage of disease course; this role is frequently played by children or spouses, which generates negative physical health outcomes, emotional distress, and enormous economic pressure on their family members. Moreover, healthcare systems also bear huge financial expenditures to provide services for patients with dementia.

Therefore, researchers have expended significant effort to characterize the pathogenesis of AD. Despite such dedication and investment, the pathological mechanisms of AD remain poorly understood. To date, the universally acknowledged pathological presumptions of AD are the amyloid-β (Aβ) protein cascade and tau protein theories. As the major pathological events of AD, extracellular senile plaques (SPs) of abnormally folded Aβ and intracellular neu-rofibrillary tangles (NFTs) comprising hyperphosphorylated tau impair the structure and function of neurons, which eventually causes atrophy of the cerebral cortex and hippocampus. Apart from Aβ and hyperphosphorylated tau proteins, researchers have reported other pathogenetic events in this disease course. Genetic susceptibility [2], neuroinflammation [3], gut microbiota dysbiosis [4], vascular factors [5], and iron dyshomeostasis, accompanied by lipid peroxidation and oxidative injury [6], have also been implicated in the pathology of AD.

Recently, we noted that iron dyshomeostasis, lipid peroxidation, and oxidative injury are vital regulators that trigger ferroptosis. Ferroptosis, a novel regulatory cell death (RCD) type, was first identified and defined by Stockwell et al., in 2012 [7]. It has recently attracted increasing attention in neurological disorders, and some studies have shown that it may participate in the pathological process of AD. Therefore, understanding the regulatory mechanisms of ferroptosis can generate new insights into AD pathogenesis.

PATHOPHYSIOLOGY OF AD

To date, the universally acknowledged pathological presumptions of AD involve two proteins in the brain: Aβ and tau. Aβ is a neurotoxic polypeptide derived from the amyloid-β protein precursor (AβPP). Full-length AβPP is synthesized in the endoplasmic reticulum and transported to the cell surface via the Golgi apparatus and trans-Golgi network [8]. AβPP in the body is processed via two pathways—the amyloidogenic and non-amyloidogenic pathways; AβPP hydrolysis process is mainly mediated by α-, β-, and γ-secretases. Under physiological conditions, AβPP is first cleaved by α-secretase to release the soluble (sAβPPα) and α-carboxy-terminal (CTFα) fragments through the non-amyloidogenic pathway. CTFα is then cleaved by γ-secretase to produce the p3 peptide and the AβPP intracellular domain (AICD) [9]. However, in the presence of multiple pathological factors, AβPP is sequentially cleaved by β-secretase and γ-secretase through the pathological amyloidogenic pathway to release insoluble Aβ polypeptide. Abnormally folded Aβ accumulates in the vicinity of neurons to form SPs, which further induces the downstream pathological cascade process. Recently, increasing evidence suggests that θ- and η-secretase to be involved in AβPP processing [10]. Beta-site AβPP cleaving enzyme 1 (BACE1) is universally acknowledged as the major β-secretase [11]. As a homolog of BACE1, BACE2 has also gained prominence worldwide [10, 13]. Although the amino acid sequences of BACE2 and BACE1 are highly homologous, BACE2 influences AβPP processing differently from BACE1 [12, 13]. BACE2 is a θ-secretase that cleaves AβPP within Aβ domain at the θ cleavage site, downstream of the α site that abolishes Aβ production [12, 13]. Consistently, the co-overexpression of APP/BACE2 in vivo is not associated with the increase in Aβ generation [14]. Higashi et al. innovatively proposed that membrane type 1 matrix metalloproteinase (MT1-MMP) is involved in AβPP processing, cleaving AβPP to release the truncated, soluble AβPP, the 95-kDa AβPP fragment [15]. Subsequently, MT3-MMP and MT5-MMP were also confirmed to significantly affect the cleavage of AβPP [16]. MT5-MMP, identified as a η-secretase, processes AβPP₆₉₅ at amino acids 504/505 to release the truncated, soluble AβPP fragment (sAβPP-η) and the C-terminal fragment (CTF-η) [17]. CTF-η is further cleaved by α- and β-secretase to generate corresponding Aη-α and Aη-β peptides. Aη-α decreases long-term potentiation and inhibits neuronal activity [17], which might imply the pathological effect of η-secretase in AD.

Tau protein is one of the major microtubule-ass-ociated proteins. It plays a vital role in nervous system development, combining with tubulin proteins and promoting their aggregation to form microtubules besides maintaining their stability. It is also involved in intracellular axonal transport and establishing cell polarity. However, when tau proteins are hyperphosphorylated at multiple sites, they can dissociate from microtubules, leading to their breakdown and aggregation into paired helical filaments, resulting in the formation of NFTs [18].

Besides these extracellular SPs and intracellular NFTs, some other factors are involved in the pathophysiology of AD. Apolipoprotein E (ApoE) is a polymorphic protein that participates in the transformation and metabolism of lipoproteins, and is mainly synthesized in the liver and brain tissue. Its role in the brain may be to redistribute lipids in cells and maintain cholesterol balance. However, for sporadic AD, which accounts for more than 90%of patients with AD, the ɛ4 isoform of APOE (APOE4) is the main genetic risk factor. Possessing of a single copy of APOE4 increases the risk of developing AD by approximately 3-fold, while individuals with two copies have an approximately 12-fold increased risk. Mounting evidence has demonstrated that APOE4 is implicated in the mechanism of AD pathology. Weakened Aβ clearance, impaired synaptic function via disruption of ApoE receptor signaling, loss of the blood-brain barrier (BBB) integrity, and enhancement in the inflammatory response are potential mechanisms for the role of APOE in AD pathogenesis [2, 19].

Recently, research into the role of gut dysbiosis in AD progression has increased rapidly. Increasing clinical and animal studies have been instrumental in illuminating the crucial role of the microbiota-gut-brain axis in the pathogenesis of AD. Studies have demonstrated the vital role of gut microbiota in neurodevelopment of germ-free mice, by influencing BBB integrity, immune function, and microglial activation [20]. Alterations in gut microbiota composition and diversity in transgenic mouse models of AD have been reported [21, 22]. Geng et al. reported that altered microbiota composition is accompanied by peripheral accumulation of phenylalanine and isoleucine, which leads to the differentiation and proliferation of proinflammatory T helper 1 (Th1) cells in patients with AD and mouse models. Importantly, peripheral Th1 cells infiltrate the central nervous system to stimulate the M1 microglia activation, which might initiate neuroinflammation and exacerbate neurodegeneration with respect to AD disorders [4]. Collectively, the identification of this dysregulated microbiota-gut-brain axis opens promising cues for the targeted treatment of AD. GV-971, a sodium oligomannate, can improve cognition impairment by reshaping the balance of the gut microbiota, regulating amino acid metabolism, and inhibiting neuroinflammation [4]. In China, GV-971 has been approved for the clinical treatment of patients with mild to moderate AD [23].

Much research has focused on the important role of the immune system in the pathophysiology of AD. Excessive Aβ undoubtedly initiates the neuroinflammation observed in AD; therefore, AD is also regarded as a chronic inflammatory disease. As the main innate immune cells in the brain, on the one hand, microglia function as important scavengers to clear neurotoxic Aβ polypeptides. Alternatively, inc-reasing evidence suggests that apart from exerting beneficial effects in AD, activated microglia also elicit harmful effects by releasing inflammatory mediators and cytokines [3]. Continuous patholog-ical stimulation from excessive Aβ causes a change in microglial polarization from the M2 anti-infla-mmatory state to the M1 proinflammatory state, exacerbating neuroinflammation processes [24]. Notably, the widely used M1/M2 macrophage classification is now considered less valid [25]. Based on transcri-ptome based network analysis, at least nine mac-rophage activation patterns have been proposed [26]. Considering the important role of microglia in AD, targeting dysfunctional microglial will be a promising therapeutic strategy. To this end, eliminating mic-roglia in AD transgenic mice showed promising outcomes, including easing neuroinflammation, inhi-biting neuronal loss, and improving cognitive deficit [27–29]. Besides, complement [30] and some peri-pheral innate immune cells, such as monocytes/macrophages, natural killer cells, and dendritic cells are involved in AD pathogenesis [31–33].

The complexity of AD has prompted researchers to consider various pathophysiological pathways. Because the brain is rich in unsaturated fatty acids, it is more vulnerable to lipid peroxidation damage than other organs. Consequently, lipid peroxidation and oxidative injury are currently being considered as potential treatment targets in AD [6]. Besides, as a crucial transition metal, iron affects several vital biological processes in mammals. However, iron overload can cause organ dysfunction by producing reactive oxygen species (ROS). Cumulative evidence suggests that iron overload worsens the pathological process of AD by forming free radicals and extending lipid peroxidation [34]. In conclusion, although tre-mendous progress has been made in understanding the pathological mechanisms of AD, further research is required to completely overcome this complex disease.

FERROPTOSIS

Ferroptosis is a recently described iron-dependent form of RCD, which has been discovered in various neurological disorders [7, 36]. Before proposing the concept of ferroptosis, Stockwell’s group discovered two RAS-selective lethal (RSL) compounds, erastin and RSL3, which were later confirmed to be inducers of ferroptosis. Ferroptotic cell death induced by erastin is iron-dependent, oxidative, and non-apoptotic, and can be suppressed by iron chelators and antioxidants [7]. It differs in morphology, biochemistry, and genetics from apoptosis, necrosis, and autophagy [7]. The distinct morphological feature of ferroptosis is an alteration in mitochondrial ultrastructure, such as an increase in membrane density and loss of structural integrity [7, 37]. Although little is known about the exact mechanisms of ferroptosis, we have noted that iron, lipid, antioxidant, and glutamine (Gln) metabolism are closely associated with it [35, 38–40].

IRON METABOLISM IN FERROPTOSIS

A certain amount of iron is essential for maintaining normal physiological functions in all living organisms, including mitochondrial respiration, cell proliferation, and neural signaling [41, 42]. However, when the level of iron exceeds the physiological needs of the body, the superfluous iron will trigger harmful reactions that can damage tissues, cells, and proteins, through a series of pathological events [43, 44]. Therefore, strict regulation of iron metabolism is necessary.

Ferroptosis is characterized by iron-dependent cytosolic and lipid ROS accumulation, which can be suppressed by iron chelators or antioxidants [7]. The dysfunction of numerous genes or proteins involved in iron metabolism can predispose tissues to ferroptotic damage. Stockwell’s laboratory has identified six high-confidence genes, one of which is iron-associated gene-iron response element- binding protein 2 (IREB2) [7]. Silencing IREB2 can rescue erastin-induced ferroptosis, which influences iron uptake, metabolism, and storage.

Transferrin receptor 1 (TfR1) significantly affects iron uptake and can trigger the release of iron from transferrin (Tf). Ferritin—comprising ferritin heavy chain 1 (FTH1) and ferritin light chain (FTL)—functions as an iron storage protein. As a crucial regulator of iron content, ferritin protein synthesis is increased in response to excess iron, which can protect cells against the iron overload-associated damage. These proteins synergistically regulate the homeostasis of iron metabolism in cells. Since iron plays a pivotal role in ferroptosis, preventing cellular iron overload by reducing iron uptake and increasing iron storage may suppress ferroptosis. Chang et al. found that neuroblastoma SH-SY5Y cells are refractory to erastin-induced ferroptosis by overexpressing mitochondrial ferritin [45]. Ferritinophagy, a selective autophagic phenomenon that specially targets ferritin to release intracellular iron, induces ferroptosis [46–48]. Aberrant levels of iron regulatory molecules in oncogenic-RAS families manifest as increased TfR1 mRNA levels and decreased FTH1 and FTL mRNA levels, which may explain why RSL-induced ferroptotic cell death is accompanied by increased iron content [49]. Altered levels of TfR1 and FTH1 were also found in heat shock protein B1 (HSPB1)-knockdown tumor cell lines. As a negative mediator of ferroptosis, when HSPB1 expression is downregulated, the growth of cancer cells is more easily inhibited by erastin. In HSPB1-knockdown cancer cells, iron overload caused by an increase and decrease in the expression of TfR1 and FTH1, respectively, can increase their susceptibility to the lethal activity of erastin [50]. Gao et al. confirmed that the extracellular iron carrier protein transferrin significantly influences the execution of ferroptosis [51]. Iron is not only a trigger in the aberrant accumulation of ROS, but also an indispensable element in the ROS-generating enzymes, such as nicotinamide adenine dinucleotide phosphate hydride (NADPH) oxidases (NOXs), and lipoxygenases (LOXs). Cumulative evidence suggests that these iron-associated enzymes are genuine biomarkers of sensitivity to ferroptotic cell death [40, 52–54]. Although the exact mechanistic implication of iron in ferroptosis remains elusive, it is undoubtedly significant in this cell death modality.

LIPID METABOLISM IN FERROPTOSIS

Lipids are an indispensable elements for maintaining the organization and function of biomembranes. However, lipids, especially polyunsaturated fatty ac-ids (PUFAs), are the first targets of peroxidation, and they interact with reactive oxygen or nitrogen species to generate various toxic products. PUFAs are preferentially oxidized because they possess bis-allylic protons, and the hydrogen atoms at these pos-itions are readily absorbed by some species, including ferryl and hydroxyl radicals, and peroxynitrite to produce numerous lipid radicals. Subsequently, a chain reaction of lipid peroxidation occurs by hydrogen atom transfers to the organic substrate [54–57]. Lipid peroxidation proceeds via three distinct pathways: enzymatic oxidation, free radical-mediated oxidation, and non-enzymatic, non-radical oxidation [55, 58–60]. A broad range of enzymes, such as LOXs, cyclooxygenases (COXs), cytochrome P450s, and NOXs, mediate lipid peroxidation by oxidizing their corresponding substrates, aggravating the oxidative stress injuries. Free radical-mediated oxidation is another important lipid peroxidation pathway. Divalent metals, especially ferrous ions, significantly influence this process via the Fenton/Haber-Weiss reaction. Increasing evide-nce also shows the involvement of singlet oxygen with paired electrons and ozone in lipid peroxidation by non-enzymatic, non-radical oxidation mechanisms [58, 61]. Thanks to pioneering studies by Stockwell et al., studies on lipid peroxidation in ferroptosis have increased recently. Erastin and RSL3 could trigger ferroptotic cell death by disrupting the antioxidant capability of cells. Many lipid peroxides are involved in erastin- or RSL3-induced ferroptosis [7]. Peroxidation of PUFAs, such as arachidonic acid (AA) and linoleic acid, sensitizes cells to RSL3-induced ferroptosis. LOXs play an important role in erastin-induced ferroptosis via the oxidation of PUFAs [54]. Several studies have also shown that some lipid metabolism-related genes, such as acyl-CoA synthe-tase family member 2 (ACSF2), citrate synthase (CS), acyl-CoA synthetase long-chain family member 4 (ACSL4), and lysophosphatidylcholine acyltransferase 3 (LPCAT3), are associated with ferroptosis [7, 63]. AA and adrenic acid (AdA), two representative PUFAs, can be driven into respective acyl-CoA derivatives (AA-CoA and AdA-CoA) by ACSL4, and then converted by LPCAT3 into phosphatidylethanolamines (PEs) via esterification [64], which is involved in the ferroptotic signal pathway. The suppression of this reaction by genetic ablation or pharmacological inhibition of ACSL4 has been proven to have an anti-ferroptotic effect in vitro [53]. Another study also demonstrated that Acsl4 knockout in breast cancer cells inhibited lipid oxidation and ferroptosis, revealing the essential role of ACSL4 in ferroptosis [65]. Given these results, elucidation of the lipid peroxidation mechanism may provide a major insight into the regulation of ferroptosis.

ANTIOXIDANT METABOLISM IN FERROPTOSIS

To cope with oxidative stress, aerobic organisms have established a comprehensive network of cellular antioxidant defense systems, including enzymatic and non-enzymatic constituents. Here, we highlight the value of the glutathione (GSH)/glutathione peroxidase 4 (GPX4) axis in ferroptotic cell death. GSH is an important intracellular non-enzymatic antioxidant formed by the sequential reaction of glutamate, cysteine, and glycine under the catalysis of glutamate-cysteine ligase and glutathione synthetase. It is a major substrate of the selenoprotein GPX4, also referred to as phospholipid hydroperoxide glutathione peroxidase (PHGPx) [66]. At the expense of GSH, GPX4 reduces lipid hydroperoxides to their corresponding lipid alcohols. Upon GPX4 inhibition, lipid peroxides accumulate in cells to a lethal level, resulting in ferroptotic cell death. Results from previously published studies indicate that GPX4 dysfunction is responsible for ferroptotic cell death. Stockwell et al. used a chemoproteomic technique to show that GPX4 is a candidate target protein for RSL3 [38]. Erastin, a ferroptosis-inducing small molecule, can deplete GSH by blocking cystine uptake via inhibition of the glutamate-cystine antiporter system x_c^-. Consequently, erastin has a proferroptotic effect by indirectly inhibiting GPX4 activity in cancer cells [7, 67]. Besides specific tumor-related res-earch, mounting evidence has illustrated that GPX4 is a particularly important modulator in protecting cells against ferroptotic attack in other patholog-ical conditions, including acute renal disorder [35], hepatocellular degeneration [68], motor neuron degeneration [69], T-cell immunity [70], and neurodegenerative diseases [71].

Another group of anti-ferroptotic antioxidants contains the non-enzymatic radical-trapping antioxidants (RTAs), which include endogenous vitamin E, exogenous ferrostatin-1 (Fer-1), and liproxstatin-1 (Lip-1). Vitamin E, whose family comprises hom-ologs of tocopherols and tocotrienols, is a lipid-sol-uble antioxidant. The findings of several studies have strongly suggested that vitamin E is an important factor for impedance ferroptosis. Conrad et al. showed that sufficient vitamin E (DL-α-tocopheryl acetate) supplementation could ease pathological changes in the liver and prolong the lifespan of Gpx4-knockout mice [68]. Another study found that both tocopherols and tocotrienols could hinder ferroptosis in murine Gpx4-knockout embryonic fibroblast (MEF) cells, potentially via inhibition of LOX activity [53]. Besides, Kopf et al. also suggested a synergistic interrelationship between GPX4 and vitamin E to inhibit lipid peroxidation-mediated ferroptosis in T-cell immunity, in both in vivo and in vitro models [70]. Shrader et al. reported that α-tocopherol hydroquinone, the metabolite of vitamin E, can inhibit 15-lipoxygenase activity by converting non-heme ferric ion to a ferrous state [72], which could impede the RSL3-induced ferroptosis. Fer-1, screened from a compound library and named by Stockwell’s laboratory, functions as a potent ferroptosis inhibitor by scavenging free radicals [7]. The authors of the study proposed that the anti-ferroptotic mechanism of Fer-1 is associated with the N-cyclohexyl moiety, which can anchor Fer-1 within lipid membranes and thus, curb ferroptosis. Conrad et al. discovered that the spiroquinoxalinamine derivative Lip-1 can inhibit ferroptosis in Gpx4-knockout MEFs [35]. Lip-1 could protect cells from RSL3-induced renal proximal tubule epithelial cell death. Furthermore, it may alleviate ischemia/reperfusion injury in liver tissue [35]. To date, both Fer-1 and Lip-1 have been widely employed in medical research to specifically suppress ferroptosis.

Studies by both Conrad et al. and Olzmann et al. have shown that the apoptosis-inducing factor mitochondria-associated 2 (AIFM2) gene has an anti-ferroptotic effect in the absence of Gpx4; hence, they renamed it ferroptosis-suppressor-protein 1 (FSP1) [73, 74]. Using the CRISPR-Cas9 technique, they identified FSP1 as a ferroptosis-resistant gene in RSL3-treated osteosarcoma cells and revealed that N-myristoylation was essential for its anti-ferroptotic activity. Olzmann et al. also showed that FSP1 was recruited to the plasma membrane via N-myristoylation, which functioned as a coenzyme Q₁₀ (CoQ) oxidoreductase to reduce CoQ, halt lipid peroxides, and suppress ferroptosis. The FSP1/ CoQ₁₀ pathway suggests new therapeutic avenues to treat cancer, especially ferroptosis-resistant cancers. These studies may provide new insights to explore the pathological role of ferroptosis in various diseases.

GLUTAMINE METABOLISM IN FERROPTOSIS

Gln metabolism metabolism may also affect ferroptotic cell death modulation. In cancer cells, Gln is an essential nutrient that participates in metabolic pathways. α-ketoglutarate, an important intermediate product derived from Gln, can be incorporated into the Krebs cycle via anaplerotic reactions. Aminooxyacetic acid (AOA), a small molecule transaminase inhibitor, can specifically inhibit the transaminase-mediated transamination process from Gln to α-ketoglutarate. Erastin-induced tumor cell ferroptosis can be rescued by coincubation with AOA, and this protective effect can be reversed by exogenous α-ketoglutarate supplementation [7]. Furthermore, Gln metabolism can generate a specific lipid precursor involved in ferroptosis. Using an AOA reagent and RNA interference technique targeting glutamate-oxaloacetate transaminase 1 (GOT1), Gao et al. showed that ferroptotic cell death could be inhibited. Additionally, with the help of the receptors SLC38A1 and SLC1A5, L-Gln is absorbed into mammalian cells and ferroptosis can be avoided by inhibiting these receptor functions [51]. These results further confirmed that Gln and the glutaminolysis pathway are essential for ferroptosis. Glutamine is an important precursor for the synthesis of GSH, a major substrate of the glutathione peroxidase family, which plays a vital role in ferroptotic cell death. The aforementioned investigations suggested that modulating Gln metabolism is a potential therapeutic target for cancers and other ferroptosis-related diseases.

IMPLICATIONS OF FERROPTOSIS IN AD

AD is heterogeneous and can be elicited by many factors. Besides the accumulation of extracellular SPs and intracellular NFTs, iron dyshomeostasis, lipid peroxidation, and oxidative injury are also implicated in the pathology of AD. These conditions also play an important role in ferroptosis, which suggests that ferroptosis may be associated with AD, although there are relatively few direct studies investigating this possible association. Here, we describe the potential association between AD and ferroptosis in terms of iron metabolism, lipid peroxidation, and the GSH/GPX4 axis, and review some studies that have explored the potential relation between ferroptosis and AD.

IRON DYSHOMEOSTASIS IN AD

Aging is an independent risk factor for neurodegenerative diseases, such as AD. The level of iron is increased with aging, and iron accumulates to a large extent within specific brain regions in AD, compared with age-matched controls [75]. Thus, it can be said that increased iron content runs through normal aging to AD, with markedly altered iron levels and disturbed iron homeostasis in the latter [44]. Mounting evidence has demonstrated that aberrant iron accumulation [76–79] and abnormal levels of iron regulatory molecules, such as Tf (iron transport) [80–82], ferritin (iron storage) [80, 84], and ferroportin (Fpn, iron export) [85, 86], are associated with the pathogenesis of AD. Moreover, these iron regulators are involved in ferroptotic cell death [49, 87]. Fpn is the only iron export protein in mammals that traffics iron from cells to the circulating blood. Its expression is post-transcriptionally regulated by hepcidin, which is responsible for iron homeostasis. A novel study identified ferritin/Fpn as the final effector of the ataxia-telangiectasia mutated (ATM)-metal regulatory transcription factor 1 (MTF1)-ferritin/Fpn axis, which is implicated in ferroptosis via modulation of labile iron levels [88].

The role of Fpn in ferroptosis was confirmed by the inhibition of its expression in erastin-incubated SH-SY5Y cells, which aggravated erastin-induced ferroptosis [87]. Fpn and transferrin might be considered as significant regulators associated with fer-roptosis in breast cancer [89]. Stockwell’s et al. found that RSL compounds induced ferroptosis, accompanied by the upregulation of TfR1 and down-regulation of FTH1 and FTL, which indicated that dysfunctional iron regulators TfR and ferritin are ferroptosis-mediating risk factors [49]. The autop-hagic degradation of ferritin—termed ferritinophagy and regulated by nuclear receptor co-activator 4 (NCOA4)—is associated with ferroptosis [46, 90]. Knockdown of NCOA4 inhibited ferritinophagy, res-ulting in increased FTH1 expression and easing erastin-induced HT-1080 cell death [48]. Excess free iron accelerates Aβ plaque formation, increases ferritin expression, and impairs spatial memory in AD mouse models [91, 92]. These results were suppor-ted by the protective effect of iron chelation in patients with AD [93, 94]. Deferoxamine (DFO), an iron chelator, is a potent inhibitor of ferroptosis [7]. In a histochemical study, Mufson et al. found that cells around the SPs in AD brain tissue were immunoreactive to iron, transferrin, and ferritin. These iron- and ferritin-containing cells around the SPs were presumed to be microglia [80]. A close association between iron and microglia has been validated by several in vitro and in vivo studies [95–97]. In vitro induction of inflammatory phenotype microglia with interferon-γ (IFN-γ) was acco-mpanied by increased iron accumulation, ferritin expression, and glycolysis and reduced Aβ phagocytosis [95]. These changes also occurred in APP/presenilin 1 (PS1) transgenic mice [95]. These studies revealed the complex relationship between iron metabolism, microglial activation, and neuroinflammation in AD. Other studies have focused on Tf. By analyzing blood samples from patients with AD, Beckman et al. [81] and van Rensburg et al. [82] found that a transferrin variant C2 (TFC2) is strongly associated with an increased risk of AD. Namekata et al. reported that the TFC2 variant acted in concert with the APOE4 allele in the Japanese population [98]. However, Zatta et al. reported that the TFC2 allele was significantly associated with AD only in the absence of the APOE4 allele [99]. Paradoxically, a study of a Korean sample revealed that the TFC2 variant did not increase the risk of AD, irrespective of whether the APOE4 allele existed [100]. These discrepancies might be attributed to the different regions and ethnicities, but more effort is needed to validate the potential interactions between the TFC2 and APOE4 alleles in AD. Additionally, lactotransferrin [101, 102] and melanotransferrin [103] also appear to be involved in SPs. Excessive brain iron is not only confirmed to be associated with AD pathogenesis, but could be regarded as an index to predict the conversion of mild cognitive impairment (MCI) to AD. Upon longitudinal follow-up, findings obtained from numerous studies have indicated that the cerebrospinal fluid (CSF) ferritin [41, 105], and serum non-heme iron [106, 107] levels could function as biomarkers for predicting the progression of MCI patients to AD.

Owing to the paramagnetic property of iron, magnetic resonance imaging (MRI)—especially the quantitative susceptibility mapping (QSM)—which offers an indirect representation of brain iron levels, has become a noninvasive method to detect iron deposition in patients with AD nowadays. Accumulating studies have shown that the susceptibility values of deep brain nuclei, particularly the bilateral caudate nucleus, putamen [108, 109], and globus pallidus [110] were significantly higher in patients with AD than in healthy controls. The QSM values of these brain regions are negatively correlated with neuropsychological scale scores [110]. Taken together, we reason that iron dyshomeostasis is an important pathological event in AD and may induce ferroptosis, which further causes a serious pathological cascade.

The two hallmark pathologies in the AD brain are intracellular NFTs and extracellular SPs; therefore, we have attempted to summarize the association between the iron and tau or AβPP to clarify the potential molecular mechanisms of iron dyshomeostasis in AD and the pathological process of ferroptosis (Fig. 1).

OPEN IN VIEWER

Fig. 1

The potential role of ferroptosis in AD. (1) Iron dyshomeostasis in AD, (1a) association between APP and iron, (1b) association between tau and iron, (2) lipid peroxidation in AD, and (3) the GSH/GPX4 axis in AD.

ASSOCIATION BETWEEN AβPP AND IRON

To date, a growing body of evidence show that iron is involved in the pathophysiological processes of AβPP. Iron is transported throughout the blood circulation by binding to Tf. Through the binding of the iron-Tf complex to the TfR, iron is then transported from serum to cells, where it is synthetically used or stored as ferritin. Iron regulatory proteins (IRPs) are sensitive cytoplasmic iron sensors that interact with iron regulatory elements (IREs) and post-transcriptionally regulate iron homeostasis [9]. Under conditions of iron depletion, IRPs bind to the IREs in the 5’-untranslated region (UTR) of ferritin mRNA and 3’-UTR of TfR mRNA, inhibiting the translation of ferritin mRNA and stabilizing the TfR mRNA to promote iron absorption and prevent iron storage, respectively. In contrast, during iron overload, IRPs separate from the IREs located in the 5’-UTR of ferritin mRNA and 3’-UTR of TfR mRNA to activate the ferritin mRNA translation and promote the degradation of TfR mRNA [9].

Like ferritin, AβPP is also a metalloprotein that can be post-transcriptionally regulated by IRP-IRE interaction [111]. In the absence of iron, IRPs bind tightly to IREs at the AβPP mRNA 5^′-UTR site, repressing AβPP translation. During iron overload, AβPP translation increases due to the separation of the IRP-IRE complex [111]. AβPP is processed by two distinct pathways: the α-, γ-secretase-mediated non-amyloidogenic pathway and the β-, γ-secretase-mediated amyloidogenic pathway. BACE1, also called Asp 2, is a transmembrane aspartic protease [112, 113] that mediates the amyloidogenic AβPP processing pathway. Inhibiting β-secretase activity can lower the labile iron pool [114]. Recent evidence has confirmed that nuclear factor erythroid 2-related factor 2 (Nrf2), whose mRNA levels are significantly reduced in patients with AD and animal models [115, 116], inhibits BACE1 expression by binding to the antioxidant response elements of BACE1 promoters, thereby repressing Aβ production [116]. Nrf2 is a transcription factor that regulates a considerable number of genes, many of which are associated with ferroptosis, including iron and glutathione metabolism [117]. Moreover, Nrf2 activation can protect tissues and cells from ferroptosis [118–120]. Mounting evidence suggests that a disintegrin and metalloprotease (ADAM) 9 [121], ADAM10 [121–126], ADAM17/tumor necrosis α-converting enzyme (TACE) [121, 125–127], and prohormone convertase 7 (PC7) [124] promote the α-secretase pathway, resulting in a markedly increased neuroprotective sAβPPα. Besides, the α-secretase-mediated non-amyloidogenic AβPP pathway is controlled by furin. Furin is an important cellular proprotein convertase of the proprotein convertase family; it proteolytically catalyzes the maturation of proprotein substrates, which requires an optimum pH and depends on calcium ion concentration [122]. Mook-Jung et al. confirmed that furin mRNA levels are significantly reduced both in brains of patients with AD and transgenic mice, and furin could shift AβPP to the α-secretase-mediated non-amyloidogenic pathway via the cleavage of ADAM10 and ADAM17 to inhibit Aβ generation [126]. The expression of furin protein is negatively correlated with intracellular iron levels, which suggested that furin is also implicated in governing iron metabolism regulation. Furin is present in the upstream region of hemojuvelin (HJV), and soluble HJV (sHJV) is produced through furin cleavage at the C-terminus domain [128]. Upon iron overload, the expression of furin is downregulated, accompanied by reduced sHJV. As a compensation response to iron overload, hepcidin expression is upregulated to accelerate the degradation of Fpn and decrease iron transport from cells to blood circulation. Taken together, furin serves as an upstream modulator of the HJV-hepcidin-Fpn axis to regulate iron metabolism, which is also a cofactor for ADAMs to promote AβPP shifting to the α-secretase-mediated non-amyloidogenic pathway.

The concentration of Fpn is controlled by hepcidin, which is mainly secreted by the liver in response to the levels of iron [129]. Recent evidence supports the viewpoint that hepcidin is implicated in iron homeostasis in the brain and is also associated with neurodegenerative diseases, such as AD [130, 131]. Hepcidin has been detected in neurons and glial cells in many regions of the brain [132, 133], and ameliorates Aβ-induced oxidative damage and inflammatory reactions in vitro and in vivo [131]. Studies have shown that in addition to inflammation stimulus, HJV is also an upstream regulator of hepcidin expression in iron metabolism [134, 135]. Inactivation of the Hjv gene can lead to elevated iron levels. Hjv-mutant mice fail to regulate hepcidin expression in response to iron content [134]. Chung et al. reported high levels of HJV in the brain tissue and plasma of acute ischemic stroke patients, along with severe functional disability, which is consistent with animal experiments in vitro and in vivo [135]. Hepcidin expression is slightly elevated in the ischemic cortex of Hjv knockout mice, compared with significantly increased expression in wild-type mice with ischemic stroke [135]. Overall, these studies imply an indirect relation between the HJV-hepcidin-Fpn axis and AβPP, which may be synergistically involved in the pathological process of AD.

AβPP might also function as a ferroxidase to facilitate iron efflux in cortical neurons via interaction with Fpn [136], but this view has been challenged by Hagen et al. [137, 138]. However, AβPP is indeed deficient in the ferroxidase activity, but it critically affects peripheral and brain iron content regulation [139] and can facilitate iron export by stabilizing Fpn at the cell surface [140, 141]. Similar outcomes were also observed in another study, where AβPP knockout led to age-dependent iron accumulation in both the liver and brain, accompanied by decreased brain Fpn levels [139]. The physiological location of non-amyloidogenic AβPP processing is the cell surface, where it regulates iron export by stabilizing Fpn. When AβPP is internalized from the cell surface to the endosome, amyloidogenic AβPP processing will be initiated [114]. To this end, the endocytotic shift for pathologic AβPP processing contributed to the retention of neuronal iron. Disrupting the conditions of AβPP trafficking to endosomes will increase cell-surface AβPP and Fpn, with corresponding reduction in iron accumulation in neurons [114]. We have learned from the aforementioned section that Fpn is involved in ferroptosis, which implies that AβPP might also have an indirect association with ferroptotic cell death [114]. In conclusion, AβPP and its cleavage process undoubtedly play an important role in iron homeostasis, which may indirectly promote ferroptosis, mediating abnormal pathological events in AD.

ASSOCIATION BETWEEN TAU AND IRON

Hyperphosphorylated tau-mediated intracellular NFTs are another important pathological event in AD. Oxidative stress is one of the main indirect mechanisms responsible for tau hyperphosphorylation [142]. Iron dyshomeostasis evidently increases the levels of oxidative free radicals through the Fenton/Haber-Weiss reaction and intensifies the extent of oxidative stress [77]. Emerging evidence has shown that iron accumulates in neurons with NFTs in AD brains. NFTs are associated with iron redox, a process primarily governed by ferric iron. Reducing ferric iron to ferrous iron reverses the process of tau aggregation [143], which might be due to Fe^3 + binding to the histidine residues in the Aβ peptide [144] and then sending this signal to the downstream mediator tau protein [145]. However, ferrous iron is the redox-active state involved in the Fenton/Haber-Weiss reaction, which has also been implicated in tau-mediated neurodegeneration [146, 147]. Connor et al. reported that iron exposure increased tau phosphorylation through the glycogen synthase kinase 3β (GSK-3β) pathway [146]. Consistent with these results, Wang et al. showed that compared with ferrous iron treatment alone, pretreatment with divalent metal transporter 1 (DMT1) inhibitor exhibits significantly reduced levels of phosphorylated tau, represses ROS generation, and decreases activities of cyclin-dependent kinase 5 (CDK5) and GSK-3β in SH-SY5Y cells [147]. Intranasal DFO treatment also inhibits iron-induced tau phosphorylation via the CDK5 and GSK-3β pathways in APP/PS1 double transgenic mice [148].

From the above discussion, we learn that AβPP interacts with Fpn to facilitate iron export, and tau protein may act as a downstream mediator of AβPP in this process. Tau-knockout mice develop age-dependent neurodegeneration after 6 months of age, manifesting as brain atrophy, iron retention, and neuronal loss, via perturbed AβPP-mediated iron export, and this process can be rescued by iron chelation [145]. A possible explanation is that there is a compensation mechanism for iron homeostasis, independent of tau-related systems in young tau-knockout mice, which is lost with aging [149]. However, recently, it was confirmed that tau-knockout mice could resist Aβ-induced toxicity at both 3 and 12 months of age, and inhibit iron elevation, which indicates that tau is implicated in iron retention [150]. In this study, we also noted that Aβ peptide not only reduced soluble tau content but also caused the hyperphosphorylation of the remaining tau. A drop in native tau resulted in an end to the trafficking of AβPP to the neuronal surface to stabilize Fpn, impeding the sole iron egress and eventually leading to iron retention. Tau hyperphosphorylation and iron retention might act as the downstream effectors of Aβ intoxication to promote AD progression; therefore, knockout of intoxicated tau could impede pathological cascade events. Significant increases in iron have been reported in middle cerebral artery occlusion (MCAO)-induced ischemic stroke rat models, concomitant with a marked decrease in tau levels in the lesioned hemisphere after reperfusion [149]. Based on the use of ferroptosis inhibitors Lip-1 and Fer-1, MCAO-induced infarct volumes and cognitive impairment were alleviated, which supports the role of ferroptosis in brain ischemia/reperfusion injury; however, this research did not explore the potential interaction between tau/iron and ferroptosis. Another study provided indirect evidence of iron accumulation, lipid peroxidation, and inflammation, which are the hallmarks of ferroptosis in P301S Tau transgenic mice [151], indicating a potential pathological role of ferroptosis in the context of AD.

Taken together, perturbed brain iron metabolism is implicated in the Aβ and hyperphosphorylated tau pathologies, thereby accelerating AD progression. An imbalance of iron may increase hydroxyl free radicals and induce neuroinflammation, eventually causing lipid peroxidation and redox dyshomeostasis, which are strongly associated with the ferroptotic pathway. Hence, suppressing the ferroptosis process by controlling iron levels may provide a potentially therapeutic avenue for AD.

LIPID PEROXIDATION IN AD

Lipids are the major component of cell membr-anes; once excessively oxidized, the integrity and fluidity of cellular membranes are disrupted. As the main component of phospholipids, PUFAs, especially those containing a (1Z, 4Z) pentadiene moiety, are most susceptible to lipid peroxidation [59]. When the antioxidative system cannot handle excessive lipid peroxidation, perturbed redox homeostasis occurs, leading to oxidative stress damage in a diverse set of pathological conditions. Due to the presence of a large number of PUFAs, neurons may be more predisposed to ROS assault [66], leading to a cascade of lipid peroxidation, which may conversely generate more ROS, creating a vicious circle. Membrane phospholipids and fatty acids are significantly decreased in some specific brain regions that contain abundant SPs and NFTs in patients with AD [152–154]. This loss of membrane phospholipids might occur through increased lipid peroxidation [152].Toxic Aβ peptide [155–157] and abundant brain iron [158, 159] could induce excessive production of free radicals that interact with unsaturated fatty acids via lipid peroxidation, resulting in the loss of membrane phospholipids and oxidative stress damage seen in AD pathology. Cell culture and animal studies, and autopsy specimens of AD have shown that some enzymes involved in lipid peroxidation, including LOXs [160–162], COXs [163, 164], cytochrome c [165, 166], and NOXs [167, 168], play pivotal roles in the etiology of AD pathology. Lipid peroxidation can generate a diverse set of aldehyde byproducts, including malondialdehyde, hydroxynonenal (HNE), and acrolein [169]. A growing body of work has shown elevated levels of HNE [170–172] and acrolein [171, 172] in AD brains. The increase in HNE and acrolein also occurs in MCI and early stage AD, which suggests that lipid peroxidation is an early event in the pathogenesis of AD [171, 172]. Ferroptosis is a non-apoptotic, iron-dependent, and peroxidation-driven form of regulated cell death. Most ferroptosis inducers, such as erastin, RSL3, CIL56, and FIN56, can cause lethal accumulation of lipid peroxides [60], and this suggests the crucial role of lipid peroxidation in the ferroptotic pathway. Stockwell’s team revealed that ferroptosis is driven by the peroxidation of PUFAs [54]. Natural PUFAs (H-PUFAs) are highly predisposed to peroxidation due to the presence of bis-allylic protons. Deuterated PUFAs (D-PUFAs) replace the bis-allylic hydrogens with deuterium, which can delay the radical chain reaction of lipid peroxidation compared with H-PUFAs. Substituting D-PUFAs for H-PUFAs protects G401 renal carcinoma cells from both erastin- and RSL3-induced ferroptosis. The protective effect of D-PUFAs has also been shown in the context of AD. A D-PUFAs diet prevents the formation of lipid peroxidation products and reduces the concentration of toxic Aβ peptide in APP/PS1 transgenic mice [173]. Thus, we have learned that lipid peroxidation plays a crucial role in the pathological condition of AD and is also a dominant modulator in the ferroptotic pathway, highlighting the possible success of anti-ferroptotic therapies targeting lipid peroxidation in AD (Fig. 1).

GSH/GPX4 AXIS IN AD

Ferroptosis is a novel regulated cell death modality characterized by three pivotal hallmarks: abundant and accessible cellular iron, the loss of GPX4 activity, and lipid peroxidation. Using a pharmacological approach, Stockwell et al. screened several ferroptosis-inducing compounds and revealed two main molecular mechanisms initiating ferroptosis: repression of system x_c^- and inhibition of GPX4 activity [54]. Erastin induces the ferroptotic pathway through the repression of system x_c^-, which limits cystine uptake and leads to decreased levels of cellular GSH. Excessive glutamate can also inhibit the glutamate-cystine antiporter, which imports cystine and exports glutamate in a 1:1 ratio. Notably, glutamate-induced cell death shared some common features with ferroptosis, both of which are oxidative, iron-dependent processes [7]. We have learned that glutamate excitotoxicity is also an underlying cause of AD neuropathology [174], suggesting that ferroptosis might also be involved in the pathological process of AD. RSL3 binds directly to the active selenocysteine of GPX4 to inhibit the activity of this antioxidase. In the context of antioxidant defense, GPX4 uses glutathione as a reductant to detoxify lipid hydroperoxides in the membranes. Taken together, this suggests that the GSH/GPX4 axis plays an important anti-ferroptotic role under various circumstances. Reduced GSH levels have been documented in both animal models [175, 176] and autopsy specimens [177, 178] of AD. GPX4 levels have also been shown to be reduced in AD mice compared with normal mice [71]. Overexpression of GPX4 suppresses lipid peroxidation and protects cortical neurons against Aβ-induced cytotoxicity [66]. Ablation of GPX4 is lethal both in neonatal [179] and adult mice [180], and it can lead to neuronal loss in the hippocampal region, accompanied by elevated astrocyte activation [180], suggesting an important role of GPX4 in neurodegenerative disorders. More recently, conditional neuron-specific Gpx4-knockout mice have been reported to: 1) manifest cognitive impairment, histologically accompanied by degeneration of the hippocampal neurons; and 2) exhibit features of ferroptosis, including increased lipid peroxidation, activation of ERKs, and overt neuroinflammation. More significantly, Lip-1 can improve these pathological alterations [181]. A recent study by Po-Wah So et al. revealed that iron dyshomeostasis, lipid peroxidation, and perturbed glutathione metabolism are found in AD brains, providing strong evidence of ferroptosis-related processes in AD [6] (Fig. 1).

Recent discoveries have shown that several pathways regulate ferroptosis, including the mevalonate and transsulfuration pathways, as well as other additional pathways, such as the heat shock factors 1 (HSF1)-HSPB1 pathway [60]. HSPB1, also called human HSP27, is a member of the small heat shock protein family, which are part of the heat shock protein family. Heat shock factors function as transcriptional factors responsible for modulating the synthesis of heat shock proteins under harmful stimuli. Sun et al. showed that HSPB1 overexpression can repress era-stin-induced ferroptotic cell death in cancer cells, and that this process was HSF1-dependent [50], highlighting the pivotal role of HSF1-HSPB1 in ferroptosis resistance. In response to harmful stimuli such as iron overload, oxidative stress, or neuroinflammation, HSP27 is upregulated in AD brains [182], and the protective effect of HSP27 has been documented in both experimental cell culture and animal models of AD [183, 184]. Whether heat shock proteins can alleviate ferroptotic damage in the context of AD is unclear, but this may provide a new insight into the potential role of ferroptosis in the pathological process of AD.

PROMISING FIELDS AND OBSTACLES OF ANTI-FERROPTOTIC THERAPIES IN AD

As mentioned above, iron, lipid, amino acid, and antioxidant metabolisms are vital modulators in fer-roptosis. Therefore, targeting excessive iron, lipid peroxidation, and death-promoting ROS may be an effective method of ferroptosis prevention. Iron-dependent cell death is a crucial feature of erastin-induced ferroptosis [7]. Although the role of iron in ferroptosis remains poorly understood, it is certain that iron chelators can rescue erastin-induced cell death [7]. Excessive iron deposition occurs in the specific brain areas of patients with AD [108–110]. Chelation of iron can suppress downstream ROS increase, lipid peroxidation aggravation, and oxidative stress damage, and is thus, a promising therapeutic strategy for delaying the progression of diseases. DFO has received the Food and Drug Administration (FDA) approval for treating β-thalassemia major and sickle cell patients, with well compliance [185]. Although DFO has not been approved for the clinical treatment of patients with AD, studies have identified the adjuvant potential of iron chelator in preclinical or clinical studies of AD [186]. The effectiveness and adverse effects of DFO are related to the doses and routes of administration (oral, intranasal, intramuscular, and subcutaneous treatment). Clues from DFO chelating transfusion-related iron in thalassemia patients indicated that DFO doses >2.5 g per infusion can cause numerous systemic toxicities, including ototoxicity, ocular toxicity, and liver and kidney function impairment, and doses >60 mg/kg can cause neurological disorders [187]. A two-year, single-blind clinical study reported that intramuscular-administered DFO improved daily living skills of patients with AD with a minor complication, such as weight loss and poor appetite [93]. Besides DFO, other iron chelators, such as deferiprone [188], and deferasirox [189], have been studied for AD treatment. Although many preclinical trials have demonstrated the benefit of iron chelation therapy in delaying the progression of AD, much research is required before iron chelators are available for clinical treatment. Some intractable problems, such as drug dosage, adverse effects, BBB permeability, and patient tolerability need to be addressed.

Increased lipid-based ROS is another feature of erastin- and RSL3-induced ferroptosis [7]. Erastin induces ferroptotic cell death by inhibiting system x_c^-, leading to sequential alteration of cystine uptake inhibition, glutathione depletion, and GPX4 activity suppression. RSL3 exhibits partly different cell death mechanism from that of erastin, in which GPX4 activity is directly inhibited. Both erastin and RSL3 trigger impaired antioxidant capacity, resulting in increased ROS by lipid peroxidation. Therefore, lipophilic antioxidants, such as α-tocopherol, β-carotene, and butylated hydroxytoluene, function as strong suppressors of ferroptosis [60]. As the isoform of lipid-soluble vitamin E, α-tocopherol has received wide attention because of its antioxidant properties. Increasing evidence suggests that α-tocopherol functions as a promising antioxidant therapeutic strategy for AD treatment, which reduces Aβ production and attenuates neuroinflammatory and oxidative stress damage [190, 191]. Several clinical studies have also yielded consistent results that α-tocopherol supplementation slows cognitive functional decline in patients with AD [192, 193]. Supplementation dose of clinical trials fluctuated from 800 to 2000 international units per day (IU/d), without unexpected side effects [194].

As mentioned above, we learned that Fer-1 and Lip-1 are identified as potent ferroptosis inhibitors that act by suppressing lipid peroxidation as RTAs [7, 39]. Accumulating evidence affirms the anti-ferroptotic role of Fer-1 and Lip-1 in both in vivo and in vitro studies for the treatment-related pathologies [7, 35]. However, the studies on Fer-1 or Lip-1 are still in the preclinical phase, and no relevant clinical studies exploring the therapeutic benefits of Fer-1 and Lip-1 are available. A notable point is that anti-ferroptotic therapies of Fer-1 and Lip-1 are harmful to patients undergoing cancer treatment, but it can ease the pathological events of AD, and vice versa. Thus, in the future, the target organ of ferroptotic inhibitors for AD treatment should be considered and strictly controlled for the applicable population.

CONCLUSION

AD is a group of pathologically and clinically heterogeneous diseases, and multiple factors promote its onset and progression. Presently approved clinical drugs for AD control symptoms but cannot cure its progression. Given the heavy burden caused by AD on healthcare systems, families, and individuals, there is a pressing need to explore the underlying mechanisms responsible for AD pathophysiology. Hitherto, several diverse molecular mechanisms and pathways have been shown to be involved in the pathogenesis of AD, including iron dyshomeostasis, lipid peroxidation, or perturbed glutathione metabolism; coincidentally, all of these are strongly related to ferroptotic cell death. This novel RCD modality, first coined in 2012 by Stockwell et al., has attracted widespread attention from researchers focusing on various human diseases [60], such as cancer, acute kidney failure, and Huntington’s disease. However, few studies on ferroptosis-related processes in AD have been conducted, but some indirect evidence, such as iron dyshomeostasis, lipid peroxidation, and perturbation of the GSH/GPX4 axis in AD assures us that ferroptosis is implicated in the pathogenesis of AD. Henceforth, more studies are needed to validate this view, and the results from such studies may alter the course of AD and bring benefits to patients with AD.