Glutathione depletion induced by erastin leads to a decreased activity of a selenium enzyme, GPX4, which neutralizes lipid peroxides. GPX4 catalyzes the reduction of lipid hydroperoxides (LOOH) to the corresponding alcohols (LOH) with reduced glutathione (GSH), whose regeneration from the oxidized glutathione is provided by NADPH-dependent glutathione reductase. The inhibition or depletion of GPX4 is central to the initiation of ferroptosis as its function relies on the oxidation of the vital cellular antioxidant, GSH. Direct inhibition of GPX4 by RAS-selective lethal (RSL3) exerts a more potent effect than erastin. Within 8 and 4 h of the treatment by 10 μM erastin and 1 μM RSL3, respectively, an increase in lipoxygenase enzymes (specifically LOX-12 and LOX-15) is seen by immunoblotting (Shintoku et al., 2017). This suggests that both treatments push the cells to make more of these enzymes by activating their expression or enhancing their translation. Of note, a similar phenomenon was observed long before the term “ferroptosis” was coined in a glutathione depletion model induced by glutamate in immature neurons, resulting in a 3-fold increase in LOX-12 protein detected at 8 h of incubation (see Figure 3C in (Li et al., 1997)). Thus, it is not surprising, given the substantial similarity in the hallmarks of ferroptosis observed both in cancer cell lines and immature neurons, as reported recently (Zille et al., 2019).

The exact mechanism leading to this phenomenon is yet to be determined. Still, we can speculate that this effect is related to the change in the functioning of an iron chaperone Poly (rC)-binding protein (PCBP1). It binds iron from the chemically reactive, labile iron pool (LIP) in the cytosol and delivers it to ferritin and non-heme iron enzymes via direct, metal-mediated protein-protein interactions. PCBP1, independent of its iron-binding activity (Patel et al., 2021), represses the translation of 15-lipoxygenase (Lee et al., 2022) (Figure 1). The transcriptional and translational repressing activity of PCBP1 is controlled by phosphorylation. When another ferroptosis inducer, sorafenib (an FDA-approved multi-kinase inhibitor), is added, the phosphorylation of PCBP1 changes within 60 min (Werth et al., 2020). Hence, we may speculate that at the ferroptosis initiation step, there has to be a change in PCBP1 phosphorylation that results in the de-repression of LOX-15 translation on the ribosome and the production of more LOX protein. Of note, PCBP1 not only regulates LOX-15 but also inhibits autophagy by binding to the CU-rich elements on the 3′-untranslated region (3′-UTR) of Beclin 1 (BECN1) mRNA (Lee et al., 2022). Hence, de-repression of BECN1 will lead to iron-mediated ferritinophagy. In addition, AMPK-mediated BECN1 phosphorylation results in blocking system Xc- (Song et al., 2018). Although the de-repression of LOX-15 and BECN1 is likely to happen simultaneously, the activation of the LOX-15-catalyzed reaction will trigger ferroptosis initiation, whereas ferritinophagy will represent the later propagation step. The translational repression of LOX-15 and BECN1 makes PCBP1 a seminal ferroptosis regulator by inhibiting iron-mediated ferritinophagy and lipid peroxidation (Figure 1).

A known kinase cascade involving PI3K-RAC1-PAK1 leads to the phosphorylation of PCBP1, eventually causing its degradation (Meng et al., 2007). However, the detailed pathway resulting in PCBP1 phosphorylation/dephosphorylation in ferroptosis initiation by various inducers requires further investigation. Another path to catalytic activation of LOX is well-documented for the LOX-12 enzyme, which is known to be activated by glutathionylation (Park et al., 2009), the process triggered by glutathione depletion, and by decreasing the ratio of reduced to oxidized glutathione.

Lipid peroxidation stands as the significant characteristic and driving force of ferroptosis (Stockwell et al., 2020). This oxidative process can occur either enzymatically, driven by non-heme iron dioxygenases - lipoxygenases, as discussed above, or non-enzymatically (Ayala et al., 2014) through a radical chain reaction with likely participation of the ferrous-ion catalyzed reaction with hydrogen peroxide to generate hydroxyl-radical, the so-called Fenton reaction (Stockwell et al., 2020; Dang et al., 2022). The role of the Fenton reaction in lipid auto-oxidation is still debated. Polyunsaturated fatty acids (PUFAs) are especially prone to autoxidation because the C–H bonds of the methylene groups flanked by C-C double bonds are among the weakest C–H bonds known (ca. 76 kcal/mol). Increasing the number of double bonds in the lipid leads to a corresponding increase in the rate of its autoxidation (Conrad and Pratt, 2019).

Lipid hydroperoxides, specifically those of PUFAs, like arachidonoyl (AA) and adrenoyl (AdA) acids, induce severe damage to the lipids in the plasma membrane, leading to ferroptosis (Dang et al., 2022). The secondary oxidation products are aldehydes such as malondialdehyde (MDA) and 4-hydroxynonenal (4-HNE), which have even more damaging chemistry. AA and its elongation product, AdA, are incorporated into membrane phospholipids via acyl-CoA synthetase long-chain family member 4 (ACSL4) and lysophosphatidylcholine acyltransferase 3 (LPCAT3) catalyzed reactions. Through genome-wide screening approaches, ACSL4, LPCAT3, and cytochrome P450 reductase (POR) (Zou et al., 2020) have been identified as genes critical to ferroptosis. Phosphatidylethanolamine (PE)-linked arachidonic acid (PE-AA) and adrenic acid (PE-AdA) are considered the most vulnerable phospholipids to peroxidation. LOX-15 generates specific oxidation products like 15-hydroperoxy-AA-PEs or 15-hydroperoxy-AdA-PEs), which play pivotal roles in ferroptosis; moreover, (15-hydroperoxy)-di-acylated PE species acting as death signals (Kagan et al., 2017). High levels of 15-hydroperoxy-AA-PEs in cell membranes make them prone to oxidative cleavage, impairing functional proteins and cellular integrity (Ren et al., 2020; Ou et al., 2022). Additionally, 15-hydroxyeicosatetraenoic acid, the major 15-LOXs metabolite of AA, induces endothelial tight junction disruption and blood-brain barrier (BBB) dysfunction (see review on BBB in ferroptosis and references therein (Chen X. et al., 2022).

Classification of the experimentally observed cell death as ferroptosis is primarily based on its inhibition by a small molecule inhibitor, ferrostatin-1 (Fer-1) (Figure 2A), that supposedly works as a radical trapping antioxidant. The inhibition of LOX-catalyzed peroxidation with Fer-1 has not been observed, thus supporting Fer-1 action on non-enzymatic peroxidation. However, a modest inhibition (ca. 2-fold) has been reported for the catalytic activity of a complex formed by LOX-15 and phosphatidylethanolamine-binding protein 1 (PEBP1) (Anthonymuthu et al., 2021). The design of inhibitors explicitly targeting the enzyme complex yielded two lead compounds, FerroLOXIN-1 and 2 (Figure 2B), which effectively suppressed ferroptosis in vitro and in vivo without radical scavenging or iron chelation. The inhibition by FerroLOXINs occurs via a specific interaction with the LOX-PEBP1 complex: they may change the substrate binding mode to the complex or block oxygen from reaching the catalytic center, preventing the peroxidation process from happening (Dar et al., 2023).

The debate about whether specific or non-specific peroxidation plays a role in ferroptosis continues because not all substances that scavenge radicals can stop ferroptosis. For example, the concentration of Ferrostatin-1 needed to stop ferroptosis is in a low micromolar range, whereas non-specific radical scavengers like N-acetyl cysteine or lipoic acid work in a non-physiological sub-millimolar to millimolar range (Liu Y. et al., 2020). It is becoming increasingly clear that the time of inhibitor administration matters for the outcome, but, unfortunately, no detailed time-course investigation of ferroptosis inhibition with any inhibitor was published to let us judge the actual time window of inhibitors, their protective effects, and their actual targets in the ferroptosis pathway.

The role of mitochondria in ferroptosis is still under debate since GPX4 inhibition-induced ferroptosis occurs in the cancer cell line with depleted mitochondria (Tait et al., 2013; Gao et al., 2019). However, the same study showed that mitochondrial depletion inhibits ferroptosis induced by erastin (Gao et al., 2019), meaning that mitochondria play a significant role in cysteine deprivation/glutathione depletion-induced cell death. Blocking the system Xc-, which leads to glutathione depletion, does not always trigger ferroptosis in cancer cells. Erastin, under specific conditions of low concentration (Huo et al., 2016; Sun et al., 2020) incubation time, and the level of VDAC1 expression (Huo et al., 2016), may cause apoptosis in cancer cells, but in developing neurons, it does cause non-apoptotic cell death. The glutathione depletion-induced cell death known for a while as “oxytosis” has been widely used to model oxidative stress in neurodegeneration; however, it is now classified as ferroptosis (Tang and Kang, 2023). Researchers are trying to understand how glutathione depletion, mitochondrial iron, and imbalance in ROS are connected. Recent studies suggest that blocking the cystine transporter leads to a buildup of glutamate, which undergoes the glutamine-glutamate-αKetoglutarate (αKG) conversion in the mitochondria to feed the TCA cycle. This process, called glutaminolysis, combined with cysteine/cystine starvation, exacerbates oxidative stress and disrupts iron balance (Gao et al., 2019). In several reports, exposure to glutamate, RSL3, and erastin was reported to cause a profound loss of mitochondrial membrane potential (Neitemeier et al., 2017). However, there is no consensus on ferroptosis induction due to changes in mitochondrial membrane potential.

Mitochondria play a crucial role in utilizing and storing iron. Ferrous iron from the cytoplasm can enter the inner mitochondrial membrane (IMM) through a pathway driven by the membrane’s electrical potential or assisted by ferritin, crossing via mitoferrin one or mitoferrin 2 (Figure 3). Once inside, this iron is primarily used for the synthesis of heme and iron-sulfur clusters (ISC), as well as for storage within the mitochondria. Mitochondrial ferritin is a specific protein dedicated to storing iron within mitochondria, contributing to the establishment of mitochondrial iron pools and ensuring iron homeostasis (Sun et al., 2022). The observation that overexpression of mitochondrial ferritin protects the SH-SY5Y neuroblastoma cells from erastin-induced ferroptosis supports the hypothesis that imbalance in iron homeostasis inside mitochondria is induced by erastin treatment (Wang et al., 2016). Heme synthesis begins in the mitochondria and then continues in the cytosol, with the final insertion of iron into the porphyrin taking place in the mitochondria. However, heme synthesis does not seem to play a role in ferroptosis. The role of heme oxygenase-1, an enzyme that breaks down heme, in the initiation and propagation of ferroptosis is still uncertain (see below section 4.4). However, malfunction in iron-sulfur cluster synthesis and removal from mitochondria likely has a particular significance for the iron imbalance in mitochondria observed during ferroptosis, as discussed below.

A class of porin channels, namely, voltage-dependent anion channels (VDACs) or mitochondrial porins, are the gatekeepers in the outer membrane of mitochondria (OMM), guiding the flow of mitochondrial metabolites in and out, thereby controlling the crosstalk between the mitochondria and the rest of the cell during oxidative stress (Tang D. et al., 2021). It has been observed that erastin targets VDAC2 specifically, and without VDAC2, ferroptosis does not happen (Xie et al., 2016). After being treated with erastin for 10 h, both VDAC2 and VDAC3 are no longer detectable because erastin activates an enzyme called NEDD4 E3 ubiquitin ligase, which clears VDAC2 and VDAC3 channels, making the cells resistant to erastin (Yang et al., 2020). What is so special about VDAC2 out of the three VDACs that consider this specific isoform a direct player in the ferroptosis mechanism?

Humans encode three VDACs: for VDAC1, the crystal structure has been resolved; for VDAC2, the NMR structure data were collected, and it was concluded that its structure closely resembles that of Zebrafish VDAC2 (Eddy et al., 2019). VDACs are established cell death targets in PD and regulate apoptosis (He et al., 2022). Glutamate interacts with VDAC1 in vitro (Gincel and Shoshan-Barmatz, 2004) and, as recently demonstrated, causes mitochondrial fragmentation through its effect on VDAC oligomerization in a model of glutamate-induced glutathione depletion (Nagakannan et al., 2019). VDAC2 is unique in possessing nine cysteine residues that act like sensors on the inner side of OMM, detecting oxidative stress and changes in redox status (Supplementary Figure S1). When cells experience oxidative stress or glutathione depletion, VDAC2 can be chemically modified: it can be carbonylated by lipid-derived molecules (a novel cysteine site has been discovered) (Chen et al., 2018), glutathionylated (Reina et al., 2020), or phosphorylated by GSK3β kinase (activated by oxidative stress and translocated to mitochondria) (Tanno et al., 2014). GSK3β is considered a significant positive regulator of the mitochondrial permeability transition pore, whereas VDAC2 is an anti-apoptotic modulator (Heslop et al., 2021). Malonylation of VDAC2 at lysine 46 is a marker of mitochondria-related ferroptosis (She et al., 2023). Additionally, VDAC2 is implicated in autophagy through interaction with Beclin 1 (Kang et al., 2018). It is now agreed that VDAC2 is more of a sensor of mitochondrial ROS rather than an ion transporter (Naghdi and Hajnoczky, 2016). However, how VDAC2’s sensing ability is strictly connected to iron imbalance in the cell and mitochondria is still unclear.

Concerning recent publications on VDAC1 (Lipper et al., 2019) and VDAC2/3 interactions (Wan et al., 2015) with mitoNEET (encoded by CDIS1 gene) and direct evidence of the mitoNEET’s role in ferroptosis (Yuan et al., 2016), we hypothesize that the disruption in the mechanism of iron-sulfur cluster transfer from mitochondria could be linked to the iron homeostasis imbalance and subsequent accumulation of labile iron to drive ferroptosis. NEETs have emerged as new targets in ferroptosis (see subsection 4.3). Overall, understanding the role of VDAC2 and its interactions with other molecules could provide insights into the mechanisms of ferroptosis and iron homeostasis. However, many unanswered questions remain, including whether erastin directly interacts with VDAC2.

Iron is the most essential microelement, and iron-containing proteins contribute 5% towards all proteins in the organism. Iron plays a crucial role in numerous physiological functions within the brain, such as oxygen transport, mitochondrial respiration, DNA synthesis, and the synthesis and breakdown of neurotransmitters (Cao and Dixon, 2016; Jakaria et al., 2021). There are specific systems at the blood-brain barrier (BBB) and the blood-cerebrospinal fluid barrier to control how iron is distributed in the brain. Iron can enter the brain in two ways: transferrin receptor-mediated endocytosis and other transport systems for non-protein-bound iron (Viktorinova and Durfinova, 2021; Xu et al., 2023). For iron uptake, iron in the bloodstream binds to transferrin (TF), forming a complex with the transferrin receptor (TfR) on cell membranes, and is brought inside the cell through a process called endocytosis. Within the endosome, the six-transmembrane epithelial antigen of prostate 3 (STEAP3) assists in converting Fe3+ (ferric iron) to Fe2+ (ferrous iron), which is then released into the cell cytoplasm through divalent metal transporter 1 (DMT1) (Chen et al., 2020). The role of iron in ferroptosis has been extensively reviewed (Chen et al., 2020). While it is unclear whether the Fenton reaction or iron-containing enzymes are more critical in ferroptosis, an enzymatic reaction likely triggers the process. The LIP in cells, which includes iron not bound to proteins, is thought to play a crucial role in propagating the reactions that lead to lipid peroxidation, a hallmark of ferroptosis.

LIP describes the chemically reactive, kinetically exchangeable pool of iron used for iron cofactor synthesis, assembly, and insertion. The empirical evidence indicates that the LIP comprises iron in the ferrous state, coordinated by a polydisperse buffer system consisting of micro- and macro-molecules (Philpott et al., 2020). Cytosolic ferrous ion is coordinated by GSH forming the cytosolic LIP, with >90% of which is coordinated by iron chaperones, PCBPs that bind GSH coordinated ferrous ion with high affinity in a 1:1 ratio (Figure 4). PCBP1 and PCBP2 are the major iron chaperones that play integral roles in intracellular iron trafficking (see also Section 2.1 on LOX-15 translational repression by PCBP1). Unchaperoned free ferrous iron may react with hydrogen peroxide to generate hydroxyl radicals (Fenton reaction). While it is unclear whether the Fenton reaction or iron-containing enzymes are more critical in ferroptosis, an enzymatic reaction likely triggers the process.

Ferritin is an essential protein in the cytoplasm and the primary storage for ferrous ions (Reichert et al., 2020) through interaction with PCBPs. Ferritin is composed of ferritin light chain (FTL) and ferritin heavy chain 1 (FTH1), playing distinct roles in managing iron: FTH1 primarily converts Fe2+ to Fe3+, while FTL facilitates the entry of the next Fe2+ into the ferroxidase site (Yan HF. et al., 2021). Consequently, the amount of ferritin in a cell affects its sensitivity to ferroptosis: higher levels of ferritin means more iron is stored as Fe3+, making cells more resistant to ferroptosis because there’s less iron in the labile iron pool (Masaldan et al., 2019; Stockwell et al., 2020).

Accumulated evidence points to a link between autophagy and ferroptosis at the molecular level, as discussed in (Lee et al., 2023). A new autophagy process linked to ferroptosis is called ferritinophagy, which involves the selective removal of ferritin by a protein called nuclear receptor coactivator 4 (NCOA4). In ferritinophagy, NCOA4 recognizes ferritin and helps deliver it to the autophagosome, within which the ferritin undergoes degradation, releasing iron in the lysosome (Figure 4). When ferritinophagy is blocked by stopping autophagy or reducing the levels of NCOA4, the build-up of labile iron and ROS is prevented, ultimately preventing ferroptosis (Gao et al., 2016). NCOA4 has been identified as a potential target for developing drugs that inhibit ferroptosis by interfering with its interaction with ferritin (Figure 4). A new ferroptosis inhibitor, labeled 9a (IC50 of 300 nM), has been developed and tested in various models, showing promising results in reducing ferroptosis (Fang et al., 2021). It was about an order of magnitude less potent than ferrostatin-1 (IC50 = 40 nM) but much more effective than deferoxamine (IC50 = 3,860 nM) under the same experimental conditions in the cell models of erastin or RSL3-induced ferroptosis. In a rat model of ischemic stroke, compound 9a significantly reduced injury caused by lack of blood flow (ischemic reperfusion). This compound works by disrupting the interaction between NCOA4 and FTH1, a component of ferritin, thereby reducing the amount of available intracellular ferrous iron. Notably, NCOA4 is increased in low-oxygen conditions (hypoxia) and is regulated by hypoxia-inducible transcription factor HIF (Li et al., 2020a). However, there are some conflicting findings regarding the role of ferritinophagy in ferroptosis induced by erastin and RSL3. While compound 9a effectively inhibits ferroptosis (Fang et al., 2021), triggered by these ferroptosis inducers, a previous study suggested that ferritinophagy is elevated after erastin treatment but not influenced by RSL3 (Gryzik et al., 2021). In our viewpoint, ferritinophagy may not be the starting point of ferroptosis, but it is the final step before the process branches via LIP, and at this stage, the process can still be interrupted or stopped.

A process that connects mitochondrial iron and iron regulation in cells involves the transfer of iron-sulfur clusters (ISC) from mitochondria to iron regulatory protein 1 (IRP1) through mitoNEET and VDAC (Tam and Sweeney, 2023). Under normal conditions, IRP1 exists in its apo form with iron response element (IRE) binding ability. IRP1 binding to IRE within the 5′UTR results in translational repression (ferritin and ferroportin), whereas binding to IRE within the 3′UTR confers mRNA stability from RNase degradation (transferrin receptor). Under high iron resulting in oxidative stress, such as in pathological conditions, mitoNEET can transfer its ISC to IRP1, which loses its IRE binding ability. As a result, the translation of ferritin and ferroportin is upregulated due to a loss of IRP1 repression, whereas the transferrin receptor is downregulated due to a loss of mRNA stability.

IRP2 was identified as a critical ferroptosis regulatory gene identified using a high-throughput shRNA screening in the seminal work by Dixon et al (Dixon et al., 2012). Unlike IRP1, IRP2 lacks ISC and is primarily regulated by alterations in protein stability, which is sensitive to hypoxia and heme-induced oxidation. However, recent studies have shown that IRP2 can regulate iron levels and sensitivity to ferroptosis independently of IRP1 by sensing ISC cluster deficiency (Terzi et al., 2021). When the assembly of ISC proteins in the cytosol is inhibited, IRP2 stability and mRNA binding activity are increased. This suggests that triggering ferroptosis by restricting cysteine availability and thus limiting sulfur (from cysteine) for ISC synthesis would also mimic a response to iron starvation by increasing the mRNA binding activity of IRP1 and IRP2. However, the mRNA binding activity of neither IRP1 nor IRP2 has been characterized following the treatment with traditional ferroptosis-inducing agents like erastin or RSL3. Notably, in the context of brain function, IRP2 plays a pivotal role in regulating iron levels. Dysregulation of IRP2 is closely linked to the accumulation of iron observed in neurodegenerative conditions (Zhou et al., 2020).

Neuroinflammation is a common etiopathogenic feature in various neurodegenerative diseases, including Alzheimer’s disease (AD), Huntington’s disease (HD), Parkinson’s disease (PD), stroke, and amyotrophic lateral sclerosis (ALS). It eventually leads to different types of regulated cell death, such as apoptosis, necroptosis, pyroptosis, and ferroptosis (Moujalled et al., 2021). Disruption of iron homeostasis is associated with neuroinflammation, indicating its role in ferroptosis. The activation of inflammatory signaling pathways is closely related to the occurrence of ferroptosis, and vice versa (Chen et al., 2023a): a) activated Janus kinase signal transducer and activator of transcription (JAK-STAT) pathway results in activated STAT1, which inhibits system xC-; b) NF-kB signaling is known to promote ferroptosis in tumors and inhibit Nrf2 pathway (Gao et al., 2021), however, through TNF-α, it activates transcription of miR-155, which binds Bach1 mRNA and hinders its translation, and thus activates HMOX-1 expression (Pulkkinen et al., 2011); c) in cGAS-STING pathway, STING, an essential element in the type I interferon response and verified inducer of ferroptosis, was recently shown to interact with NCOA4 to exacerbate ferritinophagy (Wu J. et al., 2022); d) in mitogen-activated protein kinase (MAPK) pathway, p38MAPK and extracellular-signal regulated kinase (ERK) mediate both inflammatory response and ferroptosis (Wang X. et al., 2023).

Deposition of aggregated proteins/peptides is a characteristic hallmark feature of chronic neurodegenerative diseases. Such aggregates can adsorb iron and acquire catalytic properties, producing ROS and promoting ferroptosis, among other types of cell death. Specific evidence for the role of ferroptosis in a particular disease setting is difficult to establish since neurodegenerative diseases exhibit multiple standard features, such as dysregulated iron homeostasis, oxidative stress, and diminished glutathione. Perhaps the best evidence comes from the protection afforded by specific ferroptosis inhibitors in preclinical disease models and, ultimately, their use in human clinical trials (see section 4.5).

Several studies have highlighted a connection between ferroptosis and AD in the past decade (Ma et al., 2022). AD is characterized by chronic neuroinflammation, amyloid beta-protein deposition, and the hyperphosphorylation of tau protein. AD is considered a multifactorial disease with various etiologies and complicated pathophysiological processes. Previous research identified neuroinflammation and typical microglial activation as significant mechanisms underlying AD (Leng and Edison, 2021; Woodburn et al., 2021). Abnormally activated microglia release a panoply of pro-inflammatory factors that aggravate the dysregulation of iron homeostasis and neuroinflammation, forming a vicious cycle (Wang M. et al., 2023).

Both biochemical and morphological features of ferroptosis have been documented in postmortem brains of AD patients, including depletion of GSH, inactivation of GPX4, and imbalance in iron metabolism, resulting in lipid peroxidation, mitochondrial dysfunction, and ROS generation (Zhao et al., 2023). In AD, an accumulation of amyloid-beta (Aβ) and tau tangles may trigger ferroptosis by disrupting cellular iron balance and inducing oxidative stress (Jakaria et al., 2021). The involvement of iron in AD pathology is proven by the discovery of iron and ferritin around the glial cell in the senile plaques and neurofibrillary tangles. Synchrotron-based X-ray fluorescence microscopy detected abnormal iron enrichment in procured Aβ amyloid plaques from AD postmortem brain tissues (Huang, 2023). Iron interacts with amyloid-beta (Aβ) and tau by forming deposits and creating a peptide hemin complex in vitro (Pal and Dey, 2023). The Aβ- Fe³⁺ redox interaction mechanism has been validated in Aβ amyloid plaque cores from AD postmortem and transgenic mouse brain tissues (Huang, 2023). This interaction contributes to generating ROS, potentially playing a role in the ferroptotic cell death pathway (Lane et al., 2018).

An excessive build-up of iron in senile plaques and neurofibrillary tangles was found in various brain regions, particularly in the hippocampus. However, it remains uncertain whether this iron accumulation in these brain areas is linked to malfunctions in proteins regulating either the entry of iron into cells through DMT1 (an iron importer) or the removal of iron from cells via ferroportin (an iron exporter) (Crespo et al., 2014). An upregulation in ferritin FTH and FTL expression has been noted in AD brains, indicating a potential increase in LIP levels. Additionally, compensatory ceruloplasmin elevation is observed in AD, aiding in oxidizing Fe2+ to Fe3+ to facilitate FPN-mediated iron export. However, abnormal downregulation of FPN levels in AD reduces iron export, resulting in elevated LIPs within cells and triggering lipid peroxidation, a marker of ferroptosis (Ashraf et al., 2020; Zhou et al., 2020).

The amyloid precursor protein (APP) regulates neuronal iron content by promoting iron export through stabilizing FPN. On the other hand, iron is an inductor for furin, which reduces the activation of α-secretase, which leads to the activation of β-secretase and contributes to amyloid plaque production (Hwang et al., 2006; Altamura and Muckenthaler, 2009). Tau, a microtubule-associated protein, assists in transporting APP to the cell membrane and stabilizing the FPN1-APP complex. Therefore, tau inactivation affects APP transport to the cell membrane, leading to iron accumulation and oxidative stress-induced cell death, potentially linked to ferroptosis in AD (Belaidi et al., 2018; Dlouhy et al., 2019; Zhou et al., 2020). In both human AD and APP-transgenic mice brains, notably decreased levels of hepcidin and ferroportin were observed in hippocampal lysates (Raha et al., 2013).

Ayton et al suggested that iron levels in the brain could predict AD progression, especially in individuals carrying the apolipoprotein E (APOE) gene, notably the APOE-e4 risk allele. Research on APOE4 carriers suggests that this genetic variant may induce iron deposition in AD brains by reducing the delivery of high-density lipoprotein, a significant regulator of intracellular iron levels. Ayton’s results have revealed that ferritin is closely associated with the level of apolipoprotein E, and the APOE-e4 allele elevates the risk of AD (Ayton et al., 2015).

A large body of research has shown extensive mitochondrial abnormalities in AD (Wang et al., 2020). Mitochondria play multiple regulatory roles in ferroptosis, and preserving mitochondrial integrity is considered an effective strategy to prevent ferroptosis (Wu et al., 2021). A recent study by Li et al demonstrated an intriguing connection between mitophagy-dependent ferroptosis through the CD36/PINK1/Parkin pathway leading to blood-brain barrier (BBB) destruction in AD (Li et al., 2022). They showed that Aβ1-40 aggravated BBB damage by binding to CD36, a phagocytic receptor in the pericytes, a vital component of the neurovascular unit and the BBB. They observed that Aβ1-40 accumulation resulted in ferroptosis accompanied by a rise in Fe2+, increased lipid ROS production, decreased GSH-Px, and inhibited GPx4 and XcT in the pericytes. Interestingly, Aβ1-40 caused autophagy-dependent ferroptosis mediated through the PINK1/Parkin pathway, which was blocked by Mdiv-1, a putative inhibitor of mitochondrial fission protein dynamin-related protein 1 (Drp1), which causes excessive mitochondrial fragmentation. However, genetic deletion of CD36 expression alleviated Aβ1-40 mediated increased BBB permeability and autophagy-dependent ferroptosis. These data suggest a pathogenic role of mitochondrial autophagy through the CD36/PINK1/Parkin pathway in ferroptosis in AD.

To summarize, disturbances in iron balance, iron regulatory proteins, lipid peroxidation, and defective mitochondrial autophagy collectively suggest potential implications of ferroptosis in AD.

Parkinson’s disease (PD) is a common late-onset progressive neurodegenerative disorder characterized by loss of dopaminergic (DA) neurons in the substantia nigra pars compacta (SNpc) and the presence of fibrillar cytoplasmic inclusions composed of alpha-synuclein (αS) called Lewy Bodies (Thomas and Beal, 2007), as well as iron deposition (Altshuler and Hyde, 1985). Although the pathological mechanisms underlying PD are not fully understood, the generation of lipid peroxides, defects in the antioxidant system, and iron dysregulation are implicated in ferroptosis. A recent study exploring the DNA methylation patterns in a cohort of 1,132 individuals with PD and 999 controls identified a correlation between increased methylation in the promoter region of the SLC7A11 gene, responsible for encoding the cysteine-glutamate antiporter system Xc-, and the risk of PD. This aberrant hypermethylation in SLC7A11 reduces system Xc-expression, potentially causing the decline in intracellular glutathione (GSH) levels in PD cases and potentially augmenting susceptibility to ferroptosis (Vallerga et al., 2020).

The DA neurons are particularly susceptible to PD, and ferroptosis is implicated in their selective vulnerability due to iron-induced oxidative stress. Chronic inflammation in the SN accompanies conditions with progressive neurodegeneration by gradually increasing iron deposition (Thomsen et al., 2015). Microglial iron accumulation is involved in the occurrence and development of PD, and inhibition of microglial iron deposition prevents neuroinflammation (Liu et al., 2022). The SNpc and brainstem of PD patients have iron deposition, and the severity of the disease is related directly to the iron concentration (Mahoney-Sanchez et al., 2021; Lin et al., 2022). Consistent with these findings, a study that combined MRI imaging, quantitative susceptibility mapping, and regional gene profiling in a cohort of PD patients and control subjects showed a significant increase in iron accumulation in PD compared to controls. Genes related to heavy metal detoxification were differentially expressed in distinct brain cell types of PD patients, suggesting regional and selective vulnerability to iron accumulation in PD (Thomas et al., 2021). Elevated levels of DMT1, observed in the SNpc of PD patients and various PD mouse models, likely contribute to increased cellular iron intake (Salazar et al., 2008; Bi et al., 2020). Moreover, reduced levels of ferritin in the SN region have been reported in post-mortem brains of individuals with PD (Bi et al., 2020). Although this evidence suggests a significant role of iron imbalance and ferroptosis in mediating neurodegeneration in PD, how these contribute to each other and lead to neurodegeneration is still elusive. Nevertheless, given the vital link between iron accumulation and PD pathogenesis, iron chelation therapy offers the chance to provide a critical change in the PD treatment paradigm. However, a randomized study of iron chelation with deferiprone (FAIRPARK-II trial) without any DA replacement treatments unexpectedly showed worse outcomes for patients with PD (Devos et al., 2022). These results warrant further research into the biological process of iron accumulation and associated neurodegenerative mechanisms in PD pathogenesis.

Besides being a major component of Lewy body pathology, both genetic and biochemical abnormalities of αS are associated with PD pathogenesis. However, neither αS’s physiological function nor neuropathological mechanisms are entirely understood to date. It has been shown that αS binds to iron in Fe2+ and Fe3+ states, with the Fe3+ form being responsible for enhanced αS fibril formation and aggregation (Uversky et al., 2001). Iron accumulates in the SN of the midbrain, and postmortem analysis from PD patients shows the colocalization of iron and αS in Lewy bodies (Castellani et al., 2000). Further, αS mRNA contains an iron response element (IRE) within its 5′UTR, indicating that its translation is responsive to iron metabolism (Chen et al., 2019). Iron accumulation also results in lysosomal dysfunction, an organelle responsible for the clearance of αS aggregates, by inhibiting autophagosome-lysosome fusion and the activity of lysosomal hydrolases (Wildburger et al., 2020). These findings indicate a crucial connection between iron metabolism and αS aggregation in PD pathogenesis.

αS is also known to determine the sensitivity of DA neurons to ferroptosis by generating lipid peroxides. A recent report by Angelova et al showed that exogenous αS oligomers bind to the plasma membrane to drive ferroptosis by generating lipid peroxides, which can be rescued by iron chelators, D-PUFAs, or ferrostatin-1 (Angelova et al., 2020). The significance of lipids in PD is further highlighted by data showing the impact of αS on lipid metabolism, modulation of αS by lipids, and the identification of genetic determinants involved in lipid homeostasis associated with synucleinopathies (Alza et al., 2019). Consistent with these findings, endogenous αS plays a pivotal role in the survival of DA neurons by regulating phospholipid membrane composition, specifically, the ether-linked phospholipids essential for ferroptosis. Reduction of αS expression in human DA neuronal models markedly decreased the proportion of ether-phospholipids in the plasma membrane to a comparable level as when the primary ferroptotic regulator ACSL4 was reduced. Conversely, elevated levels of αS expression rendered DA neurons more vulnerable to ferroptosis-induced lipid peroxidation and cell death (Mahoney-Sanchez et al., 2022). These results directly link an established lipid peroxidation pathway essential for ferroptosis and the implicated role of αS in lipid metabolism. However, downstream pathways whereby lipid peroxidation leads to DA neuronal dysfunction and death are not fully understood.

In summary, a significant correlation exists between ferroptosis and PD pathophysiological features, such as abnormal iron metabolism, increased lipid peroxidation, decline in intracellular GSH, and downregulation of the cystine-glutamate antiporter system (Xc-), and αS aggregation. Further understanding the intricate relationship between these pathophysiological processes driving PD pathophysiology may pave the way for a better understanding of ferroptosis in PD.

In addition to the GPX4-dependent system, ferroptosis is also regulated by three other systems related to oxidation-reduction processes (Liu et al., 2023). Two systems (Figure 5), NAD(P)H quinone oxidoreductase 1 (NQO1) and ferroptosis suppressor protein 1 (FSP1), catalyze the reduction of coenzyme Q10 (CoQ₁₀) (Bersuker et al., 2019; Doll et al., 2019) to work alongside the glutathione-GPX4 pathway and prevent ferroptosis. FSP1, previously known as apoptosis-inducing factor mitochondrial 2 (AIFM2, also recognized as AMID or PRG3), does not promote cell death (Novo et al., 2021) but functions as a NAD(P)H-dependent flavo-oxidoreductase. It protects cells from ferroptosis by catalyzing the reduction of CoQ₁₀, vitamin K, and α-tocopheryl radical, which terminate the chain reaction of lipid peroxidation by neutralizing harmful lipid peroxy- and oxyradicals (Mishima et al., 2022; Li et al., 2023). Recent research has identified FSP1 as a gene targeted by nuclear factor erythroid-derived 2-related factor 2 (Nrf2) (Mishima et al., 2022). Another protective pathway is associated with the catalytic activity of dihydroorotate dehydrogenase (DHODH), which defends against both hypoxia (Spinelli et al., 2021) and ferroptosis (Mao et al., 2021) by reducing CoQ₁₀ (Figure 5).

The third protective system against ferroptosis is Guanosine triphosphate cyclohydrolase 1 (GCH1)-tetrahydrobiopterin (BH₄) (Kraft et al., 2020). Mechanistically, BH4 acts as a reducer, neutralizing harmful lipid oxy- and peroxy-radicals. When BH4 is oxidized to BH2, it can be restored to its active form, BH4, through reduction by NADPH with the assistance of dihydrofolate dehydrogenase. Additionally, BH4 is a crucial cofactor for nitric oxide synthases (NOS). In BH4-deficient conditions, superoxide is generated instead of nitric oxide (NO) (Takimoto et al., 2005), contributing to ferroptosis. GCH1 is a rate-limiting enzyme driving BH4 synthesis. Mutations in GCH1 are associated with a higher risk of dopa-responsive dystonia and PD (Yoshino et al., 2018).

Mammalian cells rely on two main pathways to synthesize GSH: importing cystine through system Xc- and synthesizing cysteine via the reverse transsulfuration pathway, the sole cysteine biosynthesis route. Cystathionine β-synthase (CBS), a gene regulated by Nrf2, catalyzes the condensation of homocysteine and serine to form cystathionine, which is then converted to cysteine by cystathione γ-lyase (CSE). Both the enzymes CBS (Liu N. et al., 2020) and CSE (Jamaluddin et al., 2022) are Nrf2 target genes. Disruption of the transsulfuration pathway is linked to several neurodegenerative diseases, including AD (Paul, 2021), PD (Corona-Trejo et al., 2023), and Huntington’s disease (Liu N. et al., 2020).

Iron chelators are thought to safeguard against ferroptosis by hindering lipid peroxidation fueled by the labile iron pool. However, it is tempting to speculate that iron chelators inhibit iron-dependent enzymes that contribute to ferroptosis initiation and progression. One of the potential candidates other than lipoxygenases could be αKG-dependent non-heme iron dioxygenases. These enzymes are essential in regulating the stability and properties of various transcription factors and in demethylating histones and DNA. Inhibition of these enzymes is observed in hypoxic conditions and when iron chelators are present. The effect of hypoxia on ferroptosis varies greatly depending on the context and cell type: hypoxia typically inhibits ferroptosis in cancer cells, whereas, in normal cells, it often triggers or enhances ferroptosis (Zheng et al., 2023). Hypoxia leads to the accumulation of HIF, which activates the expression of dozens of pro-survival genes, including those involved in iron regulation and protection against ferroptosis. HIF can be stabilized in the presence of iron chelators or inhibitors of HIF prolyl dehydrogenase (HIF PHD), which are αKG-dependent non-heme iron dioxygenase, such as adaptaquin (Karuppagounder et al., 2016). HIF PHDs exist in three forms and have multiple protein targets. A study showed that silencing PHD1, not PHD2 or PHD3, prevented ferroptotic death in HT22 cells and primary neurons (Siddiq et al., 2009). Transcriptomic analysis revealed that HIF PHD inhibition unexpectedly downregulated ATF4-dependent genes, including Trib3, Chac1, and Chop (Karuppagounder et al., 2016), all linked to ferroptosis. ATF4, like HIF, is a substrate for HIF prolyl hydroxylases, and mutant ATF4 lacking prolines, which are hydroxylated by the enzyme, blocked ferroptosis, suggesting a dominant-negative effect. Despite the established connection between ATF4 and ferroptosis, there is a lack of clear mechanistic justification for using HIF prolyl hydroxylase inhibitors to regulate the initiation and progression of ferroptosis. The commercial HIF PHD inhibitor, Roxadustat, has demonstrated contradictory effects: inducing ferroptosis in the glioblastoma (Su et al., 2022) while protecting against folic acid-induced kidney injury by reducing ferroptosis through Akt/GSK-3β-mediated Nrf2 activation (Li et al., 2020b). These conflicting findings underscore the need for further research on the involvement of HIF prolyl hydroxylase engagement in the ferroptosis pathway.

There are three proteins (Mittler et al., 2019) with a CDGSH iron-sulfur domain encoded by CDIS genes where mitoNEET is encoded by CISD1, Miner one (or NAF-1) by CISD2, and Miner two (or miNT) by CISD3. Miner two is only found in mitochondria, exists as a monomeric unit, and can fit through VDAC1, a channel in the outer membrane of mitochondria, much like VDAC2 (shown in Supplementary Figure S1). MitoNEET and NAF-1 can exist as dimers in the outer mitochondrial membrane (OMM) and endoplasmic reticulum (ER), respectively. CISD1 was identified among ferroptosis-related genes involved in asthma (Wang H. et al., 2023), and psoriasis (Wu MN. et al., 2022). Inhibition of CISD1 reduced LPS-induced ferroptosis in vivo and in vitro by preventing mitochondrial membrane potential depolarization, cellular ATP reduction, and ROS accumulation (Zhang et al., 2024). CISD1-Dependent regulation of ferroptosis is shown in glutathione depletion–induced oxidative cardiomyopathy in murine hearts (Li et al., 2024). CISD1 expression has been recently demonstrated as a ferroptosis hub gene of prognostic relevance for triple-negative breast cancer (Yuan et al., 2023) and tumor immune filtration in gastric cancer (Zang et al., 2023). NAF-1 (CISD2) is localized to the outer membrane of mitochondria, ER membrane, and mitochondrion-associated ER membranes. NAF-1 is an NF-κB antagonist (Kung and Lin, 2021) and exerts anti-inflammatory effects (Lin, 2020). NAF-1 deficiency causes accelerated aging and shortened lifespan (Chen et al., 2009), whereas persistent expression of CISD2 promotes longevity in mice (Wu et al., 2012).

Heme oxygenase-1 (HMOX-1) is an inducible enzyme known for its anti-inflammatory, antioxidant, and neuroprotective effects (Ryter, 2022). However, increased expression of HMOX1 during aging and age-related neurodegenerative diseases has been associated with neurotoxic ferric iron deposits. HMOX-1 is a primary intracellular source of iron and, thus, may play a pro-death role in ferroptosis via enhancing iron release (Ryter, 2021).

Ferrostatin’s ability to combat ferroptosis is primarily attributed to its radical neutralizing capacity to eliminate initial alkoxyl radicals formed in non-enzymatic ferrous ion-dependent lipid peroxidation reaction (Miotto et al., 2020). Although not yet approved for clinical use, ferrostatin and related compounds are under investigation for their potential to mitigate the progression of neurodegenerative diseases and other conditions associated with ferroptosis. (Skouta et al., 2014; Anthonymuthu et al., 2021). Ferrostatin may provide neuroprotection by preventing ferroptosis-induced neuronal death, potentially slowing down the advancement of the disease. In preclinical studies, ferrostatin and similar compounds have demonstrated efficacy in AD, PD (Wang Y. et al., 2023), and intracerebral hemorrhage models (Li et al., 2017). However, a derivative of ferrostatin (with an additional N-benzyl substitution at the ferrostatin-1 ortho-phenylene-diamine group) that protects against Friedreich’s ataxia (Cotticelli et al., 2019) and cerebral ischemia/reperfusion seemingly operates by activating Nrf2 signaling (Chen et al., 2023b). In the spinal cord injury model, ferrostatin-1 also works through the activation of Nrf2 (Zhou et al., 2023). In general, the ortho-phenylene-diamine scaffold in ferrostatin, when oxidized in the presence of metal traces, generates an oxidant (substituted 2,3-diamino phenazine) that could drive activation of Nrf2 by chemically modifying its inhibitor protein, Kelch-like ECH-associated protein 1 (Keap-1) (see section 4.6 below).

Protection against ferroptosis is strongly associated with the genes targeted by Nrf2, a key transcription factor that plays a pivotal role in maintaining cellular redox homeostasis. Nrf2 belongs to the CNC-bZIP family and orchestrates the expression of antioxidant and cytoprotective genes (Shakya et al., 2023). The Nrf2-Keap1 pathway is a strategic coordinator in the cellular defense against ferroptosis. This pathway effectively detects changes in cellular oxidative stress and triggers the expression of a wide range of Nrf2-regulated genes to counteract the initial stressor of ferroptosis. When exposed to oxidative or electrophilic stress, cysteine residues in Keap1 are modified, leading to its inactivation and subsequent Nrf2 accumulation. Nrf2 then translocate to the nucleus and binds to the antioxidant response element (ARE), promoting the expression of a gene battery that combats oxidative stress and cell death (Anandhan et al., 2020; Lane et al., 2021; Lane et al., 2023). Activation of Nrf2 enhances antioxidant defense mechanisms and modulates iron homeostasis directly. For instance, by increasing ferritin levels, a protein that stores iron, Nrf2 can sequester excess iron and mitigate its involvement in lipid peroxidation. This dual function of Nrf2, safeguarding against oxidative stress and iron-induced ferroptosis, underscores its potential as a therapeutic target (Dodson et al., 2019; Lane et al., 2023).

Conversely, Bach1, or BTB and CNC homology 1, a heme-responsive transcription factor, functions as a negative regulator of Nrf2 by competing for binding to the ARE. Bach1 and Nrf2 jointly regulate the availability of iron for specific cellular proteins. When Bach1 is highly active, or Nrf2 is less active, there is an excess of unstable iron, rendering cells more susceptible to ferroptosis (Nishizawa et al., 2020). This intricate balance between Nrf2 and Bach1 is crucial in cellular defenses against ferroptosis (Namgaladze et al., 2022; Nishizawa et al., 2023). Interventions, either through drugs or genetic approaches that inhibit Bach1’s function, can release the restraints on Nrf2, amplifying its protective responses. Combining Nrf2 activation with inhibition of Bach1 holds the potential for synergistic neuroprotection against ferroptosis.

Chromatin Immunoprecipitation (ChIP)-sequencing analysis of Bach1 and Nrf2 genes, which overlap with critical players in ferroptosis, reveals 22 common genes, as depicted in (Figure 6). These genes are involved in various pathways, as outlined in (Table 1). In addition to 22 common genes, 10 genes bound by Nrf2 (but not Bach1) are implicated in ferroptosis, along with 45 genes bound by Bach1 (but not Nrf2). In other words, Bach1 binds to more genes related to ferroptosis than Nrf2 does. Therefore, Bach1 is just as crucial for the ferroptosis pathway as Nrf2. Genes regulated by Nrf2 are vital molecular players in suppressing and preventing ferroptosis. At the same time, Bach1 interferes with Nrf2 binding to several gene targets (Figure 7). This interference extends to genes involved in glutathione synthesis enzymes (Gclm and Gclc), selenocysteine enzyme thioredoxin reductase 1 (Txnrd1), and proteins and enzymes related to iron homeostasis, such as heme oxygenase-1 (HMOX-1/POR) (Figure 7; Table 1). These genes are essential for combating lipid peroxidation and preventing ferroptosis (Lane et al., 2021; Yan et al., 2023). Mitochondrial aconitase (ACO2), the target for both Nrf2 and Bach1, serves as a redox sensor in the tricarboxylic acid cycle due to the high sensitivity of its active site ISC towards ROS. ACO2 is particularly susceptible to oxidation during aging and mitochondrial dysfunction. It is vital to replenish this enzyme, as it plays a pivotal role in maintaining the performance of the TCA cycle.

Another set of targets shared by Nrf2 and Bach1 is associated with lipid synthesis/storage and inflammatory reactions catalyzed by heme-containing enzymes. Neuroinflammation is a common feature of many neurodegenerative diseases, and activating Nrf2 has been shown to alleviate oxidative stress and neuroinflammation in mouse models of AD and PD. Therefore, combining Nrf2 activation with Bach1 inhibition represents a promising strategy to inhibit the inflammatory cascade and shield neurons from the detrimental effects of chronic inflammation.

The endosomal system is indispensable for normal brain development and function. Two notable endosomal targets relevant to insulin resistance and type 2 diabetes are the Zfyve28 and Exoc6 genes (as listed in Table 1). It is well-known that brain insulin resistance plays an essential role in AD development and progression (Sedzikowska and Szablewski, 2021). Regulation of Zfyve28 by Notch is well-documented (Yu et al., 2023), but its plausible link to Nrf2/Bach1 (as indicated in Table 1) is novel. Nrf2 and Bach1 share common targets in endo/exocytosis, neurogenesis, and protein degradation pathways (such as the TET1 enzyme, ubiquitin b, and c, as shown in Table 1). However, the interdependence of Nrf2/Bach1 effects on these targets in the context of neurodegenerative diseases needs to be further explored. Notably, Nrf2 and Bach1/2 transcription factors depend on each other and are associated with ferroptosis (as indicated in Table 1). There are no published studies regarding Bach1 and Bach2’s involvement in mediating ferroptosis in neurodegenerative diseases. Considering the data in Table 1, further research into the role of Bach1 and Bach2 in ferroptosis in neurodegenerative diseases of various etiology is warranted.