Ferroptosis, a newly identified iron-dependent cell death, is characterized by iron accumulation and LPO, and has different morphological, biochemical and genetic characteristics from other forms of programmed cell death. Ultrastructural analysis showed the key morphological features distinguishing ferroptosis from apoptosis, necrosis and autophagy include cell membrane rupture and blister, decreased mitochondrial size relative to normal, mitochondrial ridge reduction or disappearance, and increased membrane density [9].

LPO is a key factor inducing ferroptosis, with polyunsaturated fatty acids (PUFAs) generally serving as reaction substrates. Carbon-carbon double bonds and carbon-hydrogen bonds in PUFAs are transformed to generate PUFAs-CoA by acyl-CoA synthetase long chain family member 4 (ACSL4), and PUFAs-coenzyme A (PUFAs-CoA) are then esterified by lysophosphatidylcholine acyltransferase 3 (LPCAT3) into substrate PUFAs-PL in membrane phospholipids. Therefore, ACSL4 is considered a ferroptosis marker [10]. Free radicals interact with the hydrogen of PUFAs-PL to form the phospholipid radical (PL$\cdot$), which then reacts with oxygen to generate phospholipid peroxyl radical (PL-OO$\cdot$). Hydrogen is extracted from other PUFAs to generate lipid peroxides (PL-OOH) and PL$\cdot$, and so on. However, excessive Fe${}^{2+}$ catalyzes the reduction of PL-OOH to PL-O through the Fenton reaction, which further promotes non-enzymatic free radical chain reaction, ultimately altering cell membrane integrity and inducing ferroptosis [11].

LPO occurs in many subcellular structures such as mitochondria, endoplasmic reticulum (ER) and lysosomes [12]. Reactive oxygen species (ROS) generated by the tricarboxylic acid (TCA) cycle and electron transport chain (ETC) in the mitochondria, can target PUFAs-double bonds to produce LPO and induce ferroptosis [13]. Moreover, ER stress has a definite correlation with ferroptosis, and lysosomal ROS may also participate in ferroptosis as ferrostatin was observed in lysosomes [14, 15]. However, the specific mechanism remains undefined. Reactive lipid species generated by LPO include 4-hydroxynonenal (4-HNE) and malondialdehyde (MDA), whose contents are the highest and associated with ferroptosis [16].

Currently, three pathways are known to inhibit LPO to prevent ferroptosis, including the glutathione pathway (GSH-GPX4), the ubiquinone pathway (NADPH-FSP1-CoQ10) and the tetrahydrobiopterin pathway (GCH1-BH4) [17]. Firstly, the glutathione peroxidase 4 (GPX4) pathway is one of the commonest defense systems, which mainly depends on the Xc${}^{\mathrm{-}}$ transporter system, glutathione (GSH) synthesis and GPX4 activity. The Xc${}^{\mathrm{-}}$ system transports equal amounts of extracellular cystine and intracellular glutamic acid. Cystine then undergoes reduction by GSH to generate cysteine, thereby interfering with LPO reactions and preventing ferroptosis. Nuclear factor erythroid 2-related factor 2 (Nrf2) represents a major upstream regulatory gene of defense systems that can target the subunit solute carrier family 7 member 11 (SLC7A11), the light chain subunit of the Xc${}^{\mathrm{-}}$ system [18]. Studies have suggested that drugs may activate Nrf2 to initiate the GPX4 defense mechanism to mediate ferroptosis and reduce the occurrence and development of diseases, especially NAFLD [19, 20, 21]. This provides a promising strategy for developing drug candidates targeting Nrf2 for the treatment of NAFLD.

Secondly, ferroptosis suppressor protein 1 (FSP1) is a newly reported strong and effective inhibitor of ferroptosis. The FSP1 pathway does not depend on GSH for its inhibitory effect, but instead functions by regulating coenzyme Q10 (CoQ10) [22]. Acetylated FSP1 is targeted to the plasma membrane and CoQ10 stimulates NAD(P)H to shuttle into the lipid bilayer with the help of FSP1; then, free radicals in the droplets are regulated to capture antioxidants, maintaining the quality of lipids and preventing LPO from inducing ferroptosis [23]. Currently, relatively few studies have examined the BH4 pathway. BH4 participates in the antioxidant system and the overexpression of its rate-limiting enzyme GTP cyclohydrolase 1 (GCH1) selectively inhibits the peroxidation of some PUFAs. Therefore, ferroptosis can be inhibited through the GCH1-BH4 pathway [24, 25]. The major regulators of ferroptosis and associated mechanisms are summarized (Fig. 1).

Fig. 1.

Pathways of ferroptosis. Ferroptosis can be triggered through inhibiting the system Xc${}^{\mathrm{-}}$/GSH/GPX4 axis or the accumulation of Fe${}^{2+}$ and lipid peroxidation. Blocking or inhibiting the Xc${}^{\mathrm{-}}$/GSH/GPX4 axis can result in decreased intracellular cysteine levels and suppressing the lipid repair function of GPX4. Consequently, disruption of the antioxidant capacity of cells contributes to the initiation and progression of ferroptosis. In lipid metabolism, the most important characteristics of ferroptosis are the increase of intracellular iron ion concentration and the abnormal accumulation of lipid ROS. FSP1 can negatively regulate ferroptosis through CoQ10, thereby inhibiting the delivery of lipid peroxides. Ferritinophagy plays a crucial role in the association between ferritinophagy and ferroptosis. Intracellular ferritin is transported to autophagy lysosomes for degradation, leading to the release of free iron and ultimately triggering ferroptosis. Abbreviations: GPX4, glutathione peroxidase 4; PUFAs, polyunsaturated fatty acids; ACSL4, acyl-CoA synthetase long chain family member 4; LPCAT3, lysophosphatidylcholine acyltransferase 3; PL-OO$\cdot$, phospholipid peroxyl radical; TCA, tricarboxylic acid; GSH, glutathione; ROS, reactive oxygen species; FSP1, ferroptosis suppressor protein 1; CoQ10, coenzyme Q10; GCH1, GTP cyclohydrolase 1. ATP, adenosine triphosphate.

Necroptosis is a programmed form of cell death that can regulate necrosis with passive and active proinflammatory functions [26]. Numerous studies have consistently shown that ferroptosis is invariably accompanied by necroptosis [27]. The key steps involved in necroptosis include tumor necrosis factor (TNF) signaling, deubiquitination of receptor-interacting protein (RIP) 1, phosphorylation of RIP1 and RIP3, inactivation of caspase-8, and phosphorylation of mixed lineage kinase domain-like protein (MLKL) [28]. Ferroptosis can induce mitochondrial damage, resulting in the opening of the mitochondrial permeability transition pore. Subsequently, the phosphorylation of RIP1/3 was further intensified, ultimately culminating in necroptosis [29]. These findings demonstrate that there is a crosstalk between ferroptosis and necroptosis, but the regulatory relationship between necroptosis and ferroptosis still requires further study.

Autophagy refers to the process where the detached double membrane of the ribosome-free attachment zone of the rough endoplasmic reticulum envelops cytoplasm, organelles, proteins, and other cellular components that require degradation to form autophagosomes [30]. Subsequently, the autophagosomes fuse with lysosomes to form autophagolysosomes, wherein the enclosed contents are degraded to fulfill the metabolic requirements and facilitate organelle renewal. A growing body of evidence suggests that ferroptosis relies on the autophagy machinery for its execution, making it a form of autophagy-dependent cell death [31]. Recent studies have discovered that autophagy plays a role in promoting ferroptosis by degrading the ferroptosis-related protein, ferritin, through the autophagy-related gene (ATG) 5/ATG7-nuclear receptor coactivator 4 (NCOA4) autophagic pathway [32]. Moreover, ras-related protein Rab-7a (RAB7A)-mediated lipophagy, BECN1-mediated system Xc${}^{\mathrm{-}}$ inhibition, signal transducer and activator of transcription 3 (STAT3)-mediated lysosomal cell death, and sequestosome 1 (SQSTM1)-mediated clockophagy are involved in the regulation of ferroptosis [33]. All of these studies highlight the regulatory role of autophagy in ferroptosis, presenting a promising new treatment hypothesis by targeting autophagic ferroptosis.

Apoptosis is a well-established mode of programmed cell death involving both extracellular and intracellular pathways [34]. The extracellular pathway is triggered by receptors on the cell membrane, such as TNF-$\alpha$, while the intracellular pathway is primarily influenced by the permeability of the mitochondrial membrane and the regulation of the Bcl-2 protein family. Indeed, several studies have reported an interrelationship between ferroptosis and apoptosis that apoptosis can be converted to ferroptosis [35]. In addition, the ferroptosis inducer, erastin, activates the p53-dependent CHOP/PUMA pathway, enhancing sensitivity to apoptosis [36]. These findings indicate the existence of close link between ferroptosis and apoptosis.

Pyroptosis, also known as cell inflammatory necrosis, is a form of programmed cell death that relies on inflammatory caspases [37]. Pyroptosis is characterized by nuclear pyknosis, cell swelling, the formation of lipid membrane vacuoles at the plasma membrane, and eventual cell rupture without DNA fragmentation. Both pyroptosis and ferroptosis are associated with damage to the cytoplasmic membrane. Notably, ROS-mediated damage to the cytoplasmic membrane may contribute to reciprocal regulation between pyroptosis and ferroptosis. Furthermore, research has revealed that both types of cell death can be triggered by elevated levels of intracellular iron and ROS [38]. The aforementioned study further demonstrates that a synergistic relationship and crosstalk is existed between pyroptosis and ferroptosis, but the underlying mechanisms still need to be thoroughly investigated.

Iron participates in multiple cellular processes, including oxygen storage and transport, mitochondrial respiration, DNA replication, and intercellular signaling in the body [46]. Intracellular iron balance considers iron absorption, output, use and storage. Excess iron is mostly stored in ferritin, as a redox inactive form. Generally, Fe${}^{3+}$ is bound to transferrin and transferrin receptor 1 (Tfr1) in cells and stored in the form of Fe${}^{3+}$-ferritin complex [47]. Subsequently, hepcidin synthesized in the liver is important in maintaining iron homeostasis by controlling ferritin synthesis. A small quantity of Fe${}^{2+}$ generate a labile iron pool [48]. Iron metabolism disorders are mainly divided into two categories, including iron overload and deficiency. In case of iron overload in cells, Fe${}^{3+}$ is reduced to Fe${}^{2+}$ in lysosomes and released into the labile iron pool through the divalent metal transporter 1. Free Fe${}^{2+}$, with high oxidative potential, easily undergoes the Fenton reaction with lipid peroxides, produces hydroxyl radicals, induces oxidative stress, enhances ROS accumulation and induces ferroptosis [49]. Fe${}^{2+}$ is thus also considered a cofactor increasing the activities of multiple metabolic enzymes, promoting lipid ROS synthesis and enhancing ferroptosis. In addition, nuclear receptor coactivator 4 (NCOA4) can mediate ferritinophagy, which dissociates the Fe${}^{3+}$-ferritin complex to release Fe${}^{2+}$ into LIP [50, 51]. Iron metabolism disorder may therefore induce the production of large amounts of Fe${}^{2+}$, which is involved in the conversion of PUFAs through the LPO process, aggravating NAFLD, diabetes, IR and other diseases. In conclusion, iron is a crucial factor in ferroptosis, which requires iron metabolism.

Hepatocytes are critical for liver iron homeostasis. Iron shortage in hepatocytes and excessive iron accumulation in hepatic stellate cells mediated by Kupffer cells-induced secretion of iron-rich extracellular vesicles (EVs), can lead to NAFLD and NASH. Meanwhile, blocking EV secretion or EV iron transport can restore iron balance in the liver and alleviate NAFLD/NASH-associated hepatic steatosis and fibrosis. A previous study revealed that the lack of hepcidin inhibitor matriptase-2 may alleviate obesity, accelerate lipolysis, improve glucose tolerance and insulin sensitivity, and prevent hepatic steatosis in high-fat diet mice [52]. This suggests that regulation of iron homeostasis in the liver has a significant correlation with anti-steatosis.

Oxidative stress describes an imbalance between oxidative and antioxidant events, where the generation of free radicals exceeds the capacity of antioxidant defense systems. Iron overload promotes the production of ROS, thereby inducing oxidative stress. Furthermore, oxidative stress-related transcription factors downregulate or upregulate ferroptosis-related genes, which regulates ferroptosis [53, 54, 55]. The “multiple-hit” hypothesis considers oxidative stress one of the key mechanisms underlying NAFLD-related liver injury. When ROS are excessively produced through the external environment (e.g., physical and chemical factors) and the internal environment (e.g., mitochondrial dysfunction, ER stress and peroxidation), they target the double bonds of PUFAs, resulting in LPO and the formation of lipid peroxides, which induces ferroptosis and aggravates hepatocyte injury. Recent evidence suggests mitochondrial ROS play a role in NASH-related ferroptosis. Epigallocatechin gallate prevents liver lipotoxicity by suppressing mitochondrial ROS-related liver ferroptosis [41]. Quercetin reduces lipid accumulation, hepatic lipotoxicity and LPO in high-fat diet mice, ultimately inhibiting ferroptosis to alleviate NAFLD [56]. Besides, obeticholic acid reduced the amounts of triglyceride-containing PUFAs, while elevating free PUFAs and phosphatidylethanolamine-containing PUFAs, which are prone to oxidization to generate lipid peroxides, triggering ferroptosis in hepatocytes and causing hepatic stellate cell activation [57]. Taken together, both oxidative stress and iron overload can induce the LPO process, and even synergically worsen the degree of cell damage, resulting in NAFLD progression from simple hepatic steatosis to inflammation and then fibrosis.

GSH is the main antioxidant in mammalian cells, which is composed of glutamate, cysteine and glycine, and represents a cofactor of GPX4 [58]. However, GSH depletion disrupts the dynamic balance of intracellular redox activity, which leads to ROS accumulation and aggravates ferroptosis [59, 60]. Cysteine is the main component and thus the limiting amino acid in GSH biosynthesis; suppressing cysteine import via the cystine/glutamate antiporter SLC7A11 induces ferroptosis via GSH depletion [61]. Therefore, GSH is often used as an evaluation indicator of oxidative stress, especially in studies developing drug therapies for ferroptosis-induced NAFLD.

GPX4 was the first discovered selenium-containing protein inhibiting ferroptosis. It is mainly found in the cytoplasm and mitochondria, and plays a role in scavenging lipid hydrogen peroxide. GPX4 can convert toxic lipid hydroperoxides into corresponding nontoxic lipid alcohols, which forms a defense mechanism against LPO and plays an important regulatory role in ferroptosis [62]. In methionine and choline deficient L-amino acid diet (MCD)-fed NAFLD mice treated with the ferroptosis inducer (1S,3R)-RSL3, liver GPX4 was downregulated while it was upregulated by lipoxygenase and an apoptosis inducer. Subsequently, supplementation of the GPX4 activator sodium nitrite increased GPX4 expression and reduced the severity of NASH. These results indicated that GPX4 plays a significant role in ferroptosis-related changes in hepatocytes that cause NASH [63]. Therefore, GPX4 regulation may control the development of NAFLD inflammation before it progresses to NASH. NAFLD rats were treated with ferroptosis inducer (erastin) and inhibitor (ferrostatin 1 [Fer-1]), and steatosis in human liver cell 7702 (L02) was induced by palmitic acid (PA), respectively. Both in cultured cells and experimental animals, interfered GPX4 expression decreased the protective effects on rat liver and mitochondrial membrane integrity in L02 cells, and also regulated B-cell lymphoma-2 (Bcl-2), Bcl-2 associated x (Bax), Caspase-3 and superoxide dismutase 1 (SOD1), attenuating the treatment effect in NAFLD. This study revealed that suppressing GPX4-related ferroptosis may provide a novel approach for NAFLD treatment [64].

The Xc${}^{\mathrm{-}}$ system is known as the cystine/glutamate transporter on the cell membrane; its light chain subunit SLC7A11 is critical for the metabolism of lipid ROS. Recent data revealed p53 downregulates SLC7A11 to markedly affect the metabolism of lipid ROS in ferroptosis [65, 66]. High SLC7A11 expression promoted tumor growth partially by inhibiting ferroptosis. Nevertheless, malignant cells highly expressing SLC7A11 also endure the substantial burden imposed by SLC7A11-indcued metabolic reprogramming, resulting in enhanced dependence on glucose and glutamine, which represents possible metabolic vulnerabilities for target therapies in malignant diseases [67]. Sorafenib attenuates liver damage and extracellular matrix accumulation in CCl4-dependent liver fibrosis, with concomitant SLC7A11 and GPX4 downregulation at the protein level. SLC7A11 is upregulated by hypoxia inducible factor (HIF)-1$\alpha$, and HIF-1$\alpha$/SLC7A11 pathway suppression is essential in sorafenib-related ferroptosis in hepatic stellate cells, ultimately limiting antifibrotic activity [68]. Furthermore, BECN1 combined with SLC7A11 can directly block the activity of the Xc${}^{\mathrm{-}}$ system, indicating that SLC7A11 is a core component of the transporter [69]. As an inducer of Nrf2, SLC7A11 accelerates the transport of cystine and glutamate, promotes the synthesis of GSH, and thus achieves the purpose of suppressing ferroptosis [70, 71].

Nrf2 interacts with nuclear factors to activate downstream transcription, which regulates antioxidative events and detoxification [72]. Nrf2 regulates multiple biological events, e.g., the transcription of the antioxidant system, the detoxification of endogenous and exogenous substances, NADPH regeneration and heme metabolism, as well as many cell processes, including autophagy and unfolded protein response [73]. Mounting evidence suggests a regulatory role for Nrf2 in ferroptosis. Nrf2 constitutes an upstream effector of some essential ferroptosis-related genes such as GPX4 and Xc${}^{\mathrm{-}}$ system [74]. Consequently, Nrf2 negatively regulates ferroptosis and promotes ferroptosis resistance. After Kelch-like ECH-associated protein 1 (KEAP1) activates Nrf2 in oxidative stress, it recognizes antioxidant response elements and activates antioxidant genes, including SLC7A11. Furthermore, it upregulates antioxidant proteins associated with ferroptosis, e.g., heme oxygenase-1 (HO-1) and GPX4, and increases GSH levels [75, 76]. A recent study reported Nrf2 activation by iron-dependent, mitochondrial pro-oxidants, which upregulates bone morphogenetic protein 6 (Bmp6) in hepatic sinusoidal endothelial cells and increases the synthesis of hepcidin in adjacent hepatocytes. In Nrf2-deficient mice, impaired Bmp6-hepcidin axis was reported, causing iron accumulation and worsened liver damage. Meanwhile, pharmacological induction of Nrf2 resulted in activated Bmp6-hepcidin signaling, ameliorating iron homeostasis and countering Bmp6 suppression [77]. Overall, Nrf2 links cellular sensitivity to iron toxicity, systemic iron homeostasis and antioxidation, representing a potential target for treating iron-related diseases.

Besides, the Nrf2 contributes to reducing ROS overproduction to exert anti-inflammatory effects for NAFLD. Indeed, a small interfering RNA (siRNA) of Nrf2 or a glutathione peroxidase suppressor could abolish nuclear factor kappa-B (NF-$\kappa$B) repression and decrease the secretion of inflammatory cytokines [78]. As shown in Fig. 2, downstream effectors of Nrf2, such as HO-1, GSH and GPX4, were upregulated to reduce MDA, inhibit ROS accumulation, LPO and reduce high-fat diet-induced NAFLD [79].

Fig. 2.

KEAP1-Nrf2 pathway. When cells experience oxidative stress, the KEAP1-Nrf2 interaction is altered, preventing KEAP1 from targeting Nrf2 for degradation. As a result, Nrf2 is stabilized and accumulates in the cell. One of the key mechanisms by which Nrf2 contributes to protecting cells from ferroptosis is through the upregulation of genes involved in iron metabolism. Nrf2 induces the expression of GPX4, HO-1, and SLC7A11, releasing iron in a controlled manner. By regulating iron levels, Nrf2 helps prevent the accumulation of toxic levels of labile iron that can catalyze lipid peroxidation, a hallmark of ferroptosis. Abbreviations: KEAP1, Kelch-like ECH-associated protein 1; GPX4, glutathione peroxidase 4; HO-1, heme oxygenase-1; SLC7A11, subunit solute carrier family 7 member 11; Nrf2, nuclear factor erythroid 2-related factor 2.

In NAFLD, liver inflammation is a critical component of disease progression. FSP1 has been associated with the regulation of inflammatory responses in various cell types. Its involvement in immune cell migration and activation may influence the extent of liver inflammation in NAFLD [18]. FSP1 was initially identified as the homologous regulation factor of the mitochondrial pro-apoptotic protein apoptosis-inducing factor (AIF)/AIF mitochondrion-associated 1 (AIFM1), termed AIFM2. However, it was later confirmed that FSP1 lacks the N-terminal sequence for mitochondrial targeting, is not localized in mitochondria and does not promote apoptosis. It was then renamed by Bersuker et al. [23] as FSP1 and its cellular function was described. FSP1 is present in lipid droplets and cell membranes, and it acts as a suppressor of ferroptosis by reducing lipid peroxidation and protecting cells from oxidative damage. Mechanically, CoQ10 was identified as a major substrate using a ferroptosis activator [80]. FSP1 expression shows a positive correlation with resistance to GPX4 suppressors, and plays a critical role in maintaining cancer cell proliferation in case of no GPX4 expression. Notably, similar to FSP1, GCH1 was recently described as a new protein suppressing ferroptosis based on whole-genome activation screening [81]. GCH1 overexpression occurs in presence of all three ferroptosis activators, including RSL3, erastin and GPX4. High GCH1 amounts enhance BH4 and CoQ10 levels, synergistically acting with the FSP1 pathway to hamper the LPO process. GCH1 expression and resistance to ferroptosis are overtly correlated, suggesting the GCH1-BH4-phospholipid pathway as a potential therapeutic target in iron-related diseases.

Transferrin receptor 1 (Tfr1), or cluster of differentiation 71 (CD71), interacts with transferrin and exerts an important effect on iron uptake by hepatocytes in a specific manner [82]. In palmitate-induced IR, palmitate simultaneously upregulates Tfr1 and induces intracellular ferroptosis. Knockdown of Tfr1 prevented palmitate-induced iron uptake and IR, and also translocated Tfr1 via enhancement of calcium influx. However, supplementation of a calcium chelator substantially decreased ferroptosis by suppressing Tfr1 translocation, and ultimately enhanced insulin sensitivity [83]. It was also found that the interaction between complexes was neutralized by antibodies against Tfr1, making Tfr1 a potential target for ferroptosis alleviation in NAFLD [84].

Excess iron in the body is stored in ferritin that plays a critical role in iron metabolism and storage in cells. It is usually involved in ferritinophagy which releases Fe${}^{3+}$ and is subsequently converted to Fe${}^{2+}$ to the labile iron pool, participating in the process of cell ferroptosis. Eventually, it sequesters excess iron in a non-toxic and bioavailable form to prevent the free iron from participating in harmful Fenton reactions that generate highly reactive and toxic hydroxyl radicals [85]. Therefore, suppressing ferritinophagy may provide a novel approach and target for treating NAFLD [86]. Elevated Ferritin levels are commonly observed in patients with NAFLD. It is important to note that the relationship between Ferritin and NAFLD is complex and multifactorial, involving iron metabolism, oxidative stress, and inflammation. It was reported that prominin-2, a pentapeptide protein associated with lipid homeostasis, is induced by ferroptosis response and induces the generation of ferritin-containing multivesicular bodies (MVBs), which transports iron out of cells to suppress ferroptosis. This suggests the prominin-2/MVBs-ferritin pathway has certain significance in iron homeostasis and intracellular transport [87]. Therefore, further investigation is needed to study how to inhibit ferroptosis-induced hepatic steatosis from the perspective of iron metabolism.