Ferroptosis: Final destination for cancer?

Abstract Ferroptosis is a recently defined, non‐apoptotic, regulated cell death (RCD) process that comprises abnormal metabolism of cellular lipid oxides catalysed by iron ions or iron‐containing enzymes. In this process, a variety of inducers destroy the cell redox balance and produce a large number of lipid peroxidation products, eventually triggering cell death. However, in terms of morphology, biochemistry and genetics, ferroptosis is quite different from apoptosis, necrosis, autophagy‐dependent cell death and other RCD processes. A growing number of studies suggest that the relationship between ferroptosis and cancer is extremely complicated and that ferroptosis promises to be a novel approach for the cancer treatment. This article primarily focuses on the mechanism of ferroptosis and discusses the potential application of ferroptosis in cancer therapy.

a hot topic in a variety of diseases, especially cancer therapy. 5,6 For instance, new findings reveal that cell density can affect the sensitivity to ferroptosis, and another study showed that ferroptosis can spread through cell populations in a wave-like manner. 7,8 These factors should be considered when ferroptosis is applied to cancer therapy. In addition, some groups have tried to use nanoparticles and exosomes as carriers of erastin and drugs to precisely induce ferroptosis in tumour tissues. 9,10 These new findings and treatment attempts enrich the study of ferroptosis. Therefore, it is meaningful to review the main mechanisms underlying ferroptosis and their potential treatment value.

| A PREQUEL TO FERROP TOS IS
The cognition of ferroptosis is a cumulative process. Before Dixon defined ferroptosis, the key molecules associated with it had been reported. For example, the cystine and glutamate transport system (System Xc -) was discovered in 1986, and scholars found that exposure to high levels of glutamate or low levels of cysteine could cause a decrease in glutathione and accumulation of intracellular peroxides. 11,12 Further, Dolma team used synthetic, lethal, highthroughput screening to filtrate a mass of compounds for their potency to kill RAS-mutated tumour cells and found one chemical compound, erastin, that could cause the death of cancer cells in a non-apoptosis manner. 13 Five years later, another two small molecules, named RSL3 and RSL5, were identified and found to lead to the death of RAS-mutated cancer cells in an iron-dependent, non-apoptotic cell death manner. 14 At the same time, a new finding emerged that GPx4 depletion caused tremendous lipid peroxidation and cell death with an unrecognized cell death pattern, which was 12/15-lipoxygenase-dependent and AIF-mediated. 15 Based on these studies, the Scott J. Dixon and team expanded, extended and systemically summarized this special type of cell death, naming it ferroptosis, which is a type of RCD caused by iron-dependent lipid peroxides and shares none of the characteristic morphologic features associated with necrosis, apoptosis or autophagy-dependent cell death.

| The role of lipid peroxides in ferroptosis
The most prominent feature of ferroptosis is iron-dependent lipid peroxides. Lipid peroxides are generally viewed as eventual executioners of ferroptosis through their ability to cause plasma membrane damage. 16 Physiologically, most intracellular oxygen is reduced to H 2 O via oxidative phosphorylation in the mitochondrial inner membrane. 17 However, a small proportion of oxygen will participate in other physiological or biochemical activities, including phagocytosis, immune activation and xenobiotic metabolism, and result in harmful intermediates, such as reactive oxygen species (ROS). 18,19 Objectively speaking, a low and controlled ROS level is crucial for normal cellular and organismal function, and a moderately increased ROS level is beneficial for cancer development, but high ROS levels will cause cell injury and death. 20 Current researches studied that accumulation of polyunsaturated fatty acid (PUFA) oxides is a hallmark of ferroptosis, but the accumulation process is complicated. The process involves inhibition of deoxidation, which mainly refers to glutathione peroxidase 4 (GPx4) and ferroptosis suppressor protein 1 (FSP1), and enhancement of hyperoxidation, which is catalysed by iron and a series of enzymes.
Eventually, PUFA oxidation accumulates and leads to membrane integrity damage and ferroptosis. 19,[21][22][23][24] Lipoxygenases (LOXs) are the most important lipid oxidation enzymes for ferroptosis. LOX family members are iron-incorporating enzymes that can catalyse the dioxygenation of PUFAs. 25 Moreover, erastin-induced cell death can be prevented if arachidonate lipoxygenase (ALOX) is silenced, and several LOX inhibitors can also have this effect. [26][27][28] In addition, given that mitochondria are main site of redox reactions, it is rational to speculate that toxic ROS is generated by this organelle. However, Dixon's experiments found that mitochondria were not the source of the fatal ROS levels in erastin-treated cells. 4 To find the potential source of lethal ROS, they tested the role of the NADPH oxidase (NOX) family (NOX1-5, DUOX1, 2) in erastin-induced ferroptosis. The results demonstrated that diphenylene iodonium (DPI, an inhibitor of canonical NOX), GKT137831 (a specific inhibitor of NOX1/4), and 6-aminonicotinamde (6-AN, an inhibitor of the NADPH-generating pentose phosphate pathway) could strongly suppress erastin-induced ferroptosis in Calu-1 cells, a cell line that express a high level of NOX1 with RAS mutation. 4,29,30 These results revealed the other source of lethal ROS, but this phenomenon was limited to several cell lines ( Figure 1). 3 Molecular dynamics showed that increased lipid peroxidation directly enhances membrane permeability, changes membrane shape and curvature, and promotes increased accessibility to oxidants, eventually causing cell death. [31][32][33] A recent study demonstrated that erastin and lipid peroxidation could activate the ASK1-p38 pathway and trigger ASK1-dependent cell death in a cell type-specific manner. Further results indicated ASK1 was a downstream of lipid peroxidation in ferroptosis. However, these findings only apply in special cases. 34

| The role of glutathione in ferroptosis
System Xc − is important for the exchange of glutamate and cystine at a ratio of 1:1 across the plasma membrane. 11 As early as 1989, it was reported that inhibition of system Xc − could increase oxidative stress and mediate the toxicity of glutamate. 35 Subsequently, researchers discovered that small molecule compounds, erastin or other system Xc − inhibitors, could induce ferroptosis. 4,13,36 Indeed, cystine, imported into the cytoplasm, will be catalysed to cysteine and then to GSH (glutathione) via a multistep enzyme-catalysed reaction.
GSH is an important substrate for GPx4 to reduce phospholipid hydroperoxide (PLOOH). 37 Of note, another cysteine biosynthesis pathway, called the transsulphuration pathway, exists and generates cysteine via transfer of the sulphur atom of methionine to serine, and this pathway endows some cells with the ability to resist ferroptosis induced by system Xc − inhibitors. Further study revealed that knock-down of CARS (cysteinyl-tRNA synthetase) upregulated the transsulphuration pathway by causing cystathionine accumulation and upregulating genes associated with serine biosynthesis and transsulphuration, thereby antagonizing erastin-induced ferroptosis ( Figure 1). 38 GPx4 is the articulation point of ferroptosis-related glutathione metabolism and lipid peroxidation. GPx4 determines the fate F I G U R E 1 Main mechanisms of ferroptosis. TFRC-mediated endocytosis promotes the uptake of transferrin. Ferritin can be degraded by autophagy through ferritinophagy. The LOXs, NOXs and increased labile iron result in lipid peroxide. Cysteine can be generated from uptake of cystine via system Xcor transsulphuration pathway. GPx4 catalyses reduction reaction at the cost of GSH. CoQ10H2/α-TOH traps lipid peroxyl radicals and FSP1 catalyses the regeneration of CoQ10

DPI GKT137831
of cells because it can directly reduce the lethal phospholipid hydroperoxides. 39,40 Ras-selective lethal (RSL) small molecules, including RSL3 and erastin, can directly decrease the activity of GPx4 by binding to it or indirectly inhibiting the import of cystine, respectively, thus leading to ferroptosis. Interestingly, GPx4 can be classified into three types according to organelle location, mGPx4 (transported into mitochondria), nGPx4 (localized in nucleoli) and cGPx4 (localized in the cytosol and nucleus), and the three GPx4 types independently regulate local lipid hydroperoxides. [40][41][42][43] Mitochondrial GPx4 is associated with apoptosis by inhibiting the release of cytochrome c from mitochondria. 40,[44][45][46] While cGPx4 is associated with ferroptosis. Importantly, the Dixon team found that cells with depleted mitochondrial DNA could still trigger potent ferroptosis. 4 The organelle-specific GPx4 in cells might provide an explanation for why mitochondria are dispensable for ferroptosis. However, a recent study revealed that mitochondria are involved in cysteine deprivation-induced ferroptosis via the electron transfer chain (ETC) or TCA cycle. In addition, inhibition of glutaminolysis could lead to a parallel prohibitive effect.
These findings indicate that mitochondria play a decisive role in cysteine deprivation-induced ferroptosis but not in GPx4 inhibition-induced ferroptosis. 47

| Iron metabolism in ferroptosis
Intracellular iron is under delicate regulation to sustain iron homoeostasis. Iron regulatory proteins (IRP1 and IRP2) modulate the cellular Fe 2+ concentrations, and various proteins regulate iron import, storage, release and export. 48 Most intracellular Fe 2+ is stored in ferritin and iron-containing proteins, and the amount of free Fe 2+ , also called the cellular labile iron pool (LIP), is very limited. 31 In mammalian cells, a portion of the cellular iron can be distributed in mitochondria, the cytosol, the nucleus and lysosomes; although the amount is little, cells are sensitive to iron concentration, and a little fluctuation in concentration can cause a great response. 49,50 How to increase the level of labile iron in cells? First, uptake more iron from the extracellular environment. Dixon identified six high-confidence genes in erastin-induced ferroptosis, and IREB2 (iron response element binding protein 2) was one of them. If IREB2 was silenced, the expression of iron uptake, metabolism and storage genes, such as TFRC, ISCU, FTH1, and FTL, would decrease, and the cells exhibited resistance to erastin-induced ferroptosis. 4,51,52 Transferrin is an iron carrier protein in serum that can be transported into cells through TFRC-mediated endocytosis. Silencing of TFRC significantly inhibited conditional serum-induced ferroptosis. 36 In other words, transferrin import is required for ferroptosis.
The second way to release iron is autophagic degradation of ferritin, which is called ferritinophagy. 53 This process is primarily associated with autophagy. A recent study found that silencing of Atg5 and Atg7 (autophagy-related genes 5 and 7) decreased intracellular Fe 2+ levels and lipid peroxidation and further limited erastin-induced ferroptosis. Further study confirmed that during autophagy ferritin is degraded to release labile iron and the degradation process is regulated by NCOA4 (nuclear receptor coactivator 4), a selective cargo receptor for specific autophagy of ferritin. Blockage of autophagy or genetic inhibition of NCOA4 can inhibit ferritin degradation and suppress ferroptosis. [54][55][56] How does increased labile iron result in lipid ROS? Indeed, it primarily depends on Fenton chemistry and LOXs. Fenton chemistry also refers to the Fenton reaction. For this reaction, there must be enough labile iron in this cycle. The increased labile iron portion (LIP) in ferroptosis exactly meets this condition. In this reaction, peroxides and Fe 2+ are used to produce oxygen-centred radicals. 57 This was further confirmed by results showing that antioxidants could terminate the reaction. 58 If iron chelators, such as deferoxamine (DFO), were applied to downregulate intracellular iron, then iron-dependent lipid peroxidation formation is also restrained. 4 The indirect way for iron to generate lipid ROS is through the catalytic activity of iron-containing enzymes, and LOXs are the most important enzymes among these ( Figure 1). 59

| FSP1-NAD(P)H pathway in ferroptosis
Although GPx4 plays a crucial role in ferroptosis, certain cancer cell lines are resistant to ferroptosis caused by GPx4 inhibitors, 60,61 indicating that there might be additional factors that regulate ferroptosis. To unveil the underlying mechanism, Kirill Bersuker used a synthetic lethal CRISPR-Cas9 screen to distinguish genes in U-2 OS osteosarcoma cells treated with a GPX4 inhibitor, 62

and Sebastian
Doll generated a cDNA expression library derived from an MCF7 ferroptosis-resistant cell line and screened for genes complementing loss of GPX4. 63 Coincidentally, they simultaneously found that ferroptosis suppressor protein 1(FSP1) is an important ferroptosis suppressor that parallels GPx4. In FSP1-NAD(P)H pathway, coenzyme Q10 (CoQ10) can directly trap lipid peroxyl radicals to reduce lipid peroxides, FSP1 catalyses the regeneration of CoQ10 at the cost of NAD(P)H. Sebastian Doll also screened approximately 10,000 compounds to identify FSP1 inhibitors. iFSP1 was identified as a potent FSP1 inhibitor and can trigger ferroptosis in GPX4 knockout cells that overexpress FSP1 (Figure 1). 63  Indeed, oncogenic KRAS can regulate metabolic changes and alter cellular signalling, both of which can increase the production of intracellular reactive oxygen species (ROS), but at the same time, it also upregulates antioxidant systems to balance ROS to levels at which they are beneficial for tumour development and progression, while remaining below the threshold that causes cell death. 65 KRAS can upregulate ROS through multiple mechanisms.

| ROLE OF K R A S IN FERROP TOS IS
For example, KRAS can regulate HIF-1α and HIF-2α target genes to modulate mitochondrial metabolism or regulate the transferrin receptor (TFRC) to modulate mitochondrial respiration and ROS generation. 66,67 Notably, TFRC is a regulator in ferroptosis, as mentioned above. KRAS can also activate Rac1-NOX4 signalling to alter NADPH oxidase activities. 68,69 The NOX family also plays an important role in lipid peroxidation in ferroptosis. In addition, KRAS controls the regeneration of peroxiredoxins by inducing autophagy-specific genes 5 and 7 (ATG5, ATG7) and repressing SESN3. 70,71 Autophagy is also important in ferroptosis. With regard to antioxidant systems to sustain redox homoeostasis in the presence of KRAS-induced ROS production, a recent study showed that superoxide dismutase (Sod), glutathione peroxidase 4 (GPx4) and perox-

| ROLE OF P53 IN FERROP TOS IS
p53 is a crucial tumour suppressor that orchestrates specific cellular responses, such as transient cell cycle arrest, cellular senescence and apoptosis. 79 A recent study revealed that p53 is also involved in regulation of ferroptosis. The Le Jiang and colleagues generated a tetracycline-controlled (tet-on) p53-inducible cell line for microarray analysis to identify novel p53 target genes, and SLC7A11 (a component of system Xc − ) was identified as a novel p53 target gene. p53 could transcriptionally suppress SLC7A11 and thereby inhibit cystine uptake and sensitize H1299, U2OS and MCF7 cells to ferroptosis. In addition, p53 3KR , an acetylation-defective mutant that failed to induce cell cycle arrest, senescence and apoptosis, retained the ability to inhibit SLC7A11 expression and thereby downregulated cystine metabolism and ferroptosis upon ROS-induced stress. 80 Another study reported that p53 repressed the expression of SLC7A11 during erastin treatment by decreasing H2Bub1 occupancy on the SLC7A11 gene regulatory region. 81  indicating that p53 has a pro-survival function. 82 Wild-type p53 has been reported to delay the onset of ferroptosis in response to cystine deprivation via the p53-p21 axis to mitigate the depletion of intracellular glutathione and reduce accumulation of lethal lipid ROS. 83 In addition, certain mutant forms of p53 do not suppress the expression of SLC7A11. For example, a p53 that harboured mutations in the N-terminal domain of the gene did not downregulate SLC7A11. 84,85 Conversely, p53 promotes ROS accumulation via regulation of cytochrome c oxidase 2 (SCO2), glucose transporter (GLUT)1, GLUT4 and glutaminase 2 (GLS2). 86 90 In these conditions, p53 acts as a positive regulator of ferroptosis. Beyond doubt, this complicated genetic background of p53 will bring more challenge for cancer treatment.

| PROG RE SS IN FERROP TOS IS APPLI C ATI ON IN C AN CER THER APY
In the process of exploring how to kill RAS-mutant cancer cells, ferroptosis was accidentally discovered. The ultimate purpose of elucidating the mechanism underlying ferroptosis is to obtain better cancer treatment options. Because, based on the molecular regulation mechanisms, we can specifically target the key regulators (such as system Xc − , GPx4, autophagy and iron) to trigger ferroptosis.

F I G U R E 3
Here, we list some new findings in the utilization of ferroptosis activators as weapons to treat cancer.

| Death propagation
Although the forms of cell death vary in mechanism, from the per-

| Molecular carrier
Although RSL has been studied for many years, they are not suited for direct use in clinical applications due to their poor water solubility, renal toxicity and other toxic side effects. Thus, enhancing the specificity and delivery efficiency of these drugs is a promising re- To specifically release iron to tumour tissues, one team developed a nanoprobe that consisting of upconversion luminescence (UCL) nanoparticles as a core and a coordinated Fe 3+ /gallic acid complex as a shell. This nanoprobe can release Fe 3+ only in tumour lesions in response to the lightly acidic tumour pH and thus triggered ferroptosis. 108 Magnetosomes can also be iron carriers.
Another team engineered a magnetosome with an Fe 3 O 4 magnetic nanocluster (NC) as the core and pre-engineered leucocyte membranes as the cloak. In addition, TGF-β inhibitor (Ti) was loaded inside the membrane and PD-1 antibody (Pa) was anchored on the membrane surface. In this way, the magnetosome induced ferroptosis and regulated immunomodulation to treat cancer (Table 1). 109 CSCs. [112][113][114][115][116][117] However, iron deprivation has also been found to halt cell proliferation in mouse-induced pluripotent cells and inhibit the expression of stemness markers. 117  On the other hand, gold nanoparticles (AuNPs) are excellent drug carriers with good biocompatibility, easy synthesis and good drug functionalization capability. [121][122][123] Studies have found that salinomycin-loaded gold nanoparticles can induce ferroptosis in CSCs. 124

| Tumour microenvironment
The microenvironment is one of the most important features of tumour and plays a critical role in many aspects of tumorigenesis.
Currently, the exploration of ferroptosis is primarily focused on vitro cell lines or xenogeneic tumour cell transplantation. Thus, the role of the tumour microenvironment in ferroptosis is being neglected. Iron is the key factor in ferroptosis. An aberrantly increased labile iron portion can originate from two main sources as mentioned above: import of transferrin and degradation of ferritin via ferritinophagy. TFR mediates the import of transferrin from the extracellular environment, while FPN1 is the only ferrous iron exporter. FPN1 is primarily expressed in iron-recycling macrophage populations. 125,126 In the extracellular milieu, ferric iron is bound by apo-transferrin (TF), which is the key iron transport protein in plasma. Hepcidin is an iron regulatory hormone that acts in a negative feedback manner through binding to FPN1. 127,128 Macrophages play an important role in sustaining iron homoeostasis. Although the mechanism of iron homoeostasis in the extracellular environment is clear, the changes in response to ferroptosis are unclear and whether these changes affect the efficiency of therapy remains to be unveiled.

| CON CLUS I ON AND PER S PEC TIVE
Although the main mechanisms underlying ferroptosis have been   135 ferroptosis, which have been uncovered, provide vital therapeutic targets, such as targeting system Xc − , GPx4, autophagy and iron.
Researchers have found a number of small molecules that can induce ferroptosis by targeting these therapeutic targets. However, these compounds are not currently suited for direct use in vivo, but further application of cutting-edge technology may bring new options for cancer treatment based on ferroptosis.

CO N FLI C T O F I NTE R E S T
The authors have declared no conflicting interests.

AUTH O R CO NTR I B UTI O N
YQ conceived of the presented idea; ZY, WL and QZ wrote the manuscript; QH, ML, QS, ZZ, GF, WX and SJ searched the literature; XX and XY supervised the project.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.