3. Mechanisms of Protection from Ferroptosis
3.1. The System xc−—GSH—GPX4 Axis
Ferroptosis was initially described as a form of cell death that is induced by erastin and RSL3 through the inhibition of cystine uptake via system x
c− and GPX4, respectively
[6]. Given that cystine is an essential component of GSH, which is a co-factor of GPX4, the system x
c−-GSH-GPX4 pathway was first established as the core defense mechanism against ferroptosis.
System x
c− is a heterodimeric plasma membrane cystine/glutamate antiporter composed of SLC7A11 (xCT) and SLC3A2 (CD98)
[30]. System x
c− imports cystine while exporting intracellular glutamate. Imported cystine is reduced to cysteine, which is utilized for GSH biosynthesis. Cystine availability also promotes GPX4 protein synthesis through the activation of the mammalian target of rapamycin complex 1 (mTORC1)
[31]. A minor portion of intracellular cysteine pool can also be synthesized de novo from methionine through the transsulfuration pathway
[32]. The activation of this pathway through cystathionine β-synthase (CBS)
[33] or S-adenosyl homocysteine hydrolase (SAHH)
[34] confers resistance to ferroptosis. GSH is synthesized from cysteine via sequential enzymatic reactions involving the rate-limiting γ-glutamylcysteine synthetase (γ-GCS) and glutathione synthetase (GSS)
[35]. The inactivation of system x
c− (e.g., by erastin) or the depletion of cystine from culture media has been reported to induce ferroptosis in a wide variety of cancer models
[36][37][38]. Recently, Pardieu et al. studied the anti-leukemia effects of cystine depletion in AML and found that ferroptosis comprised only a part of the mechanisms of cell death induced
[39]. The depletion of GSH also induces a mixture of cell death modes including ferroptosis
[40][41]. Intriguingly, one recent study indicated that γ-GCS has a non-canonical, GSH-independent role in protecting cells from ferroptosis through the synthesis of γ-glutamyl peptides and the inhibition of glutamate stress
[42]. In fact, glutamate metabolism is also known as a regulator of ferroptosis; the inhibition of SLC1A5-mediated glutamine uptake, the mitochondrial glutaminase (GLS2)–mediated synthesis of glutamate from glutamine, and the glutamic-oxaloacetic transaminase 1 (GOT1)–mediated synthesis of α-ketoglutarate from glutamate, can all suppress ferroptosis
[43].
GPX4 is the only mammalian enzyme that catalyzes the reduction of lipid peroxides into nontoxic alcohols
[44]. Since the discovery that RSL3 targets GPX4 to induce ferroptosis
[7], a number of studies have established the roles of GPX4 as a central repressor of ferroptosis
[31][45][46][47]. Importantly, therapy-resistant or -persistent solid cancer cells with a mesenchymal gene signature exhibit vulnerability to GPX4 inhibition, which suggests that targeting GPX4 has therapeutic potential in certain intractable phases or types of cancers
[16][17]. Because GPX4 is a selenoprotein, its synthesis and activity are regulated by cellular selenium availability
[48]. Therefore, the incorporation of selenium within GPX4 is also important for protecting cells from ferroptosis
[49][50]. Mammalian GPX4 consists of three isoforms with distinct subcellular localizations: short, long, and nuclear isoforms of GPX4 (from here, we describe them simply as short, long, and nuclear GPX4). Whereas long GPX4 contains a mitochondria-targeting sequence and is transported into the mitochondrial matrix, short GPX4 mainly resides in the cytosol
[51]. Of note, a portion of short GPX4 is translocated into the mitochondria in a mitochondria-targeted sequence–independent manner
[52]. Only short GPX4 was initially believed to be important for cell protection against ferroptosis, mainly because short GPX4, but not the other GPX4 isoforms, rescues the lethal phenotype of GPX4 knockout in mice
[52][53][54]. However, recent studies have demonstrated that long GPX4 also plays an important role in ferroptosis protection in doxorubicin-induced cardiomyopathy
[55][56][57] and in cancer models under certain conditions, as described in the following
Section 3.3 [58][59][60].
3.2. The Ferroptosis Suppressor Protein 1-Coenzyme Q Axis
In 2019, two groups independently reported a GPX4-independent mechanism of ferroptosis protection that involves ferroptosis suppressor protein 1 (FSP1, formerly known as apoptosis inducing factor mitochondria associated 2 [AIFM2])
[61][62]. FSP1 localizes on the plasma membrane and reduces ubiquinone, or coenzyme Q (CoQ), to ubiquinol (CoQH
2). In fact, CoQ/CoQH
2 works not only as an electron carrier in the mitochondrial electron transport chain complex but also as a major endogenous lipophilic radical-trapping antioxidant that protects cells from lipid peroxidation
[63]. In this reaction, CoQH
2 reduces lipid peroxides and is oxidized to CoQ, which is then recycled back to CoQH
2 by FSP1 using NADPH. Indeed, NADPH was identified as a predictive biomarker of ferroptosis sensitivity in cancer cell lines
[64], and its depletion by the phosphatase guanosine-3′,5′-bis(diphosphate) 3′-pyrophosphohydrolase (MESH1) facilitates ferroptosis
[65]. Using Kelch-like ECH-associated protein 1 (KEAP1)–mutant lung cancer models, Koppula et al. demonstrated that the FSP1-CoQ axis is regulated by the KEAP1–nuclear factor erythroid 2–related factor 2 (NRF2) pathway
[66]. The mouse double minute 2 (MDM2)–murine double minute X (MDMX) complex also regulates FSP1 through peroxisome proliferator-activated receptor α (PPARα) in a p53-independent manner to promote ferroptosis
[67]. Recently, Mishima et al. discovered that vitamin K is another substrate of FSP1 that protects cells from ferroptosis
[68], and they demonstrated that the reduced form of vitamin K is a potent radical-trapping antioxidant that inhibits lipid peroxidation.
3.3. The Dihydroorate Dehydrogenase/Glycerol-3-Phosphate Dehydrogenase 2–Mitochondrial CoQ Axis
Two recent studies uncovered mitochondria-specific ferroptosis defense mechanisms that are independent of cytosolic anti-ferroptosis pathways. In one study, Mao et al. found that dihydroorate dehydrogenase (DHODH), an enzyme required for pyrimidine synthesis, reduces CoQ to CoQH
2 in the mitochondrial inner membrane to protect cells from ferroptosis
[59]. The same group also found that glycerol-3-phosphate dehydrogenase 2 (GPD2) is another enzyme that couples glycerol-3-phosphate oxidation with mitochondrial CoQ reduction to suppress ferroptosis
[60]. Importantly, these mitochondria-specific machineries cooperate with mitochondria-localized long GPX4, but not short GPX4, to suppress mitochondrial membrane lipid peroxidation to protect cells from GPX4 inhibition–induced ferroptosis. These studies suggest the importance of subcellular compartmentalization in ferroptosis defense mechanisms
[69].
3.4. The GTP Cyclohydrolase 1–Tetrahydrobiopterin–Dihydrofolate Reductase Axis
GTP cyclohydrolase 1 (GCH1) is another GPX4-independent regulator of ferroptosis
[20][70]. It protects cells from lipid peroxidation through the production of tetrahydrobiopterin (BH
4) and dihydrobiopterin (BH
2). BH
4 and BH
2 are endogenous radical-trapping antioxidants and are generated by dihydrofolate reductase (DHFR). BH
4 is also involved in the synthesis of reduced CoQH
2 and the depletion of PUFA-containing phospholipids.
4. Regulation of Ferroptosis by p53
The “guardian of the genome”, p53, is a well-established tumor suppressor. Owing to deletion and/or loss-of-function mutations, p53 has deficient function in approximately half of human cancers
[71]. Many other cancers overexpress MDM2 or MDMX, which results in degradation of p53
[72]. Patients who have AML with TP53 mutations—an AML subtype that is one of the most challenging to treat—have low treatment response rates and poor survival
[73]. In addition to its canonical roles of inducing cell cycle arrest, senescence, apoptosis, and DNA damage repair, p53 regulates multiple nodes of cellular metabolism, including iron, lipid, and redox metabolism, and thus plays important roles in the regulation of ferroptosis.
In 2015, Jian et al. reported that functional p53 directly suppresses SLC7A11 to induce ferroptosis and inhibit tumor development
[74]. They demonstrated that acetylation-deficient TP53
3KR mutants can efficiently induce ferroptosis while losing their ability to upregulate canonical p53 target genes (e.g., CDKN1A/p21, PUMA). Another study showed that the mutation of all four acetylation sites in TP53 (i.e., TP53
4KR) abolishes the gene’s effect against SLC7A11 and thus eliminates its tumor-suppressive function. These studies indicate that p53, through its regulation of senescence, apoptosis, and ferroptosis, has distinct functions as a tumor suppressor. Another study also demonstrates that mutant p53 (e.g., TP53
R175H) retains the ability to repress the expression of SLC7A11 by inhibiting NRF2-mediated transcription programs
[75]. Several metabolism-associated genes, including ALOX12
[76], the spermidine/spermine N
1-acetyltransferase 1 gene SAT1
[77], and the ferredoxin reductase gene FDXR
[78], are important for the regulation of p53-induced ferroptosis. Another p53 target, GLS2, promotes ferroptosis through the regulation of glutaminolysis; however, further investigation is needed to determine the specific role of GLS2 in p53-mediated ferroptosis
[43][79].
Intriguingly, p53 can also suppress ferroptosis under certain conditions. Tarangelo et al. reported that, in contrast to the aforementioned SLC7A11 suppression and ferroptosis induction mediated by p53, the stabilization of wild-type p53 by MDM2-p53 interaction, by inducing p21 and de novo GSH synthesis, delays ferroptosis induced by system x
c− inhibition or cysteine deprivation
[80]. Another direct target of p53 is calcium-independent phospholipase A2 group VI (iPLA2β), which detoxifies membrane lipid peroxidation to protect cells from ferroptosis
[81]. Kuganesan et al., dissecting the roles of p53 and downstream p21, CDKs, RB, and E2F in ferroptosis modulation
[82], speculated that multiple pro-ferroptosis and anti-ferroptosis signals may emanate from the individual nodes of p53 pathways, leading to context-dependent outcomes against ferroptosis stimuli and tumor progression.
Taken together, the role of p53 in ferroptosis regulation is highly complex and context dependent. Extensive reviews on this topic are available elsewhere
[83][84][85].
5. Ferroptosis in Leukemia
5.1. Dysregulation of Iron Homeostasis in Leukemia
Leukemia patients are characterized by systemic iron overload due to multiple factors. The most common reason is repetitive red blood cell transfusions that lead to massive iron intake over the courses of disease
[86]. Ineffective erythropoiesis and a high turnover of immature erythrocytes can also lead to iron overload, especially in AML associated with myelodysplasia. Mechanistically, ineffective erythropoiesis suppresses hepcidin leading to enhanced iron uptake
[87]. Hematopoietic stem cell transplantation can also disturb iron homeostasis by suppressing erythropoiesis and by erythroid cell lysis, which can affect patients’ survival
[88][89]. Lopes et al. recently reported on iron redistribution in AML cells
[90]. They found that AML patients have increased transferrin saturation (TSAT) and high hepcidin irrespective of transfusion history, indicating increased levels of circulating iron. Furthermore, using electron microscopy combined with energy-dispersive X-ray spectroscopy for elemental analysis and flow cytometry for LIP quantification, they found increased intracellular iron levels in AML cells. Compared to healthy bone marrow cells, AML cells show increased expression of HO-1 and ferritin light chain 1 (FTL1), indicative of intracellular iron accumulation. The expression level of the iron importer TFR1 in leukemia cells is similar to that in healthy bone marrow cells but lower than that in erythroblasts, which suggests that leukemia cells have a relatively low iron demand
[90][91].
The correlation of dysregulated iron metabolism and patient outcome in leukemia, however, remain unclear. The low expression level of the iron exporter FPN1 in AML cells is associated with chemosensitivity and better patient outcome
[92]. In another study, Trujillo-Alonso et al. showed that the expression levels of FPN1 in primary AML blasts and leukemia stem cells are lower than that in healthy bone marrow CD34
+ hematopoietic stem progenitor cells
[93]. They demonstrated that low FPN1 expression leads to increased intracellular iron and oxidative stress, which sensitized cells to the iron oxide nanoparticle ferumoxytol. Together, these data suggest that leukemia cells have impaired iron flux and intracellular iron accumulation, which may render them vulnerable to ferroptosis induction.
5.2. Ferroptosis-Related Gene Signatures in Leukemia
Since ferroptosis was first identified as a distinct type of cell death, numerous studies have identified molecular regulators of this complicated cell death process. This extensive knowledge is organized in FerrDb, a publicly available database of ferroptosis-related genes and their associated diseases
[94][95]. High-throughput gene testing with microarrays and, more recently, RNA sequencing, have enabled researchers to capture gene expression patterns as gene signatures that can be used to identify multiple clusters among patients with heterogeneous cancers based on their genetic backgrounds and to predict prognosis
[96].
On the basis of these patient datasets, including those in the Cancer Genome Atlas (TCGA), multiple groups have defined ferroptosis-related genes (FRGs) and have correlated their expression with patient survival to develop FRG scoring models that can predict prognosis and ex vivo drug sensitivities in association with their clinical characteristics in AML
[97][98][99][100][101][102][103][104], acute lymphoblastic leukemia (ALL)
[105], and CLL
[106][107]. Interestingly, many FRG signatures are associated with immune-related pathways and immune cell infiltration predicted by CIBERSORTx gene-expression profiling analyses, which suggests that ferroptosis has relevance in anti-tumor immune responses and potentially in immunotherapy in leukemia. In addition to FRGs’ expression levels, their methylation status
[108] and copy number variations
[109] have also been reported to have prognostic significance. One group studying the correlation of long non-coding RNAs (lncRNAs) and FRGs identified ferroptosis-related lncRNA signatures that predict the prognosis of AML
[110].
Although multiple FRG signatures share some genes (e.g., GPX4, AIFM2), these models are based on different datasets and statistical methods, which has resulted in distinct scoring systems with individual clinical relevance. Further validations in independent datasets are warranted to ensure the robustness of a model to be applied in clinical practice, including in terms of its prognosis prediction and treatment optimization. In addition, basic in vitro and in vivo experiments are needed to verify the expression level and determine the biological role(s) of each FRG. Of note, ferroptosis-related prognostic gene signatures developed from the pediatric AML dataset in the National Cancer Institute’s Therapeutically Applicable Research to Generate Effective Treatments (TARGET) Initiative could not be validated in an adult AML dataset from TCGA
[111], likely because of the two cohorts’ distinct genetic landscapes and clinical characteristics.
5.3. Ferroptosis Induction as a Therapeutic Strategy in Leukemia
Ferroptosis induction has been demonstrated to exert anti-tumor effects in various leukemia models. Most of the evidence is based on in vitro experiments using cell lines or patient-derived primary cells, but some of the studies include in vivo models. Not only classic ferroptosis inducers (e.g., system xc−/GSH/GPX4 inhibitors) but also natural compound derivatives and other small molecules are shown to induce ferroptosis as a mechanism of their anti-leukemia activities.
5.3.1. Inhibition of System xc-—GSH—GPX4 Axis in Leukemia
The first study of ferroptosis in AML was reported in 2015 by Yu et al., who demonstrated that the system xc- inhibitor erastin induces ferroptosis in HL-60 cells in vitro [184]. The cell death was a mixture of ferroptosis and necroptosis, as it was blocked not only by ferrostatin-1 and deferoxamine but also by necrostatin-1 and the knockdown of receptor-interacting protein 3 (RIP3). They also demonstrated the association of autophagy and p38 signaling, the inhibition of which attenuated the anti-leukemia effects of erastin. Later, erastin-induced ferroptosis in HL-60 cells was shown to depend on the cytoplasmic translocation of high mobility group box 1 (HMGB1) from the nucleus, the knockdown of which attenuated ferroptosis in vivo [185]. Recently, Pardieu et al. reported that the xCT gene SLC7A11 is a putative therapeutic vulnerability, especially in NPM1-mutated AML, and a poor prognostic factor [113]. They demonstrated that the system xc- inhibitor sulfasalazine suppressed GSH and induced oxidative stress–dependent cell death partly through ferroptosis, as indicated by its partial rescue by ferrostatin-1. The combination of sulfasalazine with daunorubicin and cytarabine was found to synergistically exert anti-leukemia effects in a patient-derived xenograft model as well as in primary AML cells. A clinical trial of sulfasalazine in combination with intensive chemotherapy in AML patients is being initiated (NCT05580861). K562 chronic myeloid leukemia cells were tested for cysteine depletion-induced ferroptosis [188]. The authors demonstrated that the concomitant inhibition of thioredoxin reductase 1 (TXNRD1) by auranofin leads to ferroptosis in these cells. Interestingly, K562 cells with imatinib resistance (K562/G0 cells) showed higher sensitivity to ferroptosis induced by cysteine depletion.
5.3.2. Ferroptosis Induction by Clinically Available Anti-Leukemia Agents
Hypomethylating agents are used as a standard of care for elderly patients with AML and for patients with myelodysplastic syndromes (MDS) [208]. One hypomethylating agent, decitabine, has been shown to induce ferroptosis and necroptosis in MDS-derived primary cells as well as cell lines [189]. It downregulates GSH and GPX4 while inducing ROS and cell death, which can be blocked by ferrostatin-1, necrostatin-1, and z-VAD-FMK. These data indicate that ferroptosis is at least partially involved in the effect of decitabine.
Wang et al. investigated the anti-leukemia effects observed for the combination of granulocyte-colony stimulating factor (G-CSF) and thrombopoietin (TPO) with low-dose chemotherapy investigated in a phase 2 trial in elderly AML patients [209]. They demonstrate that TPO induces ferroptosis through the suppression of E1A binding protein P300 (EP300)–mediated GPX4 transcription, whereas G-CSF induces pyroptosis through neutrophil elastase, which activates gasdermin D (GSDMD) in AML cells [190].
In 2020, Birsen et al. reported that, independent of its postulated effects against mutant TP53, APR-246 induces ferroptosis in AML cells during the early phases of drug exposure [191]. Ferroptosis induction by APR-246 was later confirmed in esophageal cancer cell lines, which showed an increase in GSH turnover and the suppression of mitochondrial iron-sulfur cluster biosynthesis through NFS1 [210].
Imetelstat is a first-in-class telomerase inhibitor currently under a phase 2 clinical trial against AML (NCT05583552), in addition to phase 3 clinical trial against myelodysplastic syndromes and myelofibrosis. A functional genetic screen combined with lipidomics revealed that imetelstat promotes PUFA-containing phospholipid synthesis in ACSL4– and FADS2–dependent manner, leading to AML cell ferroptosis both in vitro and in vivo [192]. However, it has not yet been shown whether the induced ferroptosis is telomerase-dependent or drug-specific off-target effects.
The tyrosine kinase inhibitor Neratinib was approved by the U.S. Food and Drug Administration in 2017 for the treatment of breast cancer, and a phase 1/2 clinical trial of this drug in pediatric patients with relapsed or refractory cancer including leukemia is ongoing (NCT02932280). Recently, neratinib has been shown to induce autophagy-dependent ferroptosis as well as G0/G1 arrest and apoptosis in HL-60 cells [193].
5.3.3. Ferroptosis Induction by Natural Compounds and Their Derivatives
Many natural compounds have been studied for their ability to induce ferroptosis in cancer cells [211]. Dihydroartemisinin is a derivative of Artemisia annua, a plant native to China, and has anti-tumor effects in many types of cancers. This compound has effects against HL-60 cells in vitro and in vivo through the inhibition of mitochondrial oxidative phosphorylation and the activation of AMPK to induce ferritinophagy and ferroptosis [194]. Typhaneoside, a flavonoid extracted from Pollen typhae, has also been reported to induce apoptosis and ferroptosis associated with ferritinophagy in AML cells [195]. Hydnocarpin D is another flavonoid demonstrated to induce apoptosis and ferroptosis through autophagy in ALL cells [196]. Perillaldehyde, the main component of Ammodaucus leucotrichus, downregulates GSH and GPX4 to induce ferroptosis in HL-60 cells and primary AML cells [197]. Glycyrrhetinic acid is a bioactive compound of licorice, and its nanoparticles have been demonstrated to induce ferroptosis in AML cells through GPX4 downregulation [198]. These nanoparticles have synergistic anti-leukemia effects with ferumoxytol, an iron oxide nanoparticle, and programmed cell death ligand 1 (PD-L1) antibody treatment. The isothiocyanate sulforaphane (SFN) is derived from cruciferous vegetables and is known to exert anti-cancer activities. Greco et al. reported that low-dose SFN induces apoptosis whereas high-dose SFN induces ferroptosis in AML cell lines [199]. Interestingly, SFN activates the apoptosis pathway when ferroptosis is impaired, indicating that these two cell death modes could be convertible. Honokiol, a derivative of the magnolia tree, induces cell death in AML cell lines; this cell death was triggered by the upregulation of HO-1 and associated with lipid peroxidation and the alteration of ferroptosis pathway genes [200]. 4-Amino-2-trifluoromethyl-phenyl retinate (ATPR), a retinoid derivative synthesized from all-trans retinoic acid (ATRA), has been shown to activate NCOA4-dependent ferritinophagy and induce ferroptosis in AML in vitro and in vivo [201]. Poricoic acid A, the component of the mushroom Poria cocos, exerts anti-ALL effects in vitro and in vivo through mitochondrial dysfunction and activation of AMPK-mTOR autophagy pathway, leading to apoptosis and ferroptosis in T-ALL cells [202].
5.3.4. Other Potential Therapeutic Strategies Inducing Ferroptosis in Leukemia
ALDH3a2, an enzyme that oxidizes long-chain aliphatic aldehydes, which are byproducts of lipid peroxidation, protects AML cells from oxidative stress. The inhibition of ALDH3a2 induces ferroptosis and exerts synergistic anti-leukemia effects with GPX4 inhibition or standard chemotherapy with cytarabine plus daunorubicin in vivo [96]. In addition, ALDH3a2 is selectively essential in leukemia progenitor cells but not in their normal counterparts, possibly owing to increased oxidative stress in leukemia cells.
Gold nanoparticles have anti-leukemia effects in part through ferroptosis induction. For example, GNR-CSP12 (gold nanorods loaded with chitosan and a 12-mer peptide) can induce ferroptosis through the suppression of global m6A RNA methylation and its combinatorial effects with tyrosine kinase inhibitors or PD-L1 checkpoint inhibitors [203].
Non-coding RNAs have also been reported to be involved in ferroptosis regulation in AML cells. A nuclear long non-coding RNA (lncRNA), LINC00618, is downregulated in AML, and its induction by vincristine treatment activates ferroptosis and apoptosis through SLC7A11 downregulation and BAX upregulation as well as caspase-3 cleavage [204]. The circular RNA circKDM4C, which is also downregulated in AML, sequesters the microRNA hsa-let-7b-5p and upregulates its downstream target, p53 [205]. The induction of circKDM4C causes AML cells to undergo ferroptosis, possibly through the downregulation of GPX4 and FTH1.
- Conclusions and Future Perspectives
Recent discoveries of various non-apoptotic RCD modes, including ferroptosis, have expanded the potential modalities to induce death in cancer cells, especially in cancers resistant to conventional therapies targeting apoptosis mechanisms. A comprehensive understanding of the complex regulatory mechanisms of RCD and its involvement in cancer pathophysiology is necessary to develop RCD modes different from apoptosis into advanced cancer therapies.
While more than 10 years of extensive research has provided us with a large body of knowledge about ferroptosis, one needs to be cautious as most of our current mechanistic insights are based on in vitro models, whose environments differ from those in vivo. Components of the in vivo environment that likely affect ferroptosis regulation include oxygen; trace metals, including iron and selenium; and various metabolites such as amino acids and fatty acids. In addition, interactions with the tumor immune microenvironment (i.e., macrophages and immune cells) and even between cancer cells have been shown to affect cell vulnerability to ferroptosis, a complexity that is difficult to recapitulate in culture models. In fact, mesenchymal stem cells (MSCs) protect leukemia cells from oxidative stress through upregulation of GSH [212,213], while its significance in ferroptosis protection remains to be elucidated. Therefore, the in vitro findings must be validated in vivo to obtain a precise understanding of the mechanisms regulating ferroptosis, for which further development of specific and stable in vivo drugs to induce ferroptosis is of urgent need. The lack of definitive markers of ferroptosis is an obstacle in studying ferroptosis in vivo and in clinical settings in humans. Hence, direct evidence of ferroptosis in cancer patients treated with anti-cancer agents including ferroptosis inducers is lacking so far [214].
Since leukemia is characterized by increased oxidative stress and iron overload, one can speculate that leukemia cells are vulnerable to ferroptosis, suggesting a therapeutic potential. However, studies focusing on ferroptosis as a therapeutic modality for leukemia are limited, as discussed here. Given the physiological roles of ferroptosis, the therapeutic windows of ferroptosis induction should be carefully determined. A recent study demonstrated the potential vulnerability of hematopoietic stem cells to ferroptosis [215], warranting precise monitoring of normal hematopoiesis upon anti-cancer treatments involving ferroptosis. Taken together, implementing the concept of ferroptosis into leukemia therapeutics requires fine-tuning ferroptosis induction in normal versus malignant cells, which should be tested and established through in vivo studies and ultimately clinical trials.
In addition to understanding the underlying molecular mechanisms of ferroptosis, some of which may be specific to leukemia cells, identifying potential biomarkers that predict sensitivity or resistance to ferroptotic stimuli is critical for the development of this unique cell death mode as anti-leukemia therapy. Advancements in single-cell analysis technologies will help in the identification of potential vulnerabilities of leukemia stem/progenitor cells with specific genetic or molecular phenotypes that can be targeted for ferroptosis induction. Pursuing synthetic lethal interactions would be one promising strategy to efficiently target those populations while minimizing potential toxicity to normal cells [216]. Lastly, recent studies have revealed the interconnectivity and plasticity of different RCD pathways including ferroptosis that could compensate each other upon cell death stimuli (detailed review elsewhere [217]). Understanding the interactions as well as the molecular mechanisms of individual RCD pathways would facilitate the development of combination therapies to overcome resistance to various RCD mechanisms.