Non-coding RNAs and ferroptosis: potential implications for cancer therapy

Ferroptosis is a recently defined form of regulated cell death, which is biochemically and morphologically distinct from traditional forms of programmed cell death such as apoptosis or necrosis. It is driven by iron, reactive oxygen species, and phospholipids that are oxidatively damaged, ultimately resulting in mitochondrial damage and breakdown of membrane integrity. Numerous cellular signaling pathways and molecules are involved in the regulation of ferroptosis, including enzymes that control the cellular redox status. Alterations in the ferroptosis-regulating network can contribute to the development of various diseases, including cancer. Evidence suggests that ferroptosis is commonly suppressed in cancer cells, allowing them to survive and progress. However, cancer cells which are resistant to common chemotherapeutic drugs seem to be highly susceptible to ferroptosis inducers, highlighting the great potential of pharmacologic modulation of ferroptosis for cancer treatment. Non-coding RNAs (ncRNAs) are considered master regulators of various cellular processes, particularly in cancer where they have been implicated in all hallmarks of cancer. Recent work also demonstrated their involvement in the molecular control of ferroptosis. Hence, ncRNA-based therapeutics represent an exciting alternative to modulate ferroptosis for cancer therapy. This review summarizes the ncRNAs implicated in the regulation of ferroptosis in cancer and highlights their underlying molecular mechanisms in the light of potential therapeutic applications.


INTRODUCTION
Non-coding RNAs: master regulators of cellular processes Non-coding RNAs (NcRNAs) are a miscellaneous group of noncoding transcripts with limited protein-coding potential that perform important cellular functions through different molecular mechanisms [1]. Initially, it was thought that they are functionally irrelevant. However, as a myriad of functional ncRNAs were identified and characterized, the central dogma of proteins being the functional end product of gene expression has drastically changed [2].
Broadly spoken, they can either be subdivided into short and long ncRNAs (a general cut off value is 200 nucleotides in length), or subdivided due to their biological roles [3]. Three major classes of functional ncRNAs are short microRNAs (miRNAs), long ncRNAs (lncRNAs), and circular RNAs (circRNAs). Depending on their intrinsic features, they may show tissue-and/or disease-specificity and may be detected in all body fluids making them interesting for their potential utilization as biomarkers [4]. In addition, ncRNAs are frequently deregulated in various diseases, including cancer [5]. In many instances their involvement in drug resistance in cancers has been reported [6][7][8][9]. Thus, targeting of ncRNAs might be a promising therapeutic option to modulate drug resistancepromoting pathways in cancer cells and improve the outcome of patients [10].
Ferroptosis: the molecular mechanisms of a recently identified form of cell death Ferroptosis is an iron-dependent, oxidative form of cell death that is biochemically and morphologically different from other types of regulated cell death [11][12][13]. Ferroptosis is caused by excessive oxidative destruction (peroxidation) of lipids in the cellular membranes. The process relies on iron, reactive oxygen species (ROS), and phospholipids containing polyunsaturated fatty acids (PUFAs) [14][15][16].
Lipid peroxidation occurs when a bisallylic hydrogen atom, located between two carbon-carbon double bonds, is removed from the PUFAs in the membrane phospholipids. The result is formation of a carbon-centered phospholipid radical, phospholipid peroxyl radical, and phospholipid peroxides, a form of lipid ROS. Phospholipid peroxides can react with iron to generate free alkoxyl and peroxyl radicals [17]. The requirement of iron in this form of cell death inspired the term ferroptosis [12]. If not converted to its corresponding alcohol, phospholipid peroxides, together with phospholipid free radicals, promote further phospholipid peroxide formation via the processes of hydrogen removal and reaction with oxygen. It is the unrestrained lipid peroxidation that is considered to be the hallmark of ferroptosis [18].
Some first hints of ferroptosis-like death have been observed in the middle of 20-th century in studies investigating metabolism and neuronal cell death. The earliest reports attributed ferroptosis either to other forms of regulated cell death, or it was not recognized as being biologically significant. It was not interpreted as sufficient evidence for a distinct cell death until early 2000s when the Stockwell lab conducted screening of lethal compounds in RAS-transformed cancer cells. They identified erastin and RAS synthetic lethal 3 (RSL3) as inducers of non-apoptotic, irondependent cell death preventable by iron chelators and lipophilic antioxidants [19,20]. The following findings, including the mechanism of action of erastin and RSL3, lead to the idea of a unique regulated cell death form. The term ferroptosis was introduced in 2012, and, thus, the field of ferroptosis research is rather officially young [12]. Despite being frequently cited as a new type of cell death, ferroptosis may actually be considered the oldest and evolutionary most conserved form of regulated cell death owing to its simple molecular requirements of iron and oxygen. In fact, ferroptosis-like death has been observed in lesscomplex species including protozoa, prokaryotes, fungi and plants [21][22][23][24].

Ferroptosis initiation and regulation
Lipid peroxidation can be initiated by non-enzymatic and enzymatic processes [25]. The non-enzymatic process is triggered by Fenton reaction, where iron and hydrogen peroxide react toward free radical formation and propagation of lipid peroxidation [15]. Numerous enzymes were implicated in the regulation of ferroptosis and are outlined in Fig. 1. Some of the key enzymes are described below.
Cyst(e)ine-GSH-GPX4-system x c is considered as the main antagonist of ferroptosis. Ferroptotic death via inhibition of system x c and GPX4 can be induced by the compounds erastin and RSL3 [19,20]. While RSL3 impairs GPX4 activity directly, erastin inhibits cystine import into the cell and thereby indirectly affects GPX4 activity. Erastin additionally targets voltage-dependent ion channels inducing mitochondrial dysfunction [19,20]. RSL3 and erastin are frequently used in experimental approaches to induce ferroptosis.
Several other opposing mechanisms of ferroptosis have been described. Ferroptosis suppressor protein 1 (FSP1) reduces lipid peroxidation and ferroptosis by reducing the oxidized form of coenzyme Q10, ubiquinone, to yield ubiquinol, which then reduces lipid radicals and prevents propagation of lipid peroxidation [42]. Tetrahydrobiopterin (BH 4 ) protects phospholipids from oxidative degradation by acting as an antioxidant and aids ubiquinone synthesis [43]. In addition, FSP1 may also indirectly reduce lipid peroxidation via regeneration of vitamin E, another strong antioxidant of lipids [44]. There is evidence that vitamin E (and possibly selenium) supplementation may promote cancer development [45]. FSP1 is indirectly upregulated by miR-4443, and several transcription factors including nuclear factor E2-related factor 2 (NRF2) [46,47]. NRF2 is an important transcription activator that regulates various genes involved in metabolism, inflammation, mitochondrial respiratory, proliferative and transport processes [48]. In particular, NRF2 is crucial for cell survival during the oxidative stress. In addition to FSP1, it promotes the expression of several other negative regulators of ferroptosis, such as ferritin heavy chain 1 (FTH1), SLC7A11, and cystathionine βsynthase (CBS) (Fig. 1) [49][50][51]. Radical-trapping antioxidants ferrostatin 1 and liproxstatin 1 are useful in experimental studies to suppress ferroptosis [52].
In addition, cytochrome P450 oxidoreductase may initiate lipid peroxidation by reducing ferric iron (Fe 3+ ) to ferrous iron (Fe 2+ ), a reaction that is crucial for the Fenton reaction and lipid peroxidation ( Fig. 1) [61]. Although phospholipid peroxides may interact with both forms of iron, ferrous iron is likely to be more important since ferric iron has poor solubility and bioavailability in cells [14].

Consequences of ferroptosis
Some of the main end products of this process are two omega-6 fatty acids, toxic 4-hydroxy-2-nonenal (4-HNE) and mutagenic malondialdehyde (MDA) (Fig. 1) [62]. Consequences of ferroptosis include formation of various secondary lipid peroxide breakdown products, modification and oxidation of proteins, and eventual breakdown of membrane integrity. Morphologically, ferroptotic cells have small mitochondria with increased mitochondrial membrane densities, reduced or vanishing mitochondrial cristae, and rupture of the outer mitochondrial membrane [52]. In addition, ferroptotic death may induce cell membrane rupture, release of intracellular content such as damage-associated molecular patterns (DAMPs), inducing sterile inflammation and can therefore be classified as a form of regulated necrosis [15]. Membrane repair is dependent on endosomal sorting complexes required for transport-III (ESCRT)-III ( Fig. 1). This protein complex, consisting of 12 subunits, assembles into the spiral filament and mediates membrane remodeling [63].

Ferroptosis, metabolic and cellular signaling pathways
The regulation of ferroptosis is strongly connected to various essential cellular processes, including metabolic pathways (iron, lipids, amino acids, and glucose metabolism), mitochondrial activity, maintenance of redox status, or response to radiation exposure. Furthermore, several key mediators of cell signaling pathways have been implicated in the regulation of ferroptosis, including multiple oncogenic and tumor-suppressive proteins (e.g., p53) [14,16].
The activity of several metabolic pathways can affect the generation of ROS and are therefore strongly associated with the induction of ferroptosis. For example, glutamine can replenish tricarboxylic acid (TCA) cycle through the generation of αketoglutarate [64]. High glutamine uptake and metabolism can result in increased TCA cycle activity and increased rate of mitochondrial respiration, leading to ROS formation and loss of the mitochondrial membrane potential. High extracellular concentration of glutamate impedes system x c function due to inhibition of cystine uptake, and eventual intracellular glutathione synthesis, therefore leading to the ferroptosis induction [12].
Hypoxia promotes ferroptosis by increasing ROS production which can directly contribute to lipid peroxidation and activation of hypoxia-inducible factors (HIFs) [14]. HIFs have been shown to drive ferroptosis in clear cell renal carcinoma [76]. Even though glucose starvation increases ROS generation, it actually suppresses ferroptosis through the activation of energy sensor AMP activated kinase (AMPK). Energy-stress-mediated AMPK activation inhibits acetyl-CoA carboxylase which blocks the conversion of acetyl-CoA to malonyl-CoA and thus synthesis of PUFAs [77,78]. In contrast, the core regulator of autophagosome formation Beclin-1 (BECN1) promotes ferroptosis by inhibiting system x c activity in energysufficient AMPK-mediated manner [79].
Tumor suppressor proteins may sensitize cells to ferroptosis. P53, an important regulator of apoptosis and autophagy, enhances ferroptosis and prevents tumor development via suppressing the transcription of system x c component SLC7A11 [80]. In addition, BRCA1-associated protein 1 (BAP1) also promotes ferroptosis by SLC7A11 downregulation [81]. Involvement of oncogenes in ferroptosis is best illustrated through the initial discovery of stronger lethality of erastin and RSL3 in RAS-mutated cancer cells, suggesting determining role of RAS-RAF-MEK pathway in ferroptosis [82].
Furthermore, ionizing radiation has been shown to upregulate ACSL4 expression in cancer leading to increased lipid peroxidation and ferroptosis [83]. Clearly, many more regulators of ferroptosis are yet to be discovered and characterized. As mentioned, and exemplified earlier in text, ncRNAs as versatile master regulators of cellular processes have also recently been linked to the regulation of ferroptosis (Fig. 2). The following sections will systematically summarize the current knowledge on the involvement ncRNAs in regulation of ferroptosis and their role in cancer.

The role of miRNAs in the regulation of ferroptosis in cancer
MiRNAs are a class of small evolutionarily conserved ncRNAs with a length of approximately 22 nucleotides. Their main function is post-transcriptional regulation of gene expression through binding to complementary target mRNA sequences, leading to translational inhibition or mRNA degradation. The final result is halted protein synthesis [84,85]. In addition, it has been reported that miRNAs may also induce gene expression by binding to target sequence and act as translational activator [86]. While they were completely unknown less than three decades ago, nowadays it has been estimated that miRNAs control the expression of over 60% of all protein-coding genes [87]. Given their substantial regulatory capacity, it seems obvious that deregulation of this tightly controlled miRNA network is frequently linked to cardiovascular, autoimmune, infectious, and neurodegenerative diseases [84]. In fact, they are increasingly recognized as major mediators of disease. The first link between miRNA and human cancer was reported in 2002 by Calin and colleagues [88]. Since then, thousands of miRNAs have been discovered, their deregulation in virtually every type of cancer has been confirmed, and their involvement in all hallmarks of cancer has been revealed [89]. Their regulatory roles in ferroptosis in cancer are not well understood yet, but there is strong evidence that miRNAs are also involved in this crucial process in cancer cells (Fig. 2). Some prominent examples are outlined below and a broader overview is provided in Table 1.
In lung cancer, miR-101-3p is found to be downregulated [26]. When available, this miRNA targets oncogenic transducin beta-like 1X-linked (TBLR1) protein. Low expression of miR-101-3p and high expression of TBLR1 result in enhanced activity of the transcription factor nuclear factor kappa B (NF-κB), which regulates ferroptosis through GPX4 and prostaglandin-endoperoxide synthase 2 (PTGS2). Interestingly, appealing results on tumor growth reduction were observed in in vivo experiments when miR-101-3p was delivered in the form of nanoparticles [26].

Ferroptosis-inhibitory miRNAs
It has been shown in melanoma cells that miR-137 inhibits lipid peroxidation and iron accumulation in vitro and in vivo by directly targeting solute carrier family 1 member 5 (SLC1A5) [93]. SLC1A5, a non-member of system x c -, is a neutral amino acid transporter of alanine, serine, cysteine, and glutamine [94]. As a consequence, reduced levels of this important glutamine transporter lead to decreased glutamine uptake, glutaminolysis, and MDA accumulation [93]. Under physiological conditions, glutamine uptake and its metabolism induce lipid ROS generation and ferroptotic cell death [64]. In addition, miR-137 is associated with TNM stage, metastasis and drug resistance in various cancers through different pathways [95][96][97][98].
In addition, miR-7-5p is highly expressed in radioresistant cell lines of ovarian, oral squamous cell and HCC. miR-7-5p downregulates mitoferrin, a protein responsible for transporting iron into mitochondria. Ferroptosis is diminished as a result of reduced iron levels [99]. This miRNA is also upregulated in radioresistant cervical cancer [59]. The observed radio-resistance in cervical and oral squamous carcinoma cell lines is, at least partly, due to miR-7-5p effect on ferroptosis. Knockdown of miR-7-5p is shown to increase ROS levels, mitochondrial membrane potential, intracellular Fe 2+ content, as well as downregulation of the iron storage protein ferritin, and upregulation ALOX12 expression [59].
MiR-4443 is upregulated in non-small cell lung cancer (NSCLC) where it contributes to cisplatin resistance [46]. In addition, this miRNA may be transferred to the sensitive cells via exosomes and make them resistant. Mechanistically, miR-4443 inhibits ferroptosis via regulation of FSP1 expression in an N 6 -methyladenosine (m6A) manner by directly targeting its gene METLL3 [46]. MiRNAs have extremely diverse regulatory roles. In addition to direct regulation of ferroptotic key players, as outlined in the above section, miRNAs may indirectly regulate this cell death process via interaction with other ncRNAs, as illustrated in the following sections.
The role of lncRNAs in the regulation of ferroptosis in cancer LncRNAs are a class of heterogeneous ncRNAs that are more than 200 nucleotides in length. They share many features with mRNA regarding transcriptional and post-transcriptional processing [100]. Despite being classified as ncRNA, the relevance of the protein-coding potential of lncRNA is growing [101]. Nevertheless, current evidence infers that lncRNAs mainly regulate cellular processes through the interaction with various other molecules, such as DNA, RNA, and proteins [102,103]. Having a much broader interactome than miRNAs, lncRNAs can also control chromatin structure, methylation status, sequestration of miRNAs, assembly or disruption of protein complexes, and post-translational modifications [100,103]. Another feature of lncRNAs is their tissue-and condition-specific (e.g., cancer-specific) expression pattern [100]. The first lncRNAs, H19 and Xist, were discovered in 1980s and 1990s, but they remained exceptions until the early 2000s when characterization of ncRNAs started to outpace protein-coding genes [104]. Dysregulated lncRNAs are involved in all hallmarks of cancers, including sustained angiogenesis and deregulated cellular metabolism [105,106]. In addition, mounting evidence suggest their importance in ferroptosis regulation, as outlined below (Fig. 2).

Ferroptosis-stimulating lncRNAs
Tumor suppressive lncRNA P53RRA, also known as LINC00472, is downregulated in various cancers including lung, liver, colon, renal and breast cancers [107][108][109][110][111]. In lung cancer, it interacts with Ras GTPase-activating protein-binding protein 1 (G3BP1) in the cytosol [38]. This cytosolic P53RRA-G3BP1 interaction displaces p53 from the G3BP1 complex. In turn, p53 is retained in the nucleus, leading to cell-cycle arrest, apoptosis, and ferroptosis. P53RRA promotes ferroptosis and apoptosis by affecting transcription of several metabolic genes, including the  [38]. Furthermore, lncRNA GA binding protein transcription factor beta subunit 1 antisense RNA 1 (GABPB1-AS1) is upregulated by erastin in HCC cells. It inhibits the translation of GA binding protein transcription factor subunit beta 1 (GABPB1) protein, which acts as an activation subunit of transcription activator nuclear respiration factor 2, also called GA-binding protein (GABP). Downregulated GABPB1 protein leads to the downregulation of peroxiredoxin-5 peroxidase (PRDX5). The resulting suppression of the cellular antioxidant capacity causes accumulation of ROS and MDA, and reduction in cell viability [112].
Metallothionein 1D pseudogene (MT1DP) is a lncRNA that regulates erastin-induced ferroptosis through NRF2 [113]. Ectopic MT1DP expression in NSCLC upregulates ROS and MDA levels, increases intracellular ferrous iron concentration, and reduces glutathione levels in cancer cells exposed to erastin. These effects are achieved through downregulation of NRF2, indirectly via stabilization of miR-365a-3p that normally targets NRF2 mRNA. Interestingly, Gai and colleagues designed folate-modified liposome nanoparticles to enhance the bioavailability and the efficiency of the targeted delivery of both erastin and MT1DP. In vivo mice studies have shown promising results for erastininduced ferroptosis through this particular pathway in NSCLC [113].

Ferroptosis-inhibitory lncRNAs
LINC00336 is a nuclear lncRNA with oncogenic functions in lung cancer, including the regulation of ferroptosis [114]. It interacts with RNA-binding protein ELAV-like RNA-binding protein 1 (ELAVL1), which stabilizes the LINC00336 via binding adenylate and uridylate (AU)-rich elements (AREs), the signal regions that determine RNA stability. In addition, it is indirectly upregulated through the p53 signaling pathway since LSH increases ELAVL1 expression. When upregulated, LINC00336 acts as an endogenous sponge of miR-6852, thus preventing miRNA-induced downregulation of CBS. The result is inhibited ferroptosis in lung cancer cells, leading to enhanced cell proliferation, colony formation, and tumor formation. LINC00336 has been shown to decrease iron concentration, lipid ROS, and mitochondrial superoxide, and increases mitochondrial membrane potential [114].
Zhang et al. investigated the effects of chronic cadmium exposure -one of the causative factors of prostate cancer -on cellular growth and ferroptosis resistance in vitro and in vivo. After the cadmium exposure, the expression of ferroptosis-related proteins (particularly GPX4, FTH1 and SLC7A11) was increased, suggesting that cadmium exposure confers ferroptosis resistance. These effects were preceded by upregulation of lncRNA OIP5-AS1 expression. OIP5-AS1 acts as an endogenous sponge of miR-128-3p to regulate the expression of SLC7A11 [39].
Nuclear enriched transcript 1 (NEAT1) is a well-known oncogenic perinuclear lncRNA that has significant roles in noncancerous diseases as well [115,116]. It is associated with several hallmarks of cancers including proliferation, cell cycle, invasion, migration and apoptosis [117]. Wu et al. found that NEAT1 is capable of binding to ACSL4 mRNA, thus reducing the expression level of this pro-ferroptotic enzyme in NSCLC [56]. While NEAT1 contributes to apoptosis, its role in ferroptosis in NSCLC seems to be independent from it. Also, its contribution to ferroptosis is mediated exclusively via ACSL4 as erastin induction does not significantly affect other ferroptotic players, such as SLC7A11, GPX4, and TfR1 levels [56]. Table 2 provides additional lncRNAs involved in ferroptosis regulation in cancer.

Ferroptosis-related lncRNAs in the prediction of therapy responses and outcomes
In addition to the above-mentioned lncRNAs with confirmed regulatory mechanisms, RNAseq investigations associated many other lncRNAs with ferroptosis. For example, signatures consisting of eight to twelve differentially expressed lncRNAs were shown to be independent prognostic factors for overall survival (OS) in breast cancer [118,119]. In the study from Zhang et al., patients with high-risk score had worse prognosis when treated with endocrine therapy, anthracycline, cyclophosphamide or paclitaxel, but not anti-HER2 therapy. In general, their tumors were immunologically cold due to inactivation of immune-related pathways and reduced tumor's immune cells infiltration [119]. Similarly, Yao et al. correlated seven ferroptosis-related lncRNAs with clinical prediction of prognosis and immunotherapeutic responses in LUAD [120]. In the same fashion, Jian et al. developed ferroptosis-related lncRNAs signature for glioma consisting of 15 lncRNAs. They also showed that patients in high-risk group had lower tumor purity, higher infiltration of immunosuppressive cells, and higher expression of immune checkpoints. In addition, those patients had no survival benefits of radiotherapy, compared to the low-risk group [121].

Inhibits ferroptosis in benign hemangioma
Sponged by miR-497-5p. Upregulates SLC7A11 and GPX4 expression [177] A. Balihodzic et al. independent prognosis factors for OS [122]. It was revealed that those novel ferroptosis-related lncRNAs in HNSCC may regulate immune and tumor-related pathways, particularly the expression of PD-1, CTLA4, LAG3, and BTLA [122]. Hence, combining immune checkpoint inhibitors with ferroptosis inducers may synergistically reduce cancer growth [123]. There is limited ongoing research that explores the relationship between immune checkpoints, radiotherapy and ferroptosis. Therefore, lncRNAs implicated in these processes should be investigated further. Nevertheless, all abovementioned signature profiles need further validation using different cohorts and complete molecular characterization before having a potential of being used as biomarkers or targets for novel medications.
The role of circRNAs in the regulation of ferroptosis in cancer CircRNAs are single-stranded, covalently closed ncRNA molecules with distinct characteristics from other ncRNAs [124]. Their existence was first reported several decades ago. Initially, it was believed that they are merely splicing-associated noise that arises from irregular splicing and represents procedural errors. Therefore, their biological relevance was initially underappreciated [125]. Generally being classified as non-coding molecules, it has been found that some circRNAs have AUG sites and may be abundantly associated with polysomes [125,126]. Similar to lncRNAs, circRNAs seem to be highly conserved and exhibit tissue-specific expression [127,128]. Fairly contrary to the linear miRNAs and lncRNAs, circRNAs are exceptionally stable thanks to their circular nature, leaving them without free ends to be degraded by exonucleases. Due to their increased stability, circRNAs can be found in exosomes and extracellular fluids, such as saliva and plasma. Therefore, they have great biomarker potential [129].
While some circRNAs (e.g., intron-containing circRNAs) are found only in the nucleus where they might play a role in transcription regulation, most circRNAs are located in the cytoplasm [125]. There they usually function as miRNA sponges. Individual circRNAs can bind to multiple miRNAs that regulate different pathways [124,125]. This feature is of particular importance for the potential therapeutic purposes. Importantly, miRNA-circRNA interactions might not always result in miRNA suppression, but also vice versa. Therefore, circRNAs may also function as miRNAs transportation or reservoir agents [124,130]. However, growing caution stands for the alteration of miRNAs' activity via sequestration by other ncRNAs (e.g., circRNAs and lncRNAs), the phenomenon called competing endogenous RNA (ceRNA) hypothesis [131]. Recent findings alert that physiological and even pathological changes in individual ceRNA expression are usually insufficient in significant miRNA activity suppression [132][133][134]. This limitation is based on the fact that individual ceRNA constitute only a small fraction of miRNAs' large target pool. Moreover, mathematical models assert that optimal ceRNA inhibition occurs when miRNA and targets are at near equimolar concentrations [135][136][137]. Many previously published studies have used supraphysiologic concentrations of transfected oligonucleotides or expression vectors that frequently exceed total cellular concentrations of their natural counterparts. This clearly suggests an overestimation of ceRNA activity and demand for better molecular models. Lastly, ceRNA as the appealing and straightforward approach in ncRNA studying may potentially hinder researchers' consideration of other confirmed ncRNAs' mechanisms of action. Nevertheless, circRNAs are important regulators of numerous normal and pathological cellular processes and diseases, including cancer [138]. So far, circRNAs have been associated with several hallmarks of cancers, including sustained proliferative signaling, evasion of growth suppressors, angiogenesis, invasion and metastasis, and evading cell death and senescence [124,138]. Growing evidence associate them with ferroptosis (Fig. 2).

Ferroptosis-stimulating circRNAs
Three circRNAs capable of ferroptosis induction are cIARS, circKDM4C and circ_0000190. cIARS is derived from the IARS gene, and it is found to be highly expressed in HCC after sorafenib Table 3. Examples of circRNAs implicated in ferroptosis regulation in cancer.
Further, circKDM4C is downregulated in AML. Normally, it sponges miRNA let-7b-5p which targets p53. In addition to indirect upregulation of p53, circKDM4C, when not retrieved from the circRNA pool, is capable of ferroptosis induction via increasing cellular iron content, upregulation of ACSL4 and PTGS2, and downregulation of GPX4 and FTH1 [32].

Ferroptosis-inhibitory circRNAs
Xian et al. found that circular ATP binding cassette subfamily B member 10 (circABCB10) is upregulated in colorectal cancer (CRC) where it acts as a sponge to miR-326. Consequently, its target C-C motif chemokine ligand 5 (CCL5) is overexpressed and contributes to carcinogenic effects, including inhibition of apoptosis and ferroptosis [141]. CCL5 has already been associated with CRC development and progression [142]. However, its exact mechanism in ferroptosis is still not reported.
Moreover, GPX4 is, as mentioned above, considered to be pivotal regulator of ferroptosis, analogous to bcl-2 in apoptosis Fig. 3 NcRNA-based therapeutics and delivery systems. Several ncRNA-based therapeutics exist, including antisense oligonucleotides, siRNAs, shRNAs, miRNA mimics, anti-miRNAs, miRNA sponges, and therapeutic circRNAs (box 1). Their major limitations and side effects can expectedly be overcome by usage of unique delivery systems that include lipid and polymer nanoparticles, antibodies, bacteriophages, exosomes and viral vectors (box 2). Newer generation of ncRNA therapeutics with convenient clinical administration (box 3) are awaited candidates due to their numerous advantages, including utilization of existing cellular processing mechanisms, capability of multiple signaling pathway targeting and promising cost-effective production (box 4). siRNA, small interfering RNA; shRNA, short hairpin RNA; antimiRs, anti-microRNAs; circRNAs, circular RNAs. Created with BioRender. [145]. CircKIF4A, upregulated in papillary thyroid cancer, acts as a sponge of miR-1231, leading to GPX4 overexpression [146]. Table 3 provides an overview of additional circRNAs involved in ferroptosis regulation in cancer.

Ferroptosis and ncRNAs interplay: therapeutic potential in cancer
Remarkably, this form of cell death has been associated with numerous pathologic processes including neurodegeneration, liver and lung fibrosis, ischemia-reperfusion injuries in brain, heart, kidneys and organ transplantation [14,16]. Nevertheless, there is major evidence of its particular relevance in cancer. It has been shown that mesenchymal and dedifferentiated cancer cells, which are resistant to cancer therapeutics and apoptosis, are highly susceptible to ferroptosis inducers [147,148]. Hence, inducing ferroptotic cell death (e.g., by pharmacologic manipulation) may help to overcome resistance of malignant cells to chemotherapy and therefore has great potential for cancer treatment. Several strategies to specifically induce ferroptosis are already being tested. One option is to target key enzymes involved in ferroptosis in cancer cells. For example, pharmacologic and genetic inhibition of system x c by blocking SLC3A2 and SLC7A11 have shown promising results in mouse models with low toxicity [149][150][151][152]. Similarly, targeting FSP1 is a promising approach due to its irrelevance in normal mice development indicating a potential broad therapeutic window [42,153].
While GPX4 is expressed in most cancer cell lines, it is essential for various organs, including kidneys and neurons [145,154,155]. Therefore, GPX4 inhibitors (e.g., RSL3) should be delivered specifically to the cancer cells to prevent side effects. Indirect ferroptosis inducers such as erastin may have low solubility and labile metabolism in the complex human body [156]. Incorporation of ferroptotic inducer compounds into protective delivery systems, such as nanoparticles, may overcome this problem. In addition, nanoparticles delivering iron, peroxides, and ncRNAs targeting key inhibitors of ferroptosis into cancer cells are already actively being tested in vitro and in vivo studies. NcRNAs are notably emerging as they ultimately carry several advantages. They are naturally occurring molecules in cells meaning their therapeutic counterparts may utilize existing cellular metabolic pathways. Additionally, ncRNAs frequently target multiple genes within one and/or more pathways causing a broader yet specific anti-cancer response, such as the case with miR-15 and miR-16 cluster that regulates various anti-apoptotic and cell cycle players, including bcl-2, mcl1 and c-JUN [157]. Lastly, ncRNA therapeutics can be fairly easy chemically synthesized shaping them as costeffective medications of the future.

CONCLUSION
Recently characterized as unique form of regulated cell death, ferroptosis has already been associated with numerous diseasesabove all with cancer. However, our knowledge about ferroptosis is still fairly limited and many open questions remain. We still do not know the complete relationship between ferroptosis and other forms of regulated cell death that share some common upstream mechanisms, such as p53. In addition, redox-independent roles of iron as well as the roles of other metals (e.g., copper) are not completely ruled out in ferroptosis induction. In addition, the exact molecular events responsible for the execution of cell death via ferroptosis are not fully understood. This is particularly pronounced in our ignorance of molecular events that occur downstream of lipid peroxidation including crucial moment(s) when activated ferroptosis cannot longer be suppressed. Finally, specific markers of ferroptosis suitable for application in live cells and intact tissues are still lacking. Furthermore, ncRNAs are a heterogenous group of non-coding transcripts with exceptional regulatory and biomarker capacities. Only a fraction of annotated ncRNAs have been investigated in the context of ferroptosis and cancer. Nevertheless, current evidence suggests that ferroptosis is frequently inhibited in cancer through the deregulation of usually tightly controlled ncRNA networks, thereby aiding cancer cell survival and progression. Hence, artificial induction of ferroptosis carries a great therapeutic potential. Albeit in their infancies, emerging innovative discoveries in both fields are paving the exciting path toward the successful utilization of novel ferroptosis-modulating ncRNAtherapeutics in cancer.

DATA AVAILABILITY
Correspondence and requests for materials should be addressed to Martin Pichler.