Crosstalk between noncoding RNAs and ferroptosis: new dawn for overcoming cancer progression

Cancer progression including proliferation, metastasis, and chemoresistance has become a serious hindrance to cancer therapy. This phenomenon mainly derives from the innate insensitive or acquired resistance of cancer cells to apoptosis. Ferroptosis is a newly discovered mechanism of programmed cell death characterized by peroxidation of the lipid membrane induced by reactive oxygen species. Ferroptosis has been confirmed to eliminate cancer cells in an apoptosis-independent manner, however, the specific regulatory mechanism of ferroptosis is still unknown. The use of ferroptosis for overcoming cancer progression is limited. Noncoding RNAs have been found to play an important roles in cancer. They regulate gene expression to affect biological processes of cancer cells such as proliferation, cell cycle, and cell death. Thus far, the functions of ncRNAs in ferroptosis of cancer cells have been examined, and the specific mechanisms by which noncoding RNAs regulate ferroptosis have been partially discovered. However, there is no summary of ferroptosis associated noncoding RNAs and their functions in different cancer types. In this review, we discuss the roles of ferroptosis-associated noncoding RNAs in detail. Moreover, future work regarding the interaction between noncoding RNAs and ferroptosis is proposed, the possible obstacles are predicted and associated solutions are put forward. This review will deepen our understanding of the relationship between noncoding RNAs and ferroptosis, and provide new insights in targeting noncoding RNAs in ferroptosis associated therapeutic strategies.


Introduction
Cancer progression including proliferation, metastasis and chemoresistance to drugs, has become serious obstacles in cancer therapy 1 . Although multiple therapeutic manners including operation, targeted therapy, chemotherapy, and radiotherapy have shown satisfactory performance, progression occurs since cancer cells dysregulate apoptosis pathways via various manners 2,3 . Therefore, new types of cancer therapy or drugs that eliminate cancer cells are urgently needed.
Ferroptosis is a type of programmed cell death discovered in 2012 4 . Unlike apoptosis, ferroptosis is characterized by excess reactive oxygen species (ROS) and intracellular iron 5 . Superabundant ROS induces peroxidation and disintegration of lipid membrane and cell death 6 . Regulation of ferroptosis mainly depends on neutral reaction between reduced glutathione (GSH) and ROS 7 . The exchange of glutamate and cystine is mediated by systemXc − , which is composed of solute carrier family 7 member 11 (SLC7A11) and solute carrier family 3 member 2 (SLC3A2), and offers the substrate cystine for GSH synthesis 8,9 . Glutathione peroxidase 4 (GPX4) catelyzes interaction between GSH and ROS to reduce intracellular oxidative stress 10 . Ferroptosis inducers can be divided into two classes based on regulation of neutral reaction to ROS. Class I ferroptosis inducers such as sorafenib, erastin and sulfasalazine, serve as blockers of systemXc − and result in a drop of GSH levels 11,12 . Class II ferroptosis inducers such as RSL3, FIN56, and ML162, inhibit function of GPX4 13,14 . Numerous studies have confirmed that ferroptosis inducers such as RSL3 and sorafenib eliminates cancer cells 15,16 . In addition, induction of ferroptosis via erastin and sulfasalazine improved effect of cytarabine and doxorubicin, and overcame cisplatin resistance of head and neck cancer 17,18 . This suggests that ferroptosis may become a new mechanism around which to design cancer therapy. However, use of ferroptosis in cancer therapy still faces obstacles. First, the specific mechanisms underlying ferroptosis and the interaction between ferroptosis and other processes, such as apoptosis, necrosis, and autophagy are not totally known, so how to control ferroptosis in cancer is in dark. Second, ferroptosis occurs in normal cells. Ferroptosis has been shown to induce the elimination of nerve cells in Parkinson's disease 19 . In addition, in acute kidney injury, ferroptosis participated in the death of renal tubular epithelial cells 20 . Therefore, use of ferroptosis inducers may generate complications. New regulatory factors should be recognized to understand the true appearance of ferroptosis in cancer.
For example, miR-675-5p promoted the metastasis of colorectal cancer cells via modulation of P53 26 . Moreover, lncRNA HOTAIR served as an enhancer in epithelial-tomesenchymal transition of breast cancer cells via competing with BRCA1 27 . In addition, circFOXO3 enhanced progression of prostate cancer through sponging miR-29a-3p 28 . However, roles of ncRNAs in ferroptosis have not been fully determined.
In this review, we focus on summarizing the ncRNAs which have been found to associate with ferroptosis regulators GSH, iron, nuclear factor (erythroid-derived 2)like 2 (NRF2) and ROS in cancer 5 . Moreover, we predict the obstacles that may limit the exploration of ncRNAs in ferroptosis in cancer therapy and offer advice for future studies. We believe that a comprehensive understanding of the interactions between ncRNAs and ferroptosis may benefit clinical therapeutics to cancer

MiRNAs and ferroptosis
MiRNAs exhibit functions by binding to the 3′untranslated regions of target mRNAs and suppressing their expression 29 . Some studies have revealed a relationship between miRNAs and ferroptosis. In radioresistant cells, miR-7-5p inhibited ferroptosis via downregulating mitoferrin and thus reducing iron levels 30 . Furthermore, miR-9 and miR-137 enhanced ferroptosis via reduction of intracellular GSH levels, miR-9 inhibited synthesis of GSH and miR-137 suppressed solute carrier family 1 member 5 (SLC1A5), a component of systemXc −31 . Moreover, miR-6852 which was regulated by lncRNA Linc00336, inhibited growth of lung cancer cells via promoting ferroptosis 32 . In the following sections, we will discuss the interactions between miRNAs and GSH, iron and NRF2 in cancer cells. The information of altered miRNAs in ferroptosis has been listed (Supplementary Table 1).

MiRNAs and GSH
GSH is a scavengerof ROS and protects lipid membrane 33 . Under physiological conditions, concentration of reduced GSH is about 10-100-fold more prevalent than the oxidized form. Under oxidative stress, reduced GSH is converted to oxidized form 34 . Biosynthesis of GSH involves three steps: exchange of glutamic acid and cystine induced by systemXc − ; synthesis of 훾glutamylcysteine by glutamic acid and cysteine catalyzed via 훾-glutamylcysteine ligase (GCL); and synthesis of GSH via 훾-glutamylcysteine and glycine catalyzed by GSH synthetase 35 . Function of GSH includes detoxification of exogenous or endogenous dangerous compounds catalyzed by GSH-S-transferases (GSTs) and GPXs 36 . Current knowledge on relation between GSH and cancer are summarized in Table 1, and the schematic diagram of these interactions is shown in Fig. 1a. MiR-18a and miR-  46 218 decreased GSH levels via targeting GCL in hepatocellular carcinoma and bladder cancer 37,38 . Furthermore, in hepatocellular carcinoma and lung cancer, miR-152 and miR-155 decreased GSH levels via targeting GST 39,40 . In addition, miR-326 and miR-27a inhibited GSH levels in cancer cells via targeting other factors such as pyruvate kinase m 2 (PKM2), SLC7A11 and zinc finger and BTB domain containing 10 (ZBTB10) [41][42][43] . Additionally, downregulation of GSH by miRNAs such as miR-21, miR-24-2, miR-497 and miR-503 has been observed in different cancer types, however, the specific mechanisms were not explored [44][45][46][47] . These findings indicate that miRNAs repress GSH levels via control of synthesis and consumption. The upregulation of GSH induced by miRNAs has been well-explored. GST was targeted by different miRNAs including miR-124, let-7a-5p, miR-92b-3p, miR-129-5P, miR-144, miR-153-1/2, miR-302c-5p, miR-3664-5p, miR-3714, miR-513a-3p, miR-590-3p/5p, miR-130b, miR-186, and miR-133a/b. These miRNAs bound to the 3′-untranslated regions of GST mRNA and inhibited GST expression, thus blocking GSH consumption and resulting in accumulation of intracellular GSH [48][49][50][51] . It is worth mentioning that miR-133a/b served as effective suppressors of GST in different cancer types, such as bladder cancer, lung cancer, prostate cancer, colorectal cancer, ovarian cancer and head and neck carcinoma. Inhibition of miR-133a/b reversed both increased GSH and insensitivity to drugs [51][52][53][54] . Furthermore, GPX family members are targeted by miRNAs and results in defect of ROS neutralization. In one report, GPX4 was decreased by miR-181a-5p in osteoarthritis 55 . However, the relationship between GPX4 and miRNAs in cancer is still in dark. Only GPX2 and GPX3 have been found to be modulated by miRNAs such as miR-17, miR-17-3p, miR-196a, and miR-921 in colorectal cancer, prostate cancer, and lung cancer [56][57][58][59] . Overall, regulation of GSH by miRNAs occurs mainly through control of GST and GPX family members. Since GSH has been shown to participate in growth of tumors and chemoresistance to drugs which induce intracellular oxidative stress, miRNAs may regulate ferroptosis and control cancer progression via modulation of GSH.

MiRNAs and iron
Iron metabolism is another key factor in ferroptosis. Excessive iron increases ROS via Fenton reaction and ROS is neutralized by iron reversely 60 . Metabolism of iron mainly includes interaction between transferrin (TF) and its receptor (TFR), import of iron via divalent metal transporter 1 (DMT1), storage of iron as ferritin and ironsulfur cluster (ISC), and export of iron via ferroportin (FPN) 61,62 . The specific realtion between miRNAs and iron is summarized in Table 2, and the schematic diagram of these interactions are shown in Fig. 1b 206 cancer, targeting of DMT1 by miR-149 and miR-19a led to decreased iron import 63 . Furthermore, in colorectal cancer and hepatocellular cancer, TFR was targeted by miRNAs including miR-141, miR-145, miR-152, miR-182, miR-200a, miR-22, miR-31, miR-320, miR-758, and miR-194 [63][64][65] . This inhibition led to disruption of interaction between TF and TFR and the following decreased iron import. Thereinto, miR-194 suppressed the expression of both TFR and FPN in colorectal cancer 63 . FPN was also targeted by miR-150, miR-17-5p, miR-20a, and miR-492 in hepatocellular carcinoma, multiple myeloma, lung cancer, and prostate cancer, respectively [66][67][68] . Furthermore, ferritin which is composed of ferritin heavy chain (FHC) and ferritin light chain (FLC), is controlled by miRNAs 69 . FHC could be targeted by miR-200b, miR-181a-5p, miR-19b-1-5p, miR-19b-3p, miR-210-3p, miR-362-5p, miR-616-3p, and miR-638 in prostate cancer, resulting in decreased intracellular iron 65,70,71 . FLC could be targeted by miR-133a in colorectal cancer and breast cancer, and knockdown of miR-133a restored the reduced iron levels inside cancer cells 63,72 . Among the miRNAs that regulate iron levels, miR-210 serves as an important member. In colorectal cancer cells, miR-210 was activated by hypoxia and then targeted ISCU to alter intracellular iron homeostasis 73 . Furthermore, transfection of miR-210 decreased the uptake of iron via TFR suppression 74 . On the contrary, miRNAs can be modulated by iron. MiR-107, miR-125b, and miR-30d were inhibited by iron in   75,76 , and miR-146a, miR-150, miR-214-3p and miR-584 were increased by iron in ovarian cancer and neuroblastoma 76,77 . This phenomenon may derive from the induction of excess ROS by iron and the subsequent regulation of miRNAs transcription. Overall, different miRNAs regulate iron levels in various directions, and the imbalance of iron leads to run-away miRNA expression.

MiRNAs and ROS
In addition to factors above, miRNAs regulate ROS via other mechanisms. The information of miRNAs that are related to ROS in cancer is listed in Table 4. MiRNAs can positively regulate ROS levels. For example, miR-21 whose expression increased with tumor grade, has been identified to enhance ROS level in lung cancer, colorectal cancer, gastric cancer, hepatocellular carcinoma, ovarian cancer, and prostate cancer [108][109][110][111][112][113] . Mechanically, miR-21 targeted STAT3, proline oxidase (POX), and programmed cell death 4 (PDCD4) to induce oxidative stress [114][115][116] . Moreover, miR-146a has attracted much attention. In ovarian cancer, miR-146a repressed SOD2 expression and inhibited proliferation of cancer cells and enhanced chemosensitivity to drugs 117 . In lung cancer, suppression of miR-146a restored catalase and inhibited ROS induction, and protected cancer cells from cisplatin-induced cytotoxicity 118 . In addition, overexpression of miR-124, miR-526b, and miR-655 led to excess ROS via thioredoxin reductase 1 in breast cancer 119,120 . Furthermore, the antioxidant enzyme SOD1 was downregulated by stable expression of miR-143 or miR-145 in colorectal cancer 121 . This indicates that miRNAs enhance intracellular ROS via different manners. On the other hand, in lung cancer, miR-99 suppressed the invasion and migration of cancer cells via targeting NOX4-mediated ROS production 122 . Additionally, miR-520 and miR-373 reduced ROS via targeting NF-κB and TGF-β signaling pathways and repressed growth and lymph node metastasis of breast cancer 123 . Other miRNAs such as let-7, miR-137, miR-193b, miR-199, and miR-26a, have been found to decrease ROS level in cancer cells via diverse targets such as heme oxygenase-1, C-MYC, and triglyceride [124][125][126][127][128] , indicating that miRNAs inhibit ROS level. Conversely, miR-133a, miR-150-3p, miR-1915-3p, miR-206, miR-34, miR-638, and miR-182 were activated by oxidative stress and then played a role in the subsequent biological processes [129][130][131][132][133] . Moreover, miR-125, miR-145-5p, miR-17-5p, miR-199, and miR-17-92, were decreased by excess intracellular ROS [134][135][136][137] . Among them, miR-125b plays a dual role in oxidative homeostasis. As discussed above, miR-125b serves as a regulator of NRF2. In addition, miR-125b could be inhibited by ROS via a DNMT1-dependent DNA  140 . Moreover, although miR-21 has been discussed as the enhancer of ROS in breast cancer, DNA damage induced by ROS led to activation of miR-21 via NF-κB, indicating the interaction between miRNAs and ROS 138 . In total, we can infer that altered levels of GSH, iron, and NRF2 are not the only methods by which miRNAs regulate ROS and vice versa in, miRNAs and ROS can also regulate each other in various pathways.

LncRNAs and ferroptosis
LncRNAs mainly serve as regulators of transcription factors in nucleus or as sponges of miRNAs in cytoplasm 139 . Linc00336 was promoted by lymphoid-specific helicase in lung cancer and inhibited ferroptosis via sponging miR-6852 32 . Furthermore, in breast cancer and lung cancer, lncRNA P53rra bound to Ras GTPaseactivating protein-(SH3domain)-Binding Protein 1 (G3BP1) and displaced P53 from a G3BP1 complex, resulting in retention of P53 in nucleus and downregulation of SLC7A11 140 . In addition, ferroptosis inducer erastin upregulated lncRNA GA binding protein transcription factor subunit beta 1 (GABPB1) antisense RNA 1 (Gabpb1-AS1), which suppressed GABPB1 and led to downregulation of peroxiredoxin-5 peroxidase and suppression of cellular antioxidant capacity in hepatocellular carcinoma 141 . Interaction between lncRNAs and ferroptosis has been listed (Supplementary Table 1), and the relationship between lncRNAs and ferroptosis associated factors is summarized in Table 5. The schematic diagram of these interactions is shown in Fig. 1.

LncRNAs and ferroptosis associated factors
Since there are only a few studies about lncRNAs and ferroptosis factors, we will discuss them together. Regulation of GSH by lncRNAs in cancer mainly depends on GST and GCL 46 . In breast cancer, knockdown of lncRNA Ror led to reduced multidrug resistance-associated Pglycoprotein and GST expression, resulting in restored sensitivity of breast cancer cells to tamoxifen 142 . Similarly, in colorectal cancer, knockdown of lncRNA Xist inhibited doxorubicin resistance via suppressing GST and increasing GSH 48 . Furthermore, in hepatocellular carcinoma cells, silencing lncRNA Neat1 inhibited IL-6-induced STAT3 phosphorylation which contributed to the increase of GST 143 . In addition, lncRNA Linc01419 bound to the promoter region of GSTP1 and recruited DNA methyltransferase, increasing promoter methylation and decreasing GST expression in esophageal squamous cell carcinoma 144 . Moreover, knockdown of lncRNA H19 resulted in recovery of cisplatin sensitivity via reduction of GCL and GST 145 . In total, regulation of GSH by lncRNAs mainly depends on GST and GCL. Moreover, in hepatocellular carcinoma, silencing of lncRNA Pvt1 inhibited  134 TFR expression and obstructed iron uptake via miR-150 146 . Furthermore, silencing of FHC in leukemia cells induced production of ROS and altered downstream genes via increasing H19 and miR-657 expression 147 . This means that lncRNAs are associated with iron metabolism in cancer cells. Moreover, in bladder cancer, suppression of NRF2 by lncRNA associated transcript in bladder cancer (Aatbc) resulted in apoptosis 148 . In multiple myeloma, metastasis associated lung adenocarcinoma transcript 1 (Malat1) which has been proved to play a role in various cancers, inhibited NRF2 via activation of their negative regulator KEAP1 149 . Furthermore, overexpression of Keap1 regulation-associated lncRNA (Kral) inhibited NRF2 via increasing KEAP1 expression, and reversed the resistance of hepatocellular carcinoma cells to 5-fluorouracil 91 . Therefore, lncRNAs regulate NRF2 expression via direct and indirect manners. On the contrary, NRF2 participates in regulation of lncRNAs. In gallbladder cancer, downregulation of lncRNA loc344887 suppressed cell proliferation and decreased migration and invasion. Further studies found that loc344887 was upregulated after ectopic expression of NRF2 150 . In a recent study, NRF2 activated smoke and cancerassociated lncRNA 1 (Scal1) and induced oxidative stress protection. Knockdown of NRF2 suppressed Scal1 and alleviated the proliferation of lung cancer cells 92 . In sum, lncRNAs can regulate NRF2 by directly controlling expression or modulating KEAP1 indirectly, and NRF2 can regulate lncRNAs expression reversely.
Other than the factors above, lncRNAs regulate ROS levels via various mechanisms. In bladder cancer, lncRNA urothelial cancer associated 1 (Uca1) decreased ROS level via targeting miR-16 which led to decreased GSH synthetase 151 . Furthermore, in hepatocellular carcinoma, downregulation of H19 increased ROS via MAPK/ERK signaling pathway and reversed chemotherapy resistance 152 . Moreover, knockdown of lncRNA growth arrest specific 5 (Gas5) in melanoma enhanced intracellular ROS via increased superoxide anion and NADPH oxidase 4 (NOX4)-oxidized GSH 153 . In lung cancer cells, the intracellular oxidative stress induced by paclitaxel was attenuated by knockdown of maternally expressed 3 (Meg3), and Meg3 overexpression induced cell death and increased sensitivity to paclitaxel in an ROS-dependent manner 154 . In total, lncRNAs influence ROS metabolism via control of GSH, iron, NRF2 and other factors, and these factors can regulate lncRNAs expression reversely.

Other ncRNAs and ferroptosis
CircRNAs, tRNAs, rRNAs, piRNAs, snRNAs, and snoRNAs are also contained in family of noncoding RNAs 21 . However, studies on the relations between these ncRNAs and ferroptosis are few. The interactions have  Table 5 Summary of GSH, iron, NRF2, and ROS associated lncRNAs in cancer.  Table 2). The schematic diagram of these interactions is shown in Fig. 1.

CircRNAs
CircRNAs are covalently closed, single-stranded RNA molecules derive from exons via alternative mRNA splicing 22 . Several studies have uncovered function of cir-cRNAs in ferroptosis. In glioma, circ-TTBK2 enhanced cell proliferation and invasion and inhibited ferroptosis via sponging miR-761 and subsequent ITGB8 activation, knockdown of circ-TTBK2 promoted erastin-induced ferroptosis 155 . Furthermore, circ0008035 inhibited ferroptosis in gastric cancer via miR-599/EIF4A1 axis. Knockdown of circ0008035 enhanced anticancer effect of erastin and RSL3 via increased iron accumulation and lipid peroxidation 156 . According to ferroptosis associated factors, in gastric cancer, circPVT1 promoted multidrug resistance by enhancing P-gp and GSTP. MRNA levels of P-gp and GSTP were obviously repressed after downregulation of circ-PVT1 in paclitaxel-resistant gastric cancer cells 157 . Moreover, high-throughout microarraybased circRNA profiling revealed that 526 circRNAs were dysregulated in cervical cancer cells, and bioinformatic analyses indicated that these circRNAs participated mainly in GSH metabolism 158 . However, associated miRNAs and downstream factors were not screened. Thus, further studies on the modulation of ferroptosis by circRNAs are needed.

TRNAs
TRNAs serve as adapter molecules between mRNAs and proteins. The interaction between tRNAs and ferroptosis includes two possible manners. First, tRNAs are required in the synthesis of ferroptosis associated factors such as SLC7A11, GPX4, and IREB2, thus the mutation of tRNAs may alter the expression of these factors and then influence ferroptosis 217 . Second, tRNAs have multiple interaction partners including aminoacyl-tRNA-synthetases, mRNAs, ribosomes and translation factors 159 . Among them, cysteinyl-tRNA synthetase plays a role in ferroptosis. In fibrosarcoma, rhabdomyosarcoma and pancreatic carcinoma, loss of cysteinyl-tRNA synthetase suppressed erastin-induced ferroptosis via increasing intracellular GSH and transsulfuration, and inhibition of the transsulfuration pathway resensitized cells to erastin 160 . Interestingly, tRNAs mutation may control ferroptosis in an opposite manner. Selenocysteine which is formed from serine at the respective tRNA, is a component of GPXs. However, in hepatoma, colorectal cancer and breast cancer, the mutation of tRNA led to decline of selenoprotein expression except GPX4 and GPX1, and weak ferroptosis alteration [161][162][163] . This indicates that tRNAs modulate GSH levels mainly via synthesis but not metabolism. In addition, tRNAs influence ROS levels via various manners. Lung cancer mouse model with deletion of selenocysteine-tRNA gene exhibited ROS accumulation and increased susceptibility to lymph nodules metastasis 164 . Additionally, Queuine-modified tRNAs promoted cellular antioxidant defense via catalase, SOD, GPX, and GSH reductase and inhibited lymphoma 165 . In total, tRNAs decrease GSH synthesis and increase ferroptosis without modulating GPX4, while on the other hand, tRNAs enhance the antioxidant defense system and then inhibit ferroptosis.

RRNAs
RRNAs constitute the structural and functional core of ribosomes 166 . Some reports have provided clues for role of rRNAs in ferroptosis. In cervical cancer, NRF2 was found to contain a highly conserved 18S rRNA binding site on 5′ untranslated region that is required for internal initiation. Deletion of this site remarkably enhanced translation, indicating that the 18S rRNA regulates NRF2 expression 167 . In another study, hepatoma cells treated with ethidium bromide exhibited a 70% decrease in the 16S/18S rRNA ratio and enhanced NRF2 expression 168 . However, whether NRF2 and 18S rRNA are mutually regulated remains unclear. Regarding ROS, nuclear mitotic apparatus protein (NuMA) is involved in cellular events such as DNA damage response, apoptosis, and P53-mediated growth arrest. In breast cancer cells, NuMA bound to 18S and 28S rRNAs and localized to rDNA promoter regions. Downregulation of NuMA expression triggered nucleolar oxidative stress and decreased pre-rRNA synthesis 169 . Furthermore, in leukemia HL-60 cells treated with iron chelator deferoxamine, rRNA synthesis in nucleoli was inhibited 170 . In conclusion, interaction between rRNAs and ferroptosis has not been completely uncovered. Role of ribosomes as the place in which proteins related to ferroptosis are synthesized may provide clues for further studies.

PiRNAs, snRNAs, and snoRNAs
PiRNAs are the class of small ncRNA molecules distinct from miRNAs in that they are larger, lack sequence conservation, and are more complex 171 . PiRNAs are involved in tumorigenesis of variety cancers 172 . However, studies on piRNAs and ferroptosis are few. In prostate cancer, piR-31470 formed a complex with piwi-like RNAmediated gene silencing 4 (PIWIL4). This complex recruited DNMT1, DNA methyltransferase 3 alpha, and methyl-CpG binding domain protein 2 to initiate and maintain the hypermethylation and inactivation of GSTP1. Overexpression of piR-31470 inhibited GSTP1 expression and increased vulnerability to oxidative stress and DNA damage in human prostate epithelial RWPE1 cells, resulting in tumorigenesis 173 . However, the GSTP1 inactivation may inhibit tumor growth via induction of ferroptosis once the tumors are formed. Clearly, further studies are needed to explore the roles of piRNAs in different stages of cancer. SnoRNAs are a class of small RNA molecules that mediate modifications of rRNAs, tRNAs, and snRNAs. The snoRNA ACA11 was overexpressed in multiple myeloma cells, increasing ROS and resulting in protein production and cell proliferation 174 . There are currently no reports on ferroptosis and snRNAs which mediate post-transcriptional splicing in gene expression. In cervical cancer and osteosarcoma, assembly chaperones and core proteins devoted to snRNA maturation contributed to recruiting trimethylguanosine synthase 1 to selenoprotein mRNAs including GPX1 for cap hypermethylation 175 . Future studies should focus on the possible regulation of snRNAs towards GPX families. In sum, further studies are needed to explore functions of circRNAs, tRNA, rRNAs, piRNAs, snoRNAs and snRNAs in ferroptosis. Furthermore, the network of factors modulating ferroptosis remains to be established. As ferroptosis is a process of dynamic equilibrium, any alteration of the associated factors may intersect with others. For example, GSH maintains the cytosolic labile iron pool via formation of iron-GSH complexes 176 . In addition, GSH regulates iron trafficking, and inhibition of GSH synthesis leads to diminished iron efflux following nitric oxide exposure 177 . Moreover, iron is exported via multidrug resistant protein 1 (MRP1), a known transporter of GSH conjugates 178 . GSH depletion, MRP1 inhibition or MRP1 knock-out leads to decreased iron release upon nitric oxide treatment 179 . Conversely, the secondary increase in ROS induced by iron stimulates GSH production, indicating that iron and GSH are interconnected 46 . Moreover, targets of NRF2 play a critical role in mediating iron/heme metabolism. Both FTL and FTH, the key iron storage protein, as well as FPN, which is responsible for cellular iron efflux, are controlled by NRF2 180,181 . In addition, a number of integral GSH synthesis and metabolism related enzymes including both the catalytic and modulatory subunits of GCLC, GCLM, GSS, and SLC7A11, are under the control of NRF2 [182][183][184] . In total, regulation of ferroptosis are linked together, modulation of GSH, iron and NRF2 by ncRNAs may result in further change of each other, and finally alter ferroptosis process.

Clinical application potential of ncRNA-associated ferroptosis
Targeting ncRNAs in cancer has yielded some promising results, however, application of ferroptosis via an ncRNA-dependent manner in clinic is facing obstacles. Inadequate understanding of specific mechanisms results in the limited use of ncRNA modifiers in ferroptosis. Furthermore, cell death occurs in a variety of ways, and numerous ncRNAs may be simultaneously regulated, thus how to ensure that the alteration of associated ncRNAs leads to ferroptosis is another problem. Moreover, ncRNAs act in various ways that may intersect with ferroptosis. For example, ferroptosis inducer miR-210 and H19 could modulate autophagy via targeting BECN1, ATG7, SIRT1, and HIF-1α [185][186][187][188] . In addition, miR-146a could regulat ROS modulator catalase and SOD2 which repressed mitochondrial function 189,190 . Alteration of autophagy or mitochondrial function resulted in multiple pathologic changes such as neuroinflammation, neurodegeneration, vessel remodeling and myocardial fibrosis, thus how to overcome these possible complications should be considered [191][192][193][194] . In addition, some pathways such as the KEAP1-NRF2 axis, is inhibited by multiple miRNAs and lncRNAs and promotes ferroptosis. Nevertheless, the repression of KEAP1-NRF2 results in the defect in cleaning of ROS and leads to susceptibility to DNA damage and tumorigenesis 195,196 . To solve these problems, future studies should address the following points. First, more ncRNAs should be identified. A ferroptosis-associated ncRNA screening platform should be established to identify the spectrum of ferroptosis associated ncRNAs and those specific to certain cancers. Second, more intensive studies using complex molecular biological experiments, such as chromosome immunoprecipitation, RNA immunoprecipitation, RNA pulldown, luciferase assays, and RNA truncation should be performed to explore the precise roles of ncRNAs in ferroptosis. Third, in order to translate fundamental experimental results into clinic, functions of ncRNAs in ferroptosis should be tested in animal models. Transgenic mouse models should be established to verify the function of ncRNAs more clearly. Fourth, in order to ensure whether ferroptosis is modulated by ncRNAs, accurate detection of ROS and iron levels, and observation of mitochondrial morphology in tumor tissues are needed. Furthermore, primary culture of tumor cells from patients should be performed to explore whether the proliferation of cancer cells is enhanced by Fer-1, which is the specific inhibitor of ferroptosis. The involvement of ncRNAs in ferroptosis in cancer can be verified in knockdown or overexpression studies. Finally, since ferroptosis occurs in not only tumors but also normal tissues, and as above, ferroptosis regulation by ncRNAs may activate other biological processes and even increase the susceptibility to tumorigenesis. Thus, both ferroptosis-related ncRNAs and associated markers of cell death, senescence, and remodeling should be assessed in patients who are suitable for ferroptosis-associated therapy. In addition, adverse events, dose-limiting toxicities and therapeutic effects should be carefully monitored through rigorous detection of organ functions, imaging of vital organs and tumors, and hematological changes during the application of ferroptosis inducers in clinic. After all, as cancer is a developmental process, the collaboration between multidisciplinary teams should be made to obtain rational therapy regimens to enhance therapeutic effect and alleviate complications.

Conclusions and perspectives
Cancer cells may be intrinsically insensitive or evolve and develop resistance to apoptosis, resulting in cancer progression 197 . Under the development of molecular biological technologies, identification of new targets or methods to eliminate cancer cells has attracted substantial attention. Ferroptosis is a recently recognized form of programmed cell death that relies on excess intracellular ROS and consequent lipid peroxidation 198 . Ferroptosis has been successfully applied to limit tumor growth and overcome the resistance of cancer cells to apoptosis, indicating that it may be useful as a new therapeutic approach 3 . Nevertheless, the application of ferroptosis inducers in cancer therapy is limited, mainly because the specific mechanisms underlying ferroptosis remain unexplored.
NcRNAs have been proved to regulate gene expression by various manners. Numerous ncRNAs have been found to regulate behaviors of cancer cells. In recent years, researchers have examined some ferroptosis-associated ncRNAs in cancer cells. Nevertheless, the specific regulatory mechanisms have not been explored. Therefore, wider and deeper studies to explore the function of ncRNAs in ferroptosis are needed. In this review, the landscape of ncRNAs associated with ferroptosis in cancer thus far is summarized. In addition, possible obstacles during application of ncRNA-associated ferroptosis in clinic are put forward and associated solutions are suggested. However, the information summarized in this review is not sufficient to support the application of ferroptosis inducers in cancer, more ncRNAs should be identified and deeper researches should be performed. In conclusion, ncRNAs may become markers to filter cancer patients who are fit for ferroptosis therapy and become therapeutic targets of ferroptosis inducers.