Abstract

   A lack of promising targets leads to poor prognosis in patients with
   lung adenocarcinoma (LUAD). Therefore, it is urgent to identify novel
   therapeutic targets. The importance of the N^6-methyladenosine (m^6A)
   RNA modification has been demonstrated in various types of tumors;
   however, knowledge of m^6A-related proteins in LUAD is still limited.
   Here, we found that insulin-like growth factor 2 mRNA binding protein 3
   (IGF2BP3), an m^6A reader protein, is highly expressed in LUAD and
   associated with poor prognosis. IGF2BP3 desensitizes ferroptosis (a new
   form of regulated cell death) in a manner dependent on its m^6A reading
   domain and binding capacity to m^6A-methylated mRNAs encoding
   anti-ferroptotic factors, including but not limited to glutathione
   peroxidase 4 (GPX4), solute carrier family 3 member 2 (SLC3A2),
   acyl-CoA synthetase long chain family member 3 (ACSL3), and ferritin
   heavy chain 1 (FTH1). After IGF2BP3 overexpression, expression levels
   and mRNA stabilities of these anti-ferroptotic factors were
   successfully sustained. Notably, significant correlations between
   SLC3A2, ACSL3, and IGF2BP3 were revealed in clinical LUAD specimens,
   further establishing the essential role of IGF2BP3 in desensitizing
   ferroptosis. Inducing ferroptosis has been gradually accepted as an
   alternative strategy to treat tumors. Thus, IGF2BP3 could be a
   potential target for the future development of new
   biomaterial-associated therapeutic anti-tumor drugs.

   Keywords: m^6A RNA methylation, m^6A reader, Ferroptosis, Therapeutic
   target, Lung adenocarcinoma

   Abbreviations: LUAD, lung adenocarcinoma; m^6A, N^6-methyladenosine;
   IGF2BP3, insulin-like growth factor 2 mRNA binding protein 3; GPX4,
   glutathione peroxidase 4; SLC3A2, solute carrier family 3 member 2;
   ACSL3, acyl-CoA synthetase long chain family member 3; FTH1, ferritin
   heavy chain 1; PKM, pyruvate kinase M1/2; ROS, reactive oxygen species;
   METTL3, methyltransferase-like 3; FTO, fat mass-and obesity-associated
   gene; YTHDF2, YTH domain family 2; qRT-PCR, quantitative reverse
   transcription PCR; IB, immunoblotting; IHC, immunohistochemistry;
   ELISA, enzyme-linked immunosorbent assay; TMA, tissue microarray assay;
   METTL14, methyltransferase-like 14; WTAP, Wilms' tumor 1-associating
   protein; ALKBH5, alkB homolog 5; VIRMA, vir-like m^6A methyltransferase
   associated; HNRNPA2B1, heterogeneous nuclear ribonucleoprotein A2/B1;
   ZC3H13, zinc finger CCCH-type containing 13; RBM15, RNA binding motif
   protein 15; MDA, malondialdehyde

Graphical abstract

   [53]Image 1
   [54]Open in a new tab

1. Introduction

   Lung cancer is one of the leading causes of cancer mortality worldwide
   [[55]1]. Although modern technology has greatly improved the efficacy
   of lung cancer treatment over the last decade, five-year survival rates
   are still only approximately 10%–20% [[56]2]. Patients with the more
   severe metastatic stage IV lung cancer have a five-year survival rate
   of less than 5% [[57]3]. One of the key reasons for this is a lack of
   effective tumor targets for treating this disease. Lung adenocarcinoma
   (LUAD) is the most common pathological type of lung cancer [[58]4];
   however, knowledge of potential targets for treatment of LUAD is
   equally limited.

   As a class of potential targets in tumors, N^6-methyladenosine
   (m^6A)-related proteins include methyltransferases, demethylases, and
   methylation recognition proteins – called “writers,” “erasers,” and
   “readers,” based on their functions. m^6A-related proteins regulate the
   biological effects of m^6A, the most common RNA methylation
   modification [[59]5]. Important roles for m^6A have been demonstrated
   in various types of tumors, including those of liver and ovarian cancer
   [[60][6], [61][7], [62][8]]. Abnormalities of m^6A-related proteins are
   crucial for the occurrence, progression, metastasis, drug resistance,
   and recurrence of tumors [[63]9]. In LUAD, m^6A and m^6A-related
   proteins are also essential for tumorigenesis [[64][10], [65][11],
   [66][12]]. To develop a cost-effective biomaterial-associated
   anti-tumor strategy, evaluation of the therapeutic value of individual
   m^6A-related proteins is particularly important. However, to the best
   of our knowledge, no such work has yet been reported.

   m^6A readers are particularly important because they are terminal
   effectors that can recognize target m^6A-methylated RNA. Insulin-like
   growth factor 2 mRNA binding protein 3 (IGF2BP3), an m^6A reader that
   belongs to the IGF2BP family, is quite well conserved between species
   [[67]13]. Prior studies confirmed that IGF2BP3 is an RNA-binding
   protein modulated by m^6A methylation, and its function is to promote
   mRNA translation and stability [[68]14]. IGF2BP3 is highly expressed in
   at least 15 kinds of tumors, indicating a remarkable effect of IGF2BP3
   on tumor occurrence and development [[69]15]. In LUAD, IGF2BP3
   accelerates cell proliferation by regulating alternative splicing of
   pyruvate kinase M1/2 (PKM) [[70]16], non-coding RNA signaling [[71]17],
   and translational activation [[72]18]. Therefore, IGF2BP3 may play an
   oncogenic role in LUAD. However, comprehensive study to further define
   its necessity in the tumorigenesis of LUAD is still lacking.

   Ferroptosis is a new type of regulated cell death that is
   iron-dependent and distinct from other types of cell death, such as
   apoptosis, necrosis, and autophagy [[73]19]. The hallmarks of
   ferroptosis include continuous accumulation of intracellular lipid
   reactive oxygen species (ROS), which eventually destroy the continuity
   of the plasma membrane [[74]19,[75]20]. Ferroptosis has been implicated
   in cancers [[76]21], and the connection between m^6A and ferroptosis
   has raised enormous interest in cancer communities. Several studies
   have proved that m^6A-related proteins, such as methyltransferase-like
   3 (METTL3), fat mass-and obesity-associated gene (FTO), and YTH domain
   family 2 (YTHDF2), regulate ferroptosis in lung and many other cancers
   [[77][22], [78][23], [79][24]]. Nevertheless, the importance of
   m^6A-related proteins has attracted attention and it is unknown which
   of these proteins are suitable targets for anti-tumor biomaterials.

   In this study, m^6A-related proteins were comprehensively evaluated in
   LUAD through multi-omics studies. IGF2BP3 was identified to have
   clinical significance and be associated with ferroptosis in LUAD cells.
   The mechanism underlying the suppression of ferroptosis by IGF2BP3 and
   potential therapeutic small molecules targeting IGF2BP3 were
   subsequently investigated. Because of the critical pro-tumorigenic role
   of IGF2BP3 in LUAD, developing more therapeutic biomaterials may be
   helpful to treat LUAD in future.

2. Materials and methods

2.1. Clinical specimens

   Fifty clinical samples of paired adjacent and LUAD tissues were
   obtained from the Department of Biobank, Shanghai Chest Hospital,
   between January 2019 and December 2020. Patient information is listed
   in [80]Supplementary Table S1. Written informed consent was obtained
   from each patient, following the guidelines of the Declaration of
   Helsinki. This study was approved by the institutional ethics committee
   of Shanghai Chest Hospital.

2.2. Cell culture

   The human bronchial epithelial cell line BEAS-2B, human LUAD cell line
   NCI–H1299, human pancreatic cancer cell line SW1990, and human colon
   adenocarcinoma cell line HCT-116 were purchased from FuHeng
   Biotechnology (Shanghai, China). Cells were validated by short tandem
   repeat analysis, and cultured in Dulbecco's Modified Eagle's Medium
   (DMEM, HyClone, Logan, UT, USA) with 10% fetal bovine serum (FBS,
   HyClone) and 1% penicillin/streptomycin (Gibco, Grand Island, NY, USA)
   in an incubator with 5% CO[2] at 37 ​°C.

2.3. Animal experiments

   Igf2bp3^−/− mice were generated by Cyagen Biosciences (Guangzhou,
   China). Mettl3^−/− mice were obtained as described in our previous
   study [[81]25]. H1299 cells with or without IGF2BP3 overexpression were
   digested and adjusted to a density of 5 ​× ​10^6 ​cells/200 ​μL. Next,
   200 ​μL of cells were injected into the right armpit of each
   4–6-week-old athymic nude mouse (Jiesijie, Shanghai, China). The weight
   and tumor size of nude mice were measured. Each group contained five
   mice. After 2 weeks, mice were injected daily with dimethyl sulfoxide
   (DMSO, Beyotime Biotechnology, Shanghai, China) with or without
   imidazole ketone erastin (IKE, 50 ​mg/kg, MedChemExpress, Monmouth, NJ,
   USA) or rigosertib (RIG, 250 ​mg/kg, Selleck, Houston, TX, USA). Mice
   were euthanized 28 ​d after implantation. Tumor growth was monitored
   and sizes were calculated as follows: 0.5 ​× ​L ​× ​W^2, where L
   indicates length and W indicates width. All animal experiments were
   approved by the institutional ethics committee of Shanghai Chest
   Hospital.

2.4. Plasmids and drugs

   IGF2BP3 single-guide RNAs (sgRNAs) were subcloned into the lentiviral
   vector LentiCrisprV2 system. IGF2BP3^WT and IGF2BP3^ΔKH3–4 were
   generated by Zuorun Biotech (Shanghai, China). Mex-3 RNA binding family
   member A (MEX3A)-overexpressing plasmids were generated by BioVision
   Technology (Shanghai, China). METTL3-overexpressing and METTL3-knockout
   plasmids were obtained from our previous studies [[82]26,[83]27].
   Glutathione peroxidase 4 (GPX4) and solute carrier family 3 member 2
   (SLC3A2) overexpressing plasmids were generated by Zuorun Biotech
   (Shanghai, China). The sequences of sgRNAs are listed in
   [84]Supplementary Table S2. RSL3 (2 ​μM, MedChemExpress), erastin
   (40 ​μM, MedChemExpress), ferrostatin-1 (2 ​μM, MedChemExpress), and
   RIG (1 ​μM, Selleck) were used as treatments in cellular experiments.
   United Stated Food and Drug Administration (FDA)-approved drugs were
   obtained from Selleck.

2.5. Quantitative reverse transcription PCR (qRT-PCR), immunoblotting (IB),
immunohistochemistry (IHC), and enzyme-linked immunosorbent assay (ELISA)

   For qRT-PCR, TRIzol reagent (Invitrogen, Carlsbad, CA, USA) was used to
   extract total RNA, and cDNA was synthesized using HiScript III RT
   SuperMix (Vazyme, Nanjing, China). qRT-PCR was performed using a
   HiScript II One Step qRT-PCR SYBR Green Kit (Vazyme). Primer sequences
   are listed in [85]Supplementary Table S2. IB, IHC, and ELISA were
   performed using conventional protocols. Primary antibodies included:
   anti-METTL3 (Abcam, Cambridge, UK), anti-MEX3A (Sigma, St. Louis, MO,
   USA), anti-IGF2BP3 (Abcam), anti-HA (Abcam), anti-GPX4 (Abcam),
   anti-SLC3A2 (Abclonal, Wuhan, China), anti-acyl-CoA synthetase long
   chain family member 3 (ACSL3, Abclonal), anti-ferritin heavy chain 1
   (FTH1, Abclonal), anti-insulin like growth factor 2 mRNA binding
   protein 1 (IGF2BP1, Abcam), anti-YTHDF2 (Abcam), and
   anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH, Cell Signaling
   Technology, Danvers, MA, USA). Tissue microarray assay (TMA) slides
   were obtained and measurement of IHC scores was performed as described
   in our previous study [[86]26]. IGF2BP3, SLC3A2, and ACSL3 proteins
   were determined in tissues using ELISA kits (Yingxin Biotech, Shanghai,
   China) in accordance with the manufacturer's instructions.

2.6. Measurements of cell death, lipid ROS, malondialdehyde (MDA), and
propidium iodide (PI) staining

   Cells were stained with SYTOX Green (1 ​μM, Invitrogen) for 20 ​min and
   then analyzed by flow cytometry to measure cell death. Lipid ROS was
   measured by adding C11-BODIPY (1.5 ​μM, Invitrogen) for 20 ​min before
   cell harvest, and then assessed by flow cytometry. For intracellular
   MDA measurement, a Lipid Peroxidation MDA Assay Kit (Beyotime) was used
   according to the manufacturer's instructions. For PI staining, cells
   were stained with PI (Servicebio, Wuhan, China) for 10 ​min and
   photographed under a microscope (Leica, Wetzlar, Germany).

2.7. m^6A dot blot

   Total RNA was extracted with Trizol reagent and denatured by heating at
   100 ​°C for 10 ​min, followed by chilling on ice immediately. Next, RNA
   was spotted on Biodyne Nylon Transfer Membranes (Pall, New York, NY,
   USA) and cross-linked by 365 ​nm UVP for 10 ​min m^6A levels were
   measured using an anti-m^6A antibody (Synaptic Systems, Goettingen,
   Germany) through IB.

2.8. RNA immunoprecipitation (RIP)

   A Magna RIP Kit (Millipore, Billerica, MA, USA) was used. Briefly,
   cells were lysed by lysis buffer and incubated with magnetic beads
   loaded with 5 ​μg of the following primary antibodies: anti-HA (Abcam),
   anti-IGF2BP3 (Abcam), anti-m^6A antibody (Synaptic Systems), or IgG
   (Beyotime Biotechnology) at 4 ​°C overnight. Next, cells were digested
   at 55 ​°C for 30 ​min using proteinase K. Finally, the remaining RNA
   was extracted using TRIzol reagent and measured by qRT-PCR.

2.9. Measurement of mRNA decay

   Cells were treated with actinomycin D (ActD, 5 ​μg/mL, MedChemExpress)
   before detection of mRNA expression using qRT-PCR at the indicated
   time.

2.10. RNA pull-down experiments

   Probes containing partial SLC3A2 mRNA with or without m^6A methylation
   were obtained from Generay (Shanghai, China). Probes were incubated
   with cell lysates at 4 ​°C overnight. Subsequently, streptavidin
   magnetic beads (MedChemExpress) were added to the reaction for 3 ​h.
   After washing, the enriched proteins were subjected to IB.

2.11. Luciferase reporter assay

   Regions encoding GPX4, SLC3A2, ACSL3, and FTH1 were inserted into the
   pmir-GLO luciferase reporter plasmid, which is often used to confirm
   putative m^6A sites [[87]25,[88]26,[89]28], and generated by Zuorun
   Biotech. Luciferase activities were detected by a dual luciferase
   reporter gene assay kit (Promega, Madison, WI, USA). Firefly luciferase
   activities were normalized to Renilla luciferase activity.

2.12. Cell proliferation assay

   Cell proliferation was analyzed with a Cell Counting Kit 8 (CCK-8)
   (Servicebio). Briefly, 6 ​× ​10^3 ​cells were added into 96-well plates
   for culture. After cells adhered to the well, the required reagents or
   drugs were added. The absorbance value of each well at 450 ​nm was
   measured with a microplate reader (BioTek, Winooski, VT, USA).

2.13. Bioinformatics

   Expression levels of m^6A-related proteins were analyzed using GEO
   DataSets ([90]http://www.ncbi.nlm.nih.gov/geo/) [[91]29]. Expression
   levels of genes in normal and tumor tissues were analyzed using GEO
   DataSets and the starBase database ([92]http://starbase.sysu.edu.cn/)
   [[93]30], GEPIA database ([94]http://gepia.cancer-pku.cn/) [[95]31],
   and Ualcan database ([96]http://ualcan.path.uab.edu) [[97]32].
   Interconnections among proteins were determined using the STRING
   database ([98]https://string-db.org) [[99]33]. The GEPIA database,
   Kaplan–Meier plotter database ([100]https://kmplot.com) [[101]34], and
   OncoLnc database ([102]https://www.oncolnc.org/) [[103]35] were used to
   analyze survival information. To identify targets of IGF2BP3, data were
   analyzed using enhanced crosslinking immunoprecipitation and
   high-throughput sequencing (eCLIP-seq), transcriptomic RNA sequencing
   (RNA-seq), and methylated RNA immunoprecipitation sequencing
   (MeRIP-seq/m^6A-seq) data, which were extracted from GEO DataSets and
   the m6a2target database ([104]http://m6a2target.canceromics.org)
   [[105]36]. Potential m^6A sites were predicted using the online tool
   SRAMP ([106]http://www.cuilab.cn/sramp) [[107]37]. Gene ontology (GO)
   and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment
   analysis were performed by DAVID ([108]https://david.ncifcrf.gov)
   [[109]38,[110]39].

2.14. Statistical analysis

   Student's t-test, one-way analysis of variance (ANOVA), two-way ANOVA,
   or Pearson correlation analysis were used for statistical analysis.
   Data are presented as mean ​± ​standard deviation (SD). ∗p ​< ​0.05,
   ∗∗p ​< ​0.01, and N.S. means no statistical significance.

3. Results

3.1. High expression of IGF2BP3 indicates poor prognosis in LUAD

   A schematic presentation of the experimental workflow to identify and
   verify promising therapeutic targets among m^6A-related proteins in
   LUAD is shown in [111]Fig. 1. First, we explored the expression
   profiles of 20 well-established m^6A-related proteins in LUAD from the
   GEO database, including METTL3, methyltransferase-like 14 (METTL14),
   methyltransferase-like 16 (METTL16), Wilms' tumor 1-associating protein
   (WTAP), FTO, alkB homolog 5 (ALKBH5), YTH domain containing 1/2
   (YTHDC1/2), YTH domain family 1/2/3 (YTHDF1/2/3), insulin-like growth
   factor 2 mRNA binding protein 1/2/3 (IGF2BP1/2/3), KIAA1429 (also
   called vir-like m^6A methyltransferase associated, VIRMA),
   heterogeneous nuclear ribonucleoprotein A2/B1 (HNRNPA2B1),
   heterogeneous nuclear ribonucleoprotein C (HNRNPC), zinc finger
   CCCH-type containing 13 (ZC3H13), RNA binding motif protein 15 (RBM15),
   and RNA binding motif protein 15B (RBM15B). We noticed that only
   IGF2BP3 was highly expressed in LUAD compared with normal or adjacent
   lung tissues by analyzing [112]GSE40275, [113]GSE7670, and [114]GSE1037
   datasets ([115]Fig. 2A). To confirm the importance of IGF2BP3 in LUAD,
   IGF2BP3 expression levels were evaluated in the [116]GSE32863 dataset
   ([117]Fig. 2B) and using the Ualcan database ([118]Fig. 2C). As
   expected, IGF2BP3 was elevated in LUAD compared with adjacent and
   normal lung tissues ([119]Fig. 2B and C). Because blocking tumor
   progression is an ideal anti-tumor approach, we examined IGF2BP3
   expression profiles in LUAD patients at distinct stages and found that
   expression of IGF2BP3 was higher in the late stages (II to IV) than the
   early stage (I, [120]Fig. 2D), suggesting that IGF2BP3 is critical for
   tumor progression. Using our own LUAD cohort (n ​= ​192), IGF2BP3
   expression levels in tumors were much higher than that of adjacent lung
   tissues ([121]Fig. 2E–G). To link IGF2BP3 expression with clinical
   outcomes in LUAD, we investigated the role of IGF2BP3 in patient
   survival and found that higher IGF2BP3 expression indicated poorer
   overall survival ([122]Fig. 2H and I). In addition to IGF2BP3 was
   upregulated in LUAD, its expression levels were associated with smoking
   history in patients with LUAD ([123]Supplementary Fig. S1).

Fig. 1.

   [124]Fig. 1
   [125]Open in a new tab

   Schematic of experimental workflow.

   In this study, we evaluated the expression profile of m^6A-related
   proteins in LUAD to uncover potential biomaterial-associated targets.
   IGF2BP3 was identified as a significant potential target by analyzing
   publicly available GEO data. Subsequently, targets of IGF2BP3 were
   extracted by analyzing publicly available eCLIP-seq, RNA-seq, and
   m^6A-seq data. The 1925 IGF2BP3 mRNA targets were further evaluated
   through GO and KEGG pathway enrichment analysis. The results
   demonstrate that IGF2BP3 might be associated with ferroptosis. Further
   experiments were performed to verify that IGF2BP3 is a key potential
   target for the development of a therapeutic drug to induce ferroptosis
   and treat LUAD. Finally, rigosertib was identified as a potent
   anti-tumor small molecule that exerts its role via reducing IGF2BP3
   expression.

Fig. 2.

   [126]Fig. 2
   [127]Open in a new tab

   High IGF2BP3 expression indicates poor prognosis in LUAD.

   (A) Venn diagram showing overlap of m^6A-related proteins altered in
   LUAD as revealed by the distinct GEO datasets indicated.

   (B) Expression of IGF2BP3 in paired adjacent–tumorous tissues in LUAD
   patients as revealed by [128]GSE32863.

   (C) Expression of IGF2BP3 in normal and LUAD tissues as analyzed using
   TCGA data and the Ualcan database.

   (D) IGF2BP3 expression profiles by tumor stages in LUAD as analyzed in
   the GEPIA database.

   (E) Representative IHC images of IGF2BP3 in paired adjacent–tumorous
   tissues from TMA. Black scale bar, 200 ​μm. Red scale bar, 50 ​μm.

   (F) Heat map showing IHC scores obtained from TMA that were evaluated
   using anti-IGF2BP3 antibodies.

   (G) Statistics of the IGF2BP3 IHC scores from TMA.

   (H, I) Overall survival curves obtained from GEPIA or Kaplan–Meier
   plotter databases based on IGF2BP3 expression in LUAD patients.

   Statistical analysis was performed using t-tests (B, C, and G), or
   one-way ANOVA (D).

   Collectively, the above results reveal that IGF2BP3 is commonly
   upregulated in LUAD and higher IGF2BP3 correlates with poorer overall
   survival.

3.2. IGF2BP3 is associated with ferroptosis

   We next investigated the function of IGF2BP3. Because IGF2BP3 is an
   RNA-binding protein and its function relies on m^6A modification, we
   wanted to determine which mRNAs can be bound and regulated by IGF2BP3.
   Expression levels of these RNAs should be manipulated by m^6A writers,
   such as METTL3, a key factor of the m^6A methyltransferase complex. To
   address this, we comprehensively analyzed eCLIP-seq data using
   anti-IGF2BP3 antibodies ([129]GSE92220), RNA-seq comparison of mRNA
   expression profiles before and after IGF2BP3 knockdown (KD,
   [130]GSE90684), and m^6A-seq to reveal which mRNAs were m^6A methylated
   after METTL3 knockdown (KD, [131]GSE102493). Through the above
   strategies, we identified 1925 potential target mRNAs of IGF2BP3
   ([132]Fig. 3A). Subsequently, these targets were used as inputs for GO
   and KEGG pathway enrichment analysis ([133]Fig. 3B–E). GO contains a
   wide variety of biological ontologies, providing a three-layer system
   to describe the function of gene products, namely: cellular component
   (CC), molecular function (MF), and biological process (BP). Cell
   components are used to describe the location of gene products in cells,
   molecular functions mostly refer to the function of a single gene
   product, and biological processes refer to an orderly biological
   process with multiple steps. GO is a pure gene set, which does not
   define the relationship between genes. In comparison, KEGG has a gene
   set, but also defines the complex relationship between genes. Using GO,
   we found that IGF2BP3 is multifunctional and may be critical for
   sustaining transformative phenotypes in tumor cells. As ferroptosis is
   a newly identified form of cell death, its induction is considered a
   new and alternative method to treat tumors [[134]21]. Thus, the
   involvement of ferroptosis, as predicted by KEGG, raised our particular
   interest ([135]Fig. 3E). According to the results shown in [136]Fig. 3A
   and E, 12 potential ferroptosis-associated targets of IGF2BP3 were
   picked up and further analyzed using STRING, a useful tool to predict
   protein-protein interactions. We found that the products of these 12
   genes are close interactors ([137]Fig. 3F). Because nuclear receptor
   coactivator 4 (NCOA4), SLC3A2, GPX4, ACSL3, and FTH1 were previously
   established to control ferroptosis in tumor cells [[138][40],
   [139][41], [140][42], [141][43]], we focused on the role of IGF2BP3 in
   ferroptosis in the current study.

Fig. 3.

   [142]Fig. 3
   [143]Open in a new tab

   Prediction of IGF2BP3 targets and their functions.

   (A) The 1925 IGF2BP3 mRNA targets, as revealed by comparing eCLIP-seq,
   RNA-seq, and m^6A-seq data from GEO and m6a2target databases.

   (B–E) Possible biological process (B), cellular component enrichment
   (C), molecular function (D), and involved pathways (E) were predicted
   by GO and KEGG enrichment analysis.

   (F) Associations between the 12 IGF2BP3 targets and KEGG-predicted
   ferroptosis-associated genes were reanalyzed using STRING.

3.3. IGF2BP3 desensitizes ferroptosis by sustaining anti-ferroptosis factors

   Next, we examined whether IGF2BP3 modulates ferroptosis and regulates
   its targets. Because IGF2BP3 is an m^6A reader, we wondered how IGF2BP3
   exerts its function in an m^6A-dependent manner. To address this, we
   generated LUAD H1299 ​cells with IGF2BP3 either overexpressed or
   knocked out (KO, [144]Fig. 4A). To avoid competitive interference in
   H1299 ​cells by endogenous IGF2BP3, we also reconstituted IGF2BP3
   expression in IGF2BP3–KO H1299 ​cells with wild-type (WT) or mutant
   IGF2BP3 without the KH3–4 m^6A-reading domain ([145]Fig. 4A). We found
   that overexpressing IGF2BP3 desensitized ferroptosis induced by RSL3
   and erastin, two pharmaceutical ferroptosis inducers. However,
   ferroptosis was sensitized when IGF2BP3 was knocked out ([146]Fig. 4B
   and C; [147]Supplementary Fig. S2A). Notably, compared with WT IGF2BP3,
   mutant IGF2BP3 without the KH3–4 m^6A-reading domain failed to rescue
   IGF2BP3–KO-induced sensitization of ferroptosis ([148]Fig. 4B and C),
   suggesting that the m^6A-reading domain of IGF2BP3 is essential for
   IGF2BP3 to suppress ferroptotic cell death. Excessive lipid
   peroxidation is the hallmark of ferroptosis. Our evaluation of lipid
   ROS revealed that, as observed for cell death, IGF2BP3 also suppressed
   lipid peroxidation in a manner dependent on the m^6A-reading domain
   ([149]Fig. 4D and E). To exclude non-ferroptotic effects,
   ferrostatin-1, a ferroptosis inhibitor, was used in parallel.
   Regardless of the IGF2BP3 status, ferroptotic changes induced by RSL3
   and erastin were successfully prevented by pretreatment with
   ferrostatin-1 ([150]Fig. 4B–E; [151]Supplementary Fig. S2A).

Fig. 4.

   [152]Fig. 4
   [153]Open in a new tab

   IGF2BP3 desensitizes ferroptosis and sustains anti-ferroptosis factors.

   (A) Verification of IGF2BP3 expression levels in control cells and
   H1299 ​cells with IGF2BP3 either overexpressed or knocked out (KO,
   upper panel). IGF2BP3 expression was also evaluated in IGF2BP3–KO
   H1299 ​cells reconstituted with either WT or IGF2BP3 without the KH3–4
   domain (IGF2BP3^ΔKH3–4, lower panel).

   (B–E) Measurements of cell death (B–C) and lipid peroxidation (D–E) by
   SYTOX Green and C11-BODIPY in control cells and H1299 ​cells under the
   indicated conditions through flow cytometry, with treatment of RSL3
   (2 ​μM, 12 ​h; B and D) or erastin (40 ​μM, 12 ​h; C and E). H1299
   cells were also pre-treated with or without ferrostatin-1 (2 ​μM,
   12 ​h) before RSL3 or erastin treatment. At least three replicates were
   carried out.

   (F) mRNA expression levels of ferroptosis-associated genes in control
   cells and H1299 ​cells with IGF2BP3 knockout (KO), as measured by qPCR.

   (G–K) Representative IB images of IGF2BP3, ACSL3, SLC3A2, GPX4, and
   FTH1 in H1299 ​cells (G–H and J) with indicated treatments or in tails
   from WT and Igf2bp3^−/− mice (I and K).

   Statistical analysis was performed using one-way ANOVA (B–E), or
   t-tests (F). Data are presented as mean ​± ​SD from three independent
   experiments. ∗p ​< ​0.05, ∗∗p ​< ​0.01, and N.S. indicates
   non-significance.

   We next assessed whether IGF2BP3 regulates the proteins identified as
   potential targets by comparing mRNA expression profiles before and
   after IGF2BP3 KO in H1299 ​cells. We found that seven of the 12
   potential targets illustrated in [154]Fig. 3F were altered
   ([155]Fig. 4F). Among these seven candidates, ACSL3, SLC3A2, GPX4, and
   FTH1 are known anti-ferroptotic proteins. Consistent with mRNA results,
   these proteins could also be positively regulated by IGF2BP3
   ([156]Fig. 4G), suggesting that protein alterations might result from
   changes in the mRNA. Interestingly, treatment of H1299 ​cells with RSL3
   and erastin led to the suppression of both IGF2BP3 and these
   anti-ferroptotic factors ([157]Fig. 4H), indicating that the alteration
   of IGF2BP3 might precede the alterations of its targets. In vivo data
   from mouse tails also confirmed that Igf2bp3 stimulates expression of
   anti-ferroptotic factors ([158]Fig. 4I). Because m^6A methylation is a
   prerequisite for the function of IGF2BP3, we next determined whether
   anti-ferroptotic factors are also regulated by m^6A. Indeed, these
   factors can be positively regulated by METTL3 in human LUAD cells and
   mice ([159]Fig. 4J and K). Collectively, IGF2BP3 desensitizes
   ferroptosis by sustaining the expression of anti-ferroptotic factors
   through its m^6A-reading function.

   To assess whether these anti-ferroptotic factors are downstream
   effectors of IGF2BP3, we overexpressed GPX4 and SLC3A2 in IGF2BP3 KO
   H1299 ​cells and evaluated whether such treatments affected ferroptosis
   sensitivity. We found that IGF2BP3 KO-sensitized RSL3 and
   erastin-induced cell death and lipid ROS accumulation could be
   partially reversed by overexpression of either GPX4 or SLC3A2
   ([160]Supplementary Figs. S2B and C). Moreover, colocalization of
   IGF2BP3 with GPX4, SLC3A2, ACSL3, and FTH1 were observed in
   H1299 ​cells ([161]Supplementary Fig. S2D), indicating that IGF2BP3 and
   its downstream effectors might closely interact.

   To further verify whether IGF2BP3 inhibits ferroptosis in vivo,
   xenograft mice were generated using control H1299 ​cells and
   H1299 ​cells overexpressing IGF2BP3. Following administration of IKE to
   mice, we found that IGF2BP3 prevented xenograft tissues from
   IKE-induced lift of MDA, which was one of the products of membrane
   lipid peroxidation ([162]Supplementary Fig. S2E), demonstrating that
   IGF2BP3 indeed suppresses ferroptotic lipid peroxidation in vivo.

3.4. IGF2BP3 recognizes m^6A-methylated mRNAs of anti-ferroptosis factors

   The fundamental role of IGF2BP3 is to recognize m^6A-methylated target
   mRNA. However, whether IGF2BP3 recognizes and binds m^6A-methylated
   mRNAs encoding anti-ferroptotic factors is unknown. Thus, potential
   m^6A sites within GPX4, SLC3A2, ACSL3, and FTH1 mRNAs were first
   predicted by SRAMP online software ([163]Fig. 5A–D). Next, sites A193
   of GPX4 mRNA, A1811 of SLC3A2 mRNA, A2564 of ACSL3 mRNA, and A361 of
   FTH1 mRNA, as well as their flanking regions, were selected for further
   analysis to investigate interactions between IGF2BP3 and
   m^6A-methylated anti-ferroptotic mRNAs because of their highest
   combined scores ([164]Fig. 5A–D). Notably, these m^6A sites are close
   to either the N- or C-terminus of the coding sequence (CDS,
   [165]Fig. 5A–D), suggesting that these are regulatory elements. In
   addition, the common m^6A motif among these anti-ferroptotic mRNAs was
   “GGACU,” which is also regarded as the consensus m^6A motif among
   species [[166]44], suggesting that our prediction results were
   reliable.

Fig. 5.

   [167]Fig. 5
   [168]Open in a new tab

   IGF2BP3 recognizes and binds m^6A-methylated GPX4, SLC3A2, ACSL3, and
   FTH1 mRNAs.

   (A–D) Prediction of m^6A sites within GPX4 (A), SLC3A2 (B), ACSL3 (C),
   and FTH1 (D) mRNAs by SRAMP online software.

   (E) The highest combined scored m^6A sites within GPX4, SLC3A2, ACSL3,
   and FTH1 mRNAs.

   (F) Consensus m^6A motifs of GPX4, SLC3A2, ACSL3, and FTH1 mRNAs.

   (G–J) RIP experiments using control IgG and anti-HA antibodies to
   examine IGF2BP3 binding to GPX4 mRNA (G), SLC3A2 mRNA (H), ACSL3 mRNA
   (I), and FTH1 mRNA (J) in H1299 ​cells expressing MEX3A-HA or
   IGF2BP3-HA.

   (K–N) RIP experiments using control IgG and anti-HA antibodies to
   examine IGF2BP3 binding with GPX4 mRNA (K), SLC3A2 mRNA (L), ACSL3 mRNA
   (M), and FTH1 mRNA (N) in IGF2BP3–KO H1299 ​cells reconstituted with WT
   or IGF2BP3^ΔKH3–4-HA.

   (O–R) RIP experiments using control IgG and anti-HA antibodies to
   examine IGF2BP3 binding with GPX4 mRNA (O), SLC3A2 mRNA (P), ACSL3 mRNA
   (Q), and FTH1 mRNA (R) in control cells and H1299 ​cells overexpressing
   METTL3.

   Statistical analysis was performed using t-test (G–R). Data are
   presented as mean ​± ​SD from three independent experiments.
   ∗∗p ​< ​0.01, and N.S. indicates non-significance.

   We next performed experimental verification. Regions encoding GPX4,
   SLC3A2, ACSL3, and FTH1 mRNA with (WT) or without (Mut) the putative
   m^6A site were cloned downstream of the Firefly luciferase gene of the
   pmir-GLO luciferase reporter plasmid ([169]Supplementary Fig. 3A). We
   found that luciferase activities from pmir-GLO plasmids containing Mut
   mRNAs were much weaker compared plasmids containing WT mRNAs
   ([170]Supplementary Figs. S3B–E), confirming that the m^6A sites within
   these anti-ferroptosis factors are genuine. RIP experiments were
   performed in H1299 ​cells expressing either exogenous IGF2BP3-HA or an
   m^6A-unrelated negative control MEX3A-HA ([171]Supplementary Fig. S3F).
   Enrichment of these anti-ferroptotic mRNAs in RIP products pulled down
   by anti-HA antibodies were notably higher in H1299 ​cells expressing
   IGF2BP3-HA compared with those expressing MEX3A-HA ([172]Fig. 5G–J),
   demonstrating that IGF2BP3 specifically binds anti-ferroptotic mRNAs.
   We also performed RIP experiments in IGF2BP3–KO H1299 ​cells
   reconstituted with either WT or IGF2BP3 without the KH3–4 domain, which
   confirmed that IGF2BP3 binding to anti-ferroptotic mRNAs relied on its
   m^6A-reading domain ([173]Fig. 5K–N). In addition to RIP experiments,
   RNA pull-down experiments using SLC3A2 probes with or without m^6A
   methylation revealed that IGF2BP3 binds to m^6A-methylated SLC3A2 mRNA
   ([174]Supplementary Fig. S3G). However, to the best of our knowledge,
   four other non-ferroptosis associated mRNAs did not bind with IGF2BP3
   ([175]Supplementary Fig. S3H), suggesting that IGF2BP3 may interact
   with anti-ferroptosis factors by a specific mechanism. By stimulating
   global m^6A methylation levels via overexpressing METTL3
   ([176]Supplementary Fig. S3I), we found that interactions between
   IGF2BP3 and anti-ferroptotic mRNAs were m^6A dose-dependent
   ([177]Fig. 5O–R). We also evaluated whether lifting global m^6A could
   affect ferroptosis sensitivity, and found that overexpressing METTL3
   only slightly suppressed RSL3 and erastin-induced cell death
   ([178]Supplementary Fig. S3J). This may result from expression of the
   cystine–glutamate antiporter (X[C]^−) system, which is essential for
   synthesis of the glutathione (GSH), an antioxidant that can be reduced
   following stimulation of global m^6A [[179]27,[180]45]. Such actions
   may offset the effects of IGF2BP3. Unfortunately, overexpression of
   IGF2BP3 did not alter global m^6A ([181]Supplementary Fig. S3K),
   demonstrating that IGF2BP3 merely acts downstream of m^6A. Altogether,
   these results suggest that pre-binding to m^6A-methylated mRNAs is
   essential for IGF2BP3 to modulate ferroptosis.

3.5. IGF2BP3 sustains the stability of anti-ferroptotic mRNAs, the
relationships of which correlate with clinical outcome

   Although IGF2BP3 recognizes and binds m^6A-methylated mRNAs encoding
   anti-ferroptotic factors, how IGF2BP3 regulates these mRNAs is unknown.
   By treating cells with ActD, which inhibits nascent mRNA transcription
   [[182]46], we evaluated the remaining intracellular mRNA in
   H1299 ​cells with or without IGF2BP3 overexpression. We found that the
   remaining mRNA levels of these anti-ferroptotic factors deceased much
   slower compared with control cells ([183]Fig. 6A–D), suggesting that
   IGF2BP3 sustains the mRNA stability of anti-ferroptotic factors.
   Similarly, overexpressing METTL3 prolonged the half-lives of these
   mRNAs ([184]Fig. 6E–H), indicating that m^6A is indispensable for the
   regulation of mRNA stability by IGF2BP3.

Fig. 6.

   [185]Fig. 6
   [186]Open in a new tab

   IGF2BP3 correlates with anti-ferroptosis factors and their clinical
   significance.

   (A–D) ActD chase experiments to analyze mRNA levels of GPX4 (A), SLC3A2
   (B), ACSL3 (C), and FTH1 (D) in control cells and H1299 ​cells
   overexpressing IGF2BP3, as measured by qPCR.

   (E–H) ActD chase experiments for analyzing mRNA levels of GPX4 (E),
   SLC3A2 (F), ACSL3 (G), and FTH1 (H) in control cells and H1299 ​cells
   overexpressing METTL3, as measured by qPCR.

   (I, J) Expression of SLC3A2 (I) and ACSL3 (J) in normal (n ​= ​59) and
   LUAD (n ​= ​515) using TCGA data and analysis with the Ualcan database.

   (K, L) Correlation between SLC3A2 and IGF2BP3 (K), and between ACSL3
   and IGF2BP3 (L) in LUAD (n ​= ​50).

   (M, N) Overall survival curves based on expression of SLC3A2 (M) and
   ACSL3 (N), as analyzed by the OncoLnc database.

   Statistical analysis was performed using two-way ANOVA (A–H), or
   Pearson correlation analysis (K–L). Data are presented as mean ​± ​SD
   from three independent experiments. ∗∗p ​< ​0.01.

   We next investigated the close interactions between IGF2BP3 and
   anti-ferroptotic factors in clinical samples. Our previous study
   [[187]27] demonstrated that YTHDC2 suppresses LUAD cells by indirectly
   inhibiting SLC3A2, leading to sensitization of ferroptosis. Therefore,
   we evaluated whether SLC3A2 is associated with other m^6A readers, such
   as IGF2BP3, in regulating ferroptosis. In addition, because few
   previous studies focused on ACSL3 and m^6A readers, we selected ACSL3
   for further study. We evaluated expression of SLC3A2 and ACSL3 in The
   Cancer Genome Atlas (TCGA) database. IGF2BP3 ([188]Fig. 2), SLC3A2, and
   ACSL3 were more highly expressed in LUAD compared with normal lung
   ([189]Fig. 6I and J). We also examined SLC3A2, ACSL3, and IGF2BP3 in
   our LUAD cohort (n ​= ​50) and found significant positive correlations
   between IGF2BP3 and SLC3A2, and between IGF2BP3 and ACSL3 ([190]Fig. 6K
   and L). Moreover, higher levels of SLC3A2 and ACSL3 indicated shorter
   overall survival ([191]Fig. 6M and N). These results demonstrate close
   relationships between IGF2BP3 and anti-ferroptotic factors that are
   indicative of poor clinical outcomes in LUAD.

3.6. Potential strategy to target IGF2BP3 and suppress tumorigenesis

   The aim of the present study was to identify therapeutic drugs that can
   target m^6A-related proteins. To achieve this, a library containing
   ∼1800 FDA-approved small molecules were screened to assess whether
   IGF2BP3 can be downregulated ([192]Fig. 7A). RIG was found to have the
   capacity to reduce IGF2BP3, but not other m^6A-related proteins
   ([193]Fig. 7B), suggesting that it might specifically inhibit IGF2BP3.
   Also, cotreatment with RIG sensitized H1299 ​cells to RSL3 and
   erastin-induced ferroptotic cell death ([194]Fig. 7C) and accumulation
   of lipid ROS ([195]Fig. 7D). We also evaluated whether administration
   of RIG was harmful to non-tumor cells by testing RIG in parallel in
   H1299 and BEAS-2B cells. As expected, RIG-impaired cell proliferation
   in BEAS-2B cells was much less obvious compared with H1299 ​cells
   ([196]Fig. 7E), suggesting that RIG administration is a safe strategy
   to treat tumors. In vivo mouse experiments also demonstrated that RIG
   has a strong capacity to reduce IGF2BP3 expression and tumor growth
   ([197]Fig. 7F). Taken together, we found that RIG is a safe drug to
   suppress LUAD, possibly by targeting IGF2BP3.

Fig. 7.

   [198]Fig. 7
   [199]Open in a new tab

   Rigosertib is a potent anti-tumor small molecule targeting IGF2BP3.

   (A) Workflow for screening small molecules that targeting against
   IGF2BP3·

   (B) Representative IB images of IGF2BP3, IGF2BP1 and YTHDF2 in
   H1299 ​cells treated with DMSO or rigosertib (1 ​μM, 24h).

   (C, D) Measurements of cell death (C) and lipid peroxidation (D) by
   SYTOX Green and C11-BODIPY in H1299 ​cells pretreated with DMSO or
   rigosertib (1 ​μM, 24 ​h) followed by treatment with or without DMSO,
   RSL3 (2 ​μM, 12 ​h), or erastin (40 ​μM, 12 ​h). H1299 cells were also
   pretreated with or without ferrostatin-1 (2 ​μM, 12 ​h). At least three
   replicates were carried out.

   (E) Cell proliferation was measured by CCK-8 assay in BEAS-2B and
   H1299 ​cells treated with DMSO or rigosertib (1 ​μM, 24h) for indicated
   time.

   (F) Xenograft mice were administrated DMSO or rigosertib (250 ​mg/kg)
   for the indicated time when tumors were obvious in mice. Tumor volumes
   were then graphed. IGF2BP3 was also measured by IB in tumors. N ​= ​5
   mice/group.

   (G) Schematic of the proposed model in the present study. Briefly, the
   m^6A reader IGF2BP3 recognizes and binds target mRNAs that can be
   m^6A-methylated by m^6A writers, such as METTL3 in LUAD cells.
   Subsequently, the stability of IGF2BP3 target mRNAs encoding
   anti-ferroptosis factors are sustained. Ferroptosis is thus suppressed
   and tumorigenesis is stimulated. Targeting IGF2BP3 using RIG or other
   potential biomaterial-associated therapeutic drugs may be a possible
   treatment for LUAD in the future.

   Statistical analysis was performed using one-way ANOVA (C, D), or
   two-way ANOVA (E, F). Data are presented as means ​± ​SD from indicated
   samples or experiments (n ​= ​3 for C, D, E, and n ​= ​5 for F).
   ∗∗p ​< ​0.01, and N.S. indicates non-significance.

   Finally, we evaluated whether IGF2BP3 plays a similar role to
   desensitize ferroptosis in tumor cells other than LUAD by knocking out
   or overexpressing IGF2BP3 in SW1990 ​cells (a pancreatic cancer cell
   line) and HCT116 ​cells (a colon adenocarcinoma cell line). Our results
   show that IGF2BP3 still desensitized SW1990 and HCT116 ​cells to
   ferroptotic cell death ([200]Supplementary Figs. S4A and B). In
   comparison, the anti-ferroptotic role of IGF2BP3 was very weak in
   BEAS-2B cells ([201]Supplementary Fig. S4C). These results demonstrate
   that the anti-ferroptosis function of IGF2BP3 is strong and common
   among tumor cells but may not be critical to suppress ferroptosis in
   non-tumor cells. Thus, once again our results emphasize that targeting
   IGF2BP3 might be useful to induce ferroptosis in tumor cells.

4. Discussion

   Although some potential targets have been revealed during studies of
   LUAD, therapeutic strategies for LUAD are still far from satisfactory.
   This ineffectiveness may largely occur because most strategies involve
   one-point to one-point targets. To enhance efficacy, it is urgent to
   develop multifunctional targets.

   To reveal such powerful targets, we focused on m^6A-related proteins.
   Through comprehensive multi-omics studies, IGF2BP3 was identified as
   being highly expressed in LUAD and positively controlling multiple
   essential regulators in the process of ferroptosis. As an m^6A reader,
   the biological effects of IGF2BP3 are ultimately executed through its
   specific recognition of m^6A-methylated RNAs [[202]47]. IGF2BP3
   modulates cancer progression by regulating cell proliferation, cell
   cycle, angiogenesis, glycolysis, metastasis, and immune escape
   [[203][48], [204][49], [205][50]]; however, whether IGF2BP3 can be
   regarded as a potent therapeutic target in LUAD is still unknown. The
   concept of IGF2BP3 as a critical LUAD regulator has gradually been
   accepted because a series of studies have reported specific effectors
   of IGF2BP3 in other tumorigenic processes in addition to ferroptosis
   [[206][51], [207][52], [208][53]]. In recent years, interactions
   between m^6A methylation and ferroptosis have garnered enormous
   interest in cancer research communities. Emerging studies have proved
   that m^6A related proteins are essential for modulating ferroptosis in
   tumors [[209][22], [210][23], [211][24],[212]27]. However, whether and
   how IGF2BP3 affects ferroptosis remains unknown until the relationship
   between IGF2BP3 and anti-ferroptosis factors is elucidated. Our study
   also demonstrates that IGF2BP3 can suppress ferroptosis by sustaining
   expression of anti-ferroptosis factors ([213]Fig. 7G). Herein, we
   provide further evidence suggesting that IGF2BP3 is a promising
   therapeutic target for LUAD treatment.

   Ferroptosis is related to iron-dependent accumulation of lipid
   hydroperoxides at lethal levels [[214]54]. Emerging evidence suggests
   that ferroptosis plays an important role in tumor development.
   m^6A-associated mechanisms have been linked to ferroptosis [[215][55],
   [216][56], [217][57]]. METTL3, METTL14, and FTO can reportedly modulate
   ferroptosis in tumors [[218][22], [219][23], [220][24]]. Notably,
   besides the currently reported m^6A reader IGF2BP3, the m^6A reader
   YTHDC2 was previously reported to be closely associated with
   ferroptosis [[221]27], thus hinting at critical roles for m^6A readers
   in modulating ferroptosis. GPX4, SLC3A2, ACSL3, and FTH1 are also
   pivotal anti-ferroptosis factors. GPX4 prevents ferroptosis by
   eliminating intracellular lipid ROS, and inhibition of GPX4 triggers
   ferroptosis [[222]58]. Increasing evidence suggests that GPX4 is
   related to the pathogenesis of tumors. GPX4 is highly expressed in
   tumorous tissues compared with normal tissues in nine types of tumors
   in TCGA, including LUAD [[223]59]. As the primary regulator preventing
   ferroptosis, GPX4 downregulation is a critical indicator of ferroptosis
   [[224]60]. SLC3A2 and solute carrier family 7 member 11 (SLC7A11) are
   two subunits of the cystine–glutamate antiporter X[C]^− system, and
   cystine is one of the major raw materials for the synthesis of the
   antioxidant GSH [[225]61]. Inhibition of SLC3A2 was previously reported
   to sensitize ferroptosis [[226]62]. Notably, our previous study
   demonstrated that SLC3A2 is regulated by m^6A by an indirect mechanism
   [[227]27]. SLC3A2 deficiency leads to impaired tumorigenesis in nude
   mice [[228]63]. ACSL3, a member of the acyl-CoA synthetase long chain
   family that activates free fatty acids to produce fatty acyl-CoAs,
   inhibits ferroptosis by activating exogenous monounsaturated fatty
   acids, which suppress ferroptosis [[229]64]. ACSL3 can also promote
   oleic acid secretion from adipocytes to inhibit lipid peroxidation and
   the ensuing ferroptotic cell death [[230]65]. FTH1 is the heavy subunit
   of ferritin, which plays an essential role in the maintenance of
   intracellular iron homeostasis to prevent harmful effects caused by
   iron overload [[231]66,[232]67]. Accordingly, inhibition of FTH1 plays
   a role in sensitizing ferroptosis. Excessive lipid peroxidation,
   unbalanced redox hemostasis, and labile iron overload are three major
   hallmarks of ferroptosis. Notably, GPX4, SLC3A2, ACSL3, and FTH1 cover
   all three of these aspects. The ideal cost-effective strategy to induce
   ferroptosis is to simultaneously inhibit these four anti-ferroptosis
   factors. Interestingly, this goal can be achieved by suppressing
   IGF2BP3 because of its common regulatory mechanism. Interestingly, the
   m^6A sites within GPX4, SLC3A2, ACSL3, and FTH1 mRNAs are close to
   either the N- or C-terminus of the CDS, suggesting that such
   localization may determine the fate of mRNA following m^6A methylation.
   In the present study, we noticed that recognition of m^6A-methylated
   mRNA by IGF2BP3 prolongs the half-lives of these mRNAs. Recently,
   several studies [[233][68], [234][69], [235][70]] pointed out that m^6A
   methylation of mRNA might be essential to maintain RNA stability,
   consistent with our conclusion.

   Because tumor cells are more sensitive to ferroptosis compared with
   normal cells, ferroptosis is a new and alternative method to treat
   cases of LUAD with resistance to conventional drugs. Therapeutic drugs
   inducing ferroptosis have been developed, some of which are undergoing
   experimental verification. For example, RSL3, ML162,
   diphenyleneiodonium chloride (DPI) compounds, FIN56, and FINO2 are
   known to target GPX4 [[236]54]. In addition, sorafenib, erastin and its
   derivatives are capable of inhibiting the X[C]^− system and preventing
   cystine uptake [[237]71], while buthionine sulfoximine and DPI2 have
   the capacity to deplete GSH [[238]72]. Although small molecules have
   been gradually discovered and developed, challenges still exist. First,
   most inhibitors have single targets, and those targets involve GPX4 or
   the X[C]^− system. Second, many inhibitors show poor (or unclear)
   pharmacological properties in animal models, limiting their clinical
   application potential [[239]73]. Last, but not least, whether these
   drugs are specific and leave normal tissue unharmed is unclear;
   accordingly, how to improve their specificity requires further
   exploration.

   Herein, we showed that a series of anti-ferroptosis factors are
   commonly regulated by IGF2BP3, indicating that IGF2BP3 exerts a wide
   regulatory mechanism to modulate ferroptosis. With the continuous
   progress of modern technology, targeting IGF2BP3 is not difficult. One
   option is RNAi agents, which are widely used with good tolerance and
   less damage to normal tissues. In addition, readily available
   FDA-approved small molecule libraries also exist, which can be used to
   screen possible drugs targeting IGF2BP3. Using such screening methods,
   we identified RIG as a potent small molecule that can suppress
   tumorigenesis by downregulating IGF2BP3. In addition, we experimentally
   confirmed that RIG is a safe drug that does not harm non-tumor cells.
   Thereby, we established a successful example of screening for drug
   targets against m^6A-related proteins to reduce tumorigenesis, and we
   may take advantage of IGF2BP3 as a promising target.

   Notably, IGF2BP3 is reported as being highly expressed in at least 15
   kinds of tumors [[240]15], thus targeting IGF2BP3 might be helpful for
   tumors other than LUAD. In summary, the efficacy of additional
   anti-IGF2BP3 drugs should be urgently verified to accelerate future
   clinical applications.

5. Conclusion

   IGF2BP3 is a key m^6A-related protein that is highly expressed in LUAD
   and associated with poor prognosis. IGF2BP3 desensitizes ferroptosis by
   sustaining m^6A-methylated mRNAs encoding anti-ferroptosis factors. Our
   discovery reveals a new potential strategy of targeting IGF2BP3 to
   inhibit LUAD by sensitizing ferroptosis.

Credit author statement

   Xin Xu: Conceptualization, Methodology, Formal analysis, Investigation,
   Data curation, Writing—original draft. Jiangtao Cui: Formal analysis,
   Investigation. Hong Wang: Formal analysis, Investigation. Lifang Ma:
   Data curation. Xiao Zhang: Data curation. Wanxin Guo: Investigation.
   Xiangfei Xue: Investigation. Yikun Wang: Investigation. Shiyu Qiu:
   Investigation. Xiaoting Tian: Methodology. Yayou Miao: Methodology.
   Mengyi Wu: Methodology. Yongchun Yu: Supervision. Yunhua Xu, Jiayi
   Wang, and Yongxia Qiao: Conceptualization, Methodology, Writing—review
   and editing, Supervision, Funding acquisition.

Declaration of competing interest

   The authors declare that they have no known competing financial
   interests or personal relationships that could have appeared to
   influence the work reported in this paper.

Acknowledgments