A lncRNA from the FTO locus acts as a suppressor of the m⁶A writer complex and p53 tumor suppression signaling

Summary

N⁶-methyladenosine (m⁶A) of mRNAs modulated by the METTL3-METTL14-WTAP-RBM15 methyltransferase complex and m⁶A demethylases such as FTO plays important roles in regulating mRNA stability, splicing and translation. Here we demonstrated that FTO-IT1 lncRNA was upregulated and positively correlated with poor survival of patients with wild-type p53-expressing prostate cancer (PCa). m⁶A RIP-seq analysis revealed that FTO-IT1 knockout increased mRNA m⁶A methylation of a subset of p53 transcriptional target genes (e.g. FAS, TP53INP1 and SESN2) and induced PCa cell cycle arrest and apoptosis. We further showed that FTO-IT1 directly binds RBM15 and inhibits RBM15 binding, m⁶A methylation and stability of p53 target mRNAs. Therapeutic depletion of FTO-IT1 restored mRNA m⁶A level and expression of p53 target genes and inhibited PCa growth in mice. Our study identifies FTO-IT1 lncRNA as a bona fide suppressor of the m⁶A methyltransferase complex and p53 tumor suppression signaling and nominates FTO-IT1 as a potential therapeutic target of cancer.

eTOC Blurb

Zhang et al. identified FTO-IT1, a lncRNA upregulated in antiandrogen- and chemotherapy-resistant prostate cancer as a bona fide inhibitor of the m⁶A METTL3-METTL14-WTAP-RBM15 ‘writer’ complex that directly interacts with RBM15, inhibits p53 target gene expression and thereby represents a key driver of prostate cancer progression and a viable therapeutic target.

Graphical Abstract

Introduction

N⁶-methyladenosine (m⁶A) is the most prevalent posttranscriptional modification of RNAs in vertebrate cells Shi et al. ^1,2. This modification is catalyzed by a ‘writer’ complex comprised primarily of METTL3, METTL14, WTAP and RBM15, in which METTL3 is the methyltransferase (MTase) responsible for casting m⁶A, whereas RBM15 and its paralogue RBM15B enable to couple a large number of mRNAs to the ‘writer’ complex for methylation ^3–5. While RNA m⁶A modification is reversible and can be demethylated by demethylases FTO and AlkB homolog 5 (ALKBH5) ^6–9, it remains largely unexplored as to how the potent activity of the MTase ‘writer’ complex itself is regulated, especially by noncoding RNAs (ncRNAs).

The FTO gene locus has been identified as an important genomic/epigenomic hub of many biological functions related to cancer and obesity ^10–12. FTO has been linked to cancer and obesity as an RNA m⁶A demethylase and a cellular sensor of amino acids and activator of the mammalian target of rapamycin complex 1 (mTORC1) ^7,13. Consistent with the reports that genomic variants such as single nucleotide polymorphisms (SNPs) in introns 1 and 2 of FTO gene are strongly associated with the risk of obesity in humans ^12,14–16, a few active enhancers have been identified within the SNP regions and linked to the transactivation of obesity-promoting genes ¹¹.

P53 is one of the most important tumor suppressors that safeguards genomic integrity and suppresses oncogenesis. P53 exerts tumor suppressor functions by primarily acting as a transcription factor that transcriptionally activates downstream target genes involved in cell cycle arrest, senescence and apoptosis such as CDKN1A (p21^WAF1), PUMA, FAS, and TP53INP1 ^17–21. P53 can also inhibit cell growth by transcriptionally activating SESTRIN (SESN) family genes such as SESN2, which are known negative regulators of the mTORC1 complex ^22,23. The importance of p53 in tumor suppression is further supported by the findings that the TP53 gene is inactivated due to genomic alterations such as gene mutation and/or deletion in approximately 50% of all human cancers including advanced prostate cancer (PCa) ^24–27. The tumor suppressor function of p53 is also regulated by protein posttranslational modifications including acetylation and methylation ^28,29. However, major pathways that influence p53 tumor suppression signaling networks beyond TP53 gene/protein itself remain elusive.

PCa is the most-commonly-diagnosed cancer among men in the United States and other Western countries. Androgen deprivation therapy (ADT) is the mainstay treatment for most advanced PCa because of their dependency on androgen/androgen receptor (AR) signaling for growth and survival ^30,31. However, the majority of these tumors relapse after ADT and become castration-resistant prostate cancer (CRPC), which is usually treated with the second-generation antiandrogens such as enzalutamide (ENZ) or the first-line chemotherapy taxane (e.g. docetaxel (DTX)) ^32–35. In addition to blocking the depolymerization of microtubules, we and others have shown that DTX also enables to inhibit AR activities in CRPC cells through various mechanisms ^36–39. Since CRPC patients often progress on the treatment of ENZ or taxane in clinic, it is important to unfold the molecular mechanisms underlying therapy resistance and identify new therapeutic targets for CRPC.

Herein, we demonstrate that FTO intronic transcript 1 (FTO-IT1) is upregulated in therapy-resistant PCa. We further show that FTO-IT1 acts as a suppressor of the m⁶A MTase ‘writer’ complex and p53 tumor suppression signaling by binding to RBM15 and diminishing mRNA m⁶A methylation and stability of a subset of p53 target genes, thereby phenocopying p53 genomic alteration or functional inactivation.

Results

FTO-IT1 is upregulated during PCa progression and negatively correlated with patient survival

To elucidate the molecular mechanisms of antiandrogen therapy resistance in PCa, we performed transcriptomic analysis in control and ENZ-resistant C4–2 cell lines (hereafter termed C4–2C and C4–2R, respectively) we generated previously ^40,41. Since drug resistance mechanisms mediated by coding RNAs and protein posttranslational modifications have been extensively studied in PCa ^42,43, we chose to focus on ncRNAs and RNA posttranscriptional modifications such as m⁶A. Among the 11 known m⁶A ‘writer’ and ‘reader’ genes, we found that ncRNAs are expressed at or near three of these gene loci and FTO-IT1, a long ncRNA (lncRNA) transcribed from intron 8 of FTO gene, was significantly upregulated in C4–2R compared to C4–2C cells (Figures 1A, 1B and S1A and Tables S1-S3). FTO mRNA was also elevated in C4–2R cells compared to C4–2C cells (Figure S1B). RNA fluorescent in situ hybridization (RNA-FISH) analysis showed that FTO-IT1 was localized in both the cytoplasm and nucleus of C4–2 cells and that ENZ treatment induced FTO-IT1 expression but had no obvious effect on FTO-IT1 cellular distribution (Figures 1C, 1D and S1C). RNA copy number analysis showed that FTO-IT1 levels were lower in C4–2C cells but much higher in C4–2R and 22Rv1, another ENZ-resistant cell line ⁴⁴ (Figure S1D).

Figure 1. Increased expression of FTO-IT1 associates with PCa progression and growth.

(A) A boxplot of RNA-seq data showing upregulation of FTO-IT1 ncRNA in C4–2R versus C4–2C cells.

(B) Diagram showing the location of FTO-IT1 in the FTO gene locus (Above) and UCSC screenshot of RNA-seq profile showing the sequencing signal of FTO-IT1 in C4–2R versus C4–2C cells (Bottom).

(C) RNA fluorescent in situ hybridization (FISH) of FTO-IT1 using FAM-labeled FTO-IT1 specific probes and antisense control probes in C4–2C and C4–2R cells. Scale bar, 10 μm.

(D) Quantification of FTO-IT1 FISH signal in C4–2C and C4–2R cells.

(E) RT-qPCR of FTO-IT1 in control and DTX-resistant 22Rv1, LNCaP and C4–2 cells.

(F) Comparison of FTO-IT1 expression levels in different stages of prostate tumors in patients from the TCGA (Firehose Legacy) dataset.

(G) RT-qPCR analysis of FTO-IT1 expression in primary PCa (n = 12) and CRPC (n = 16) patient samples.

(H) Probability of progression free survival (PFS) of FTO-IT1-high and FTO-IT1-low patients of the TCGA (Firehose Legacy) cohort with tumors expressing WT TP53 (Left), and the expression data of FTO-IT1 high and FTO-IT1 low samples (Right). P values were calculated using logrank test.

(I) Probability of overall survival (OS) of FTO-IT1-high and FTO-IT1-low patients of the WCDT cohort with tumors expressing WT TP53 (Left), and the expression data of FTO-IT1 high and FTO-IT1 low samples (Right). P values were calculated using logrank test.

A, D, E, F, G, Data shown as means ± SD. The P values were calculated using an unpaired two-tailed Student’s t-test, *P < 0.05, ***P < 0.01, ***P < 0.001. Experiments in C, D, E, G were repeated twice.

We demonstrated that synthetic androgen mibolerone induced, but ENZ inhibited AR binding in a putative enhancer (H3K4me- and H3K27ac-positive) in the FTO-IT1 locus in C4–2 cells (Figure S1E and S1F). Androgen deprivation or treatment with ENZ or AR proteolysis-targeting chimera (PROTAC) ARV-110 ⁴⁵ increased, but AR overexpression repressed FTO-IT1 expression in C4–2 cells (Figure S1G-S1I). DTX is known to suppress AR function ^36–39 and we demonstrated that the expression level of FTO-IT1, but not FTO was higher in DTX-resistant 22Rv1, C4–2, and LNCaP cells compared to control cells (Figures 1E and S2A). Thus, FTO-IT1 lncRNA is a repression target of the AR and ENZ and DTX promote FTO-IT1 de-repression and overexpression in PCa cells.

Meta-analysis of transcriptomic data in primary PCa samples from The Cancer Genome Atlas (TCGA) cohort showed that FTO-IT1 expression was higher in tumors in advanced-stages (T3b and T4) relative to early-stages (T2a to T3a) (Figure 1F). FTO-IT1 RNA was significantly upregulated in metastatic CRPC tissues compared to primary tumors (Figure 1G). In contrast, FTO mRNA was not elevated in advanced stages or CRPC samples (Figures S2B and S2C). Knockout of FTO-IT1 by CRISPR/Cas9 re-sensitized C4–2R cells to ENZ but had no obvious effect on ENZ sensitivity in C4–2C cells (Figure S2D-S2G). In contrast, FTO knockdown did not alter ENZ sensitivity in C4–2C and C4–2R cells (Figure S2H).

High FTO-IT1 expression significantly associated with poor progression-free survival (PFS) of PCa patients in the TCGA cohort, but no such effect in metastatic PCa patients of the West Coast Dream Team (WCDT) cohort (Figures S2I and S2J). By stratifying tumors with alterations often occurred in PCa such as AR amplification, TMPRSS2-ERG fusion and TP53 gene deletion/mutation, we found that high levels of FTO-IT1 significantly associated with worse PFS in TCGA and worse overall survival of patients in WCDT (Figures 1H and 1I). However, there was no such association with FTO mRNA expression in both cohorts (Figures S2K and S2L). These data indicate that high FTO-IT1 expression associates with poor disease progression only in patients with prostate cancers harboring WT TP53, implying a functional tie between FTO-IT1 overexpression and p53 signaling.

FTO-IT1 downregulates m⁶A levels on a subset of p53 target gene mRNAs

Next, we examined the effect of FTO-IT1 on mRNA m⁶A levels due to its expression from the FTO gene locus. Both dot blot and mass spectrometry analyses showed that FTO-IT1 KO substantially increased global mRNA m⁶A levels in C4–2R and 22Rv1 cell lines (Figures 2A, S2M and S2N). The effect of FTO-IT1 KO on m⁶A_m, a modification occurred at the cap of mRNAs ⁴⁶ was very minimal in both C4–2R and 22Rv1 cells (Figure S2O) and not pursued further.

Figure 2. FTO-IT1 downregulates p53 transcriptional target gene mRNA expression through m⁶A modification.

(A) Mass spectrometry analysis of global m⁶A level on mRNAs from mock and FTO-IT1 KO C4–2R and 22Rv1 cells.

(B) Volcano plot of the hyper- and hypomethylated peaks in FTO-IT1 KO versus mock KO 22Rv1 cells.

(C) Overall m⁶A frequencies along the indicated different regions of mRNAs in mock KO and FTO-IT1 KO 22Rv1 cells.

(D) Volcano plot of the upregulated and downregulated genes in FTO-IT1 KO versus mock KO 22Rv1 cells.

(E) Scatter plot showing the distribution of m⁶A peaks with significant change in both m⁶A level and expression level of corresponding genes in FTO-IT1 KO versus mock KO 22Rv1 cells.

(F) Analysis of the enrichment of pathways from three databases (WikiPathways, KEGG Pathway, and Canonical Pathways) in the genes upregulated and hypermethylated in FTO-IT1 KO compared to mock KO 22Rv1 cells.

(G) Heatmap showing the upregulated expression of p53 transcriptional target genes in FTO-IT1 KO compared to mock KO 22Rv1 cells. The red labelled genes are hypermethylated upon FTO-IT1 KO.

(H, I) IVG screenshot showing input RNA-seq and m⁶A-seq signal profiles of FAS (H) and TP53INP1 (I) gene loci in mock KO and FTO-IT1 KO 22Rv1 cells.

(J) RIP-qPCR of the indicated genes from m⁶A-immunoprecipitated mRNAs in mock KO and FTO-IT1 KO 22Rv1 cells.

(K) Analysis of stability of the indicated gene mRNAs in 22Rv1 cells treated with actinomycin D for different periods of time.

(L) Western blots of whole cell lysates (WCL) from mock KO and FTO-IT1 KO C4–2R and 22Rv1 cells. Short-exposure (S.E.) WB bands of p53 were quantified and normalized to ERK2 (loading control).

J, K, Data shown as means ± SD (n = 3 biological replicates). The P values were calculated using an unpaired two-tailed Student’s t-test; *P < 0.05, **P < 0.01, ***P <0.001. Experiments in J-L were repeated twice.

By performing bulk RNA-seq and m⁶A-methylated mRNA immunoprecipitation sequencing (MeRIP-seq), we identified a total of 10,038 and 14,739 m⁶A peaks overlapped in two replicates in mock and FTO-IT1 KO cells, respectively (Tables S4-S6). The confidence on these peaks was further evident by merging the peaks from two replicates (Tables S7 and S8). Similarly, a large number of individual m⁶A peaks (n = 3,201) were up- but a much smaller number of peaks (n = 573) were down-regulated in FTO-IT1 KO 22Rv1 cells compared to control cells (Figure 2B and Table S9). Relative to control cells, FTO-IT1 KO cells exhibited m⁶A increase primarily in coding sequence (CDS) and parts of 3’UTR (Figure 2C and Tables S6 and S9). Consistent with previous reports ^47,48, the m⁶A frequency reaches its peak near the stop codon, but the peak was higher in FTO-IT1 KO cells compared to control cells (Figure 2C). These findings reveal a role of FTO-IT1 in inhibiting the overall mRNA m⁶A levels in cells.

Bulk RNA-seq revealed 1,229 up- and 1,777 down-regulated genes in FTO-IT1 KO versus control 22Rv1 cells (Figure 2D and Table S10). Pathway analysis indicated the upregulated genes were enriched in a few pathways such as p53 transcriptional gene networks while the downregulated genes were enriched in other pathways such as PLK1 and AURORA B signaling (Figures S3A and S3B). Clustering analysis revealed 1,226 hypermethylated peaks from 649 upregulated genes (hyper-up) but only 253 hypermethylated peaks from 188 down-regulated genes (hyper-down) (Figure 2E and Table S11). This result indicates that m⁶A hypermethylation associates with more upregulated genes in 22Rv1 cells, prompting us to focus on FTO-IT1 regulation of hyper-up genes.

Similar to the hypermethylated targets, p53 transcriptional target genes were also among the top hyper-upregulated genes induced by depletion of FTO-IT1 (Figure 2F). Considering that high level FTO-IT1 expression only significantly associated with poor survival of p53 WT PCa patients (Figures 1H and 1I), we chose to focus on FTO-IT1 regulation of the p53 pathway genes. Gene Set Enrichment Analysis (GSEA) further confirmed the enrichment of p53 pathway as a hallmark change in FTO-IT1 KO cells (Figure S3C). Indeed, a group (n = 36) of p53 pathway genes including those defined by GSEA such as FAS, TP53INP1, SESN2 and MDM2 were significantly upregulated in FTO-IT1 KO cells relative to control cells (Figures 2G–2I, S3D and S3E). These results were further confirmed by MeRIP-qPCR (Figure 2J). FTO-IT1 knockout increased the stability of these mRNAs and their expression at both mRNA and protein levels (Figures 2K, 2L, S3F and S3G). FTO-IT1 KO modestly decreased the steady state level and half-life of p53 protein in both 22Rv1 and C4–2R cells, which is consistent with the modest increase in MDM2, a known E3 ubiquitin ligase targeting p53 protein for degradation (Figures 2L and S3F-S3I). FTO-IT1 KO only had very minimal effect on p53 pathway genes in C4–2C control cells (Figure S3J). On the contrary, FTO-IT1 overexpression decreased p53 target gene mRNA m⁶A level and their stability in 22Rv1 cells (Figures S3K-S3M). Similar results were observed in C4–2R (FTO-IT1 high) compared to C4–2C (FTO-IT1 low) cells (Figures S3N and S3O).

P53 signaling is known to be activated in response to DNA damage. We constructed a sgRNA targeting a gene desert region and demonstrated that DNA cut mediated by CRISPR/Cas9 in this gene desert region did not activate p53 pathway genes (Figures S3P-S3R). We also surveyed FTO-IT1 expression in a large panel of breast cancer cell lines (Table S12). FTO-IT1 expression was higher in MDA-MB-468 cells compared to most of the other cell lines (Figure S3S). FTO-IT1 KO increased overall mRNA m⁶A levels and p53 gene mRNA expression in MDA-MB-468 cells (Figures S3T and S3U), suggesting that FTO-IT1 regulation of p53 signaling may occur in other cancer types. Together, increased expression of FTO-IT1 decreases mRNA m⁶A levels and stability of a subset of p53 target genes related to cell cycle and apoptosis.

FTO-IT1 regulates cell growth and survival via m⁶A-mediated p53 target gene expression

Colony formation assays showed that FTO-IT KO 22Rv1 and C4–2R cells grew much slower than control cells (Figures 3A and 3B). FTO-IT1 knockout induced G1 cell cycle arrest in 22Rv1 cells (Figures 3C, 3D, S4A and S4B). The SESTRIN (SESN) family proteins play important roles in suppression of mTORC1 and mTORC2, activation of which induces upregulation of cell cycle drivers and downregulation of cell cycle inhibitors ²². Consistent with the upregulation of SESN2, a known p53 target gene ^22,49 and the downregulation of phosphorylation of S6K, a downstream effector of mTORC1 in FTO-IT1 deficient cells (Figures 2G and 2L), depletion of SESN2 abolished FTO-IT1 KO-induced cell cycle arrest (Figures 3E–3G), suggesting an important role of SESN2 in mediating FTO-IT1 regulation of the cell cycle. FAS and TP53INP1 are two known p53 target genes that promote apoptosis ^20,50. FTO-IT1 depletion induced apoptotic cell death (Figures 3H, 3I, S4C and S4D), and knockout of FAS or TP53INP1 individually partially blunted FTO-IT1 KO-induced apoptosis in 22Rv1 cells (Figures 3J–3O). The effects of FTO-IT1 on cell cycle and apoptosis remained significant even when cells were treated with the DNA damaging agent camptothecin (CPT) although CPT treatment did increase basal level of G1 cell cycle arrest and apoptosis as expected (Figures S4E-S4I). However, TP53 knockout almost completely abolished FTO-IT1 KO-induced cell cycle arrest and apoptosis (Figures S4J-S4N). Expression level but not mRNA m⁶A level (after normalized with input) of CDKN1A (encodes p21^WAF1), another well-recognized p53 target gene was also upregulated in FTO-IT1 KO cells (Figures 2G and S4O). Knockdown of CDKN1A partially attenuated FTO-IT1 KO-induced cell cycle arrest (Figures S4P-S4R), suggesting a role of FTO-IT1 in regulating p53 target gene expression and cell cycle progression in both mRNA m⁶A modification-dependent and independent manners.

Figure 3. FTO-IT1 promotes cell cycle progression and survival via p53 downstream pathways.

(A, B) Colony formation and quantification of mock KO and FTO-IT1 KO 22Rv1 (A) and C4–2R (B) cells.

(C, D) Cell cycle analysis of mock KO and FTO-IT1 KO 22Rv1 cells using flow cytometry. Representative flow cytometry images (C) and quantification data (D) are presented.

(E) Western blot of WCL from mock KO and FTO-IT1 KO 22Rv1 cells expressing control or SESN2-specific shRNA.

(F, G) Representative flow cytometry images (F) and quantification data (G) of mock KO and FTO-IT1 KO 22Rv1 cells expressing control or SESN2-specific shRNA.

(H, I) Apoptosis analysis of mock KO and FTO-IT1 KO 22Rv1 cells by flow cytometry. Representative flow cytometry images (H) and quantification data (I) are presented.

(J-L) Western blot (J) and apoptosis (K, L) analyses in mock KO and FTO-IT1 KO 22Rv1 cells expressing control or FAS-specific sgRNA.

(M-O) Western blot (M) and apoptosis (N, O) analyses in mock KO and FTO-IT1 KO 22Rv1 cells expressing control or TP53INP1-specific sgRNA.

A, B, D, G, I, L, O, Data shown as means ± SD (n = 3 biological replicates). The P values were calculated using an unpaired two-tailed Student’s t-test; *P < 0.05, **P < 0.01, ***P < 0.001, n.s., not significant. Experiments in A, B were repeated twice.

FTO-IT1 directly binds with RBM15

Knockout of FTO-IT1 modestly decreased FTO mRNA expression in 22Rv1 and C4–2R cells, but the changes were even more subtle at the protein level (Figures S5A-S5C). The discrepancy in the mRNA and protein levels could be due to the very long turnover time of FTO protein in both 22Rv1 and C4–2R cells (Figures S5D and S5E), consistent with a previous report in other cancer types ⁵¹. In contrast, FTO-IT1 KO had little or no effect on FTO expression in LAPC4 and PC-3 PCa cell lines (Figures S5F and S5G). FTO-IT1 deletion had no obvious impact on the expression of other m⁶A modifiers examined in both 22Rv1 and C4–2R cell lines (Figure S5C).

Since FTO-IT1 manipulation did not drastically affect the expression levels of the m⁶A writers and erasers, we sought to determine whether FTO-IT1 binds to these proteins. RNA pulldown and mass spectrometry analysis showed that FTO-IT1 uniquely bound to a total of 280 proteins (Table S13), among which RBM15 was the only m⁶A modifier protein (Figures 4A–4D). Western blot analysis confirmed that FTO-IT1 interacted strongly with RBM15, weakly with RBM15B, METTL3, METTL14 and WTAP but had no obvious association with other members of the ‘writer’ complex (Figure 4E). In vitro protein pulldown assays showed that FTO-IT1 directly bound to RBM15, but not METTL3, METTL14 and WTAP (Figures 4F–4I). CLIP-qPCR assay showed that RBM15 bound to the 3’-stem-loop region of FTO-IT1 (SL3) (Figures 4J and 4K). Reciprocally, RNA pulldown using full-length and SL3-truncated FTO-IT1 RNA confirmed that SL3 is essential for RBM15 binding (Figure 4L). RIP with GST and GST-RBM15 recombinant proteins and in vitro transcribed FTO-IT1 showed that FTO-IT1 directly binds the RRM1 domain of RBM15 (Figures 4M and 4N).

Figure 4. FTO-IT1 interaction with RBM15 and the METTL3/14 MTase complex.

(A) Diagram showing the working principle of protein-RNA pulled down assay using biotin-labeled FTO-IT1 RNA as probe.

(B) Silver staining of proteins pulled down by control (con) RNA and FTO-IT1.

(C) Venn diagram showing the FTO-IT1 interacting proteins and the major known m⁶A modifier proteins (11 m⁶A writers and erasers).

(D) Mass spectrum of the indicated unique peptide of RBM15.

(E) Western blot of indicated proteins in WCL and RNA pulldown samples. S.E. short exposure, L.E. long exposure, * non-specific band.

(F-I) Western blot of samples from RNA pulldown assays using in vitro translated RBM15, METTL3, METTL14 or WTAP.

(J) 2D structure of FTO-IT1 predicted by RNAfold webserver (http://rna.tbi.univie.ac.at//cgi-bin/RNAWebSuite/RNAfold.cgi) and the primer pairs used to detect the indicated stem-loop regions.

(K) CLIP-qPCR analysis of indicated FTO-IT1 regions using specific primers shown in (J) with the RBM15-immunoprecipitated RNAs from UV-crosslinked 22Rv1 cells transfected with FTO-IT1 expression vector.

(L) Western blot of RBM15 protein pulled down by FTO-IT1 full-length or SL3-deletion mutant (FTO-IT1ΔSL3) in RNA pulldown assays.

(M, N) GST-RBM15 recombinant protein constructs (M) and RT-qPCR analysis of samples from in vitro RNA binding assays using in vitro transcribed FTO-IT1 RNA (N, top) and GST RBM15 recombinant proteins detected by Coomassie blue staining (N, bottom). K, N, Data shown as means ± SD (n = 3 biological replicates). Experiments in E-L were repeated twice.

RBM15 selectively binds and increases m⁶A levels of a subset p53 target gene mRNAs

RBM15 is an RNA-binding-motif-containing protein that is reported to bind a large number of RNAs and facilitates RNA m⁶A methylation ⁵. By performing RBM15 CLIP-seq in 22Rv1 cells, we demonstrated that the RBM15 binding sites were highly aligned with m⁶A sites enriched by the GAC motif and peaked at or near the stop codon (Figures S6A-S6C). The 3,498 RBM15-bound mRNAs identified by CLIP-seq significantly overlapped with the well-established p53 pathway genes (Figure 5A and Tables S14-S16), Among the 68 overlapped p53 targets, 49 mRNAs were m⁶A methylated (Table S17) and 13 of them were hypermethylated upon FTO-IT1 KO, including FAS, TP53INP1, SESN2, and MDM2 (Table S17). CLIP-seq and m⁶A IP data displayed a significant consistency between RBM15 binding sites and m⁶A peak sites in the body of FAS, TP53INP1, SESN2, and MDM2 genes (Figure 5B). CLIP-qPCR confirmed that FTO-IT1 KO and overexpression (OE) increased and decreased RBM15 binding of these p53 target gene mRNAs, respectively but had mixed effects on other p53 targets (e.g. JUN, LDHB and NOTCH1) and no effect on the FTO-IT1-unaffected RBM15 binding target MYC and the RBM15-unbound target HPRT1 (Figures 5C, 5D and S6D). RBM15 knockdown not only decreased the basal m⁶A levels on these p53 target gene mRNAs and their expression, but also abolished FTO-IT1 KO-induced increase in m⁶A methylation and expression of these mRNAs (Figures 5E–5J). Given that RBM15 is a key component of the m⁶A ‘writer’ complex that regulates RNA m⁶A methylation ⁵, we sought to determine the role of the m⁶A MTase complex in FTO-IT1 regulation of m⁶A levels on p53 target gene mRNAs. Similar results were obtained by knockdown of METTL3, a catalytic subunit of the MTase complex (Figures S7A-S7C). These data indicate that the effect of FTO-IT1 on p53 target gene mRNA m⁶A modification and expression is primarily mediated through the MTase writer complex.

Figure 5. FTO-IT1 regulates p53 transcriptional target gene expression via binding with RBM15.

(A) Venn diagram showing the overlap between RBM15-interacting mRNAs identified by RBM15 CLIP-seq and typical p53 target gene mRNAs. P value was calculated by hypergeometric probability assuming 25,000 as the total gene number of humans.

(B) IVG screenshot of RBM15 CLIP-seq and m⁶A RIP-seq in the indicated p53 target gene loci.

(C) CLIP-qPCR analysis of indicated mRNAs from RBM15-immunoprecipitated RNA from UV-crosslinked control, FTO-IT1 KO, and FTO-IT1-rescued 22Rv1 cells.

(D) RT-qPCR analysis of FTO-IT1 expression in control, FTO-IT1 KO, and FTO-IT1-rescued 22Rv1 cells.

(E) Western blot of indicated proteins in WCL from 22Rv1 cells transfected with non-specific control (NC) or RBM15-specific siRNAs.

(F) RIP-qPCR of indicated genes in m⁶A-immunoprecipitated mRNAs from 22Rv1 cells transfected with NC or RBM15-specific siRNAs.

(G) RT-qPCR of indicated genes from 22Rv1 cells transfected with NC or RBM15-specific siRNAs.

(H) Western blot of indicated proteins in WCL from control and FTO-IT1 KO 22Rv1 cells transfected with NC or RBM15-specific siRNAs.

(I) RIP-qPCR of indicated genes from m⁶A-immunoprecipitated mRNAs from control and FTO-IT1 KO 22Rv1 cells transfected with NC or RBM15-specific siRNAs.

(J) RT-qPCR of indicated genes from control and FTO-IT1 KO 22Rv1 cells transfected with NC or RBM15-specific siRNAs.

C, D, F, G, I, J, Data are shown as means ± SD (n = 3 biological replicates). The P values were calculated using an unpaired two-tailed Student’s t-test; *P < 0.05, **P < 0.01, ***P < 0.001, n.s., not significant. Experiments in D-J were repeated twice.

RBM15 binds to p53 protein and regulates p53 target mRNA m⁶A level and expression

Previous studies indicate that m⁶A is co-transcriptionally deposited on mRNAs ¹. To investigate how RBM15 regulates p53 target gene mRNA m⁶A level, we performed meta-analysis of p53-interacting proteins identified by mass spectrometry in the BioGRID database and found that RBM15 was the only one of the m⁶A modifiers bound by p53 (Figure 6A). We demonstrated that at endogenous level p53 interacted strongly with RBM15, marginally with METTL3, but not other core components of the m⁶A writer complex including METTL14 and WTAP and RBM15B (Figure 6B). Reciprocal co-IP confirmed this result (Figure 6C). In addition, knockout of FTO-IT1 did not affect RBM15 and p53 protein cellular localization and RBM15 interaction with p53 and other MTase complex components examined (Figures S7D-S7F). GST pulldown assay using GST-RBM15 recombinant proteins and in vitro transcribed and translated p53 proteins showed that the SPOC domain in the C-terminal end of RBM15 directly bound p53 protein (Figures 6D and 6E). Reciprocally, we showed that p53 DNA binding domain (DBD) was required for RBM15 interaction (Figures 6F and 6G). RBM15 ChIP-qPCR analysis revealed the binding of RBM15 in these p53 target gene loci, but the binding was largely diminished by TP53 KO (Figures 6H and 6I). Co-IP assay demonstrated that deletion of the SPOC domain abolished RBM15 binding of p53 (Figures S7G and S7H). We further showed that restored expression of WT RBM15, but not the RBM15ΔSPOC mutant increased m⁶A levels and expression of p53 target mRNAs (Figures 6J–6L), highlighting the importance of RBM15 in regulating p53 target gene mRNA m⁶A levels. The defect of RBM15ΔSPOC in regulating p53 target mRNA m⁶A and expression can be alternatively explained by the inability of this mutant to interact with the MTase complex (Figure S7H), consistent with a previous report ⁵². Collectively, these data indicate that RBM15 directly binds p53 protein and regulates p53 target gene mRNA m⁶A modification and expression.

Figure 6. RBM15 binding of p53 is important for its regulation of p53 target mRNA m⁶A level and expression.

(A) Venn diagram showing the overlap between p53-interacting proteins (data from https://thebiogrid.org/) and the known m⁶A writers and erasers.

(B, C) Western blot analysis of co-IP samples using IgG or indicated antibodies from cell lysate of 22Rv1 cells.

(D, E) GST pulldown using truncated GST-RBM15 recombinant proteins (D) and in vitro translated p53 protein followed by western blot analysis and Coomassie blue staining (E). Asterisks indicate the protein bands at the expected molecular weight.

(F, G) Co-IP assay using HA-tagged truncated p53 proteins (F) and Myc-tagged-RBM15 followed by western blot analysis (G).

(H, I) Western blot of proteins in indicated cells (H) and ChIP-qPCR analysis of RBM15 binding in the indicated gene promoters (except intron 1 of MDM2) in the RBM15-immunoprecipitated chromatin from formaldehyde crosslinked control and TP53 KO 22Rv1 cells (I).

(J-L) Western blot (J), m⁶A RIP-qPCR (K) and RT-qPCR (L) analyses in 22Rv1 cells transfected with indicate siRNAs and constructs.

I, K, L, Data shown as means ± SD (n = 3 biological replicates). The P values were calculated using an unpaired two-tailed Student’s t-test; *P < 0.05, **P < 0.01, ***P < 0.001, n.s., not significant. Experiments in B, C, E, G, H, J were repeated twice.

IGF2BP proteins bind and stabilize m⁶A-modified p53 target gene mRNAs

The fate of m⁶A-methylated RNAs can be either up- or down-regulated due to their recognition by different m⁶A ‘reader’ proteins including YTHDC1–2 and YTDHF1–3) ^53–59 and IGF2BP1–3 ⁵³. IGF2BP proteins can regulate the stability of m⁶A-methylated RNAs ⁵³. Because our data show that FTO-IT1 negatively regulates the m⁶A level and stability of p53 target gene mRNA, we sought to determine whether IGF2BP proteins play a role in FTO-IT1 regulation of p53 gene mRNA expression. We found that IGF2BP1–3 bound p53 target gene mRNAs and that FTO-IT1 KO largely enhanced their binding of p53 targets but not MYC mRNA, a known IGF2BP recognition target ⁵³ (Figures S7I-S7K). Knockout of IGF2BP1–3 decreased p53 target gene expression and almost completely blocked FTO-IT1 KO-induced upregulation of these genes (Figures S7L and S7M). FTO-IT1 KO failed to increase the stability of these gene mRNAs in IGF2BP1–3-depleted cells (Figure S7N). Similarly, FTO-IT1 KO-induced G1 cell cycle arrest and apoptosis were completely reversed by IGF2BP1–3 KO in 22Rv1 cells (Figures S7O-S7R), highlighting a pivotal role of IGF2BP proteins in mediating FTO-IT1 regulation of p53 target gene mRNA stability and inhibition of PCa cell growth.

FTO-IT1 depletion inhibits PCa cell growth in vitro and in mice

We demonstrated that FTO-IT1 KO largely inhibited C4–2R tumor growth in vivo (Figures 7A–7C). Immunohistochemistry (IHC) showed that FTO-IT1 KO decreased proliferation and increased apoptosis in these tumors (Figures S8A and S8B). We designed FTO-IT1-specific antisense oligonucleotides (ASOs) and identified two most potent ones (#3 and #6) that yielded > 80% of reduction in FTO-IT1 expression (Figure 7D). Administration of these two FTO-IT1 ASOs largely increased p53 target gene expression in 22Rv1 and C4–2R cells (Figure 7E). ASO treatment largely increased cleaved PARP1 level, decreased RB phosphorylation and significantly inhibited cell proliferation and colony formation ability (Figures 7E–7G, S8C and S8D). FTO-IT1 ASOs also increased global mRNA m⁶A levels in both C4–2R and 22Rv1 cells (Figure S8E). FTO-IT1 ASO administration largely inhibited tumor growth but not mouse body weight (Figures 7H–7K). The ASO treatment also increased global m⁶A abundance, p53 target gene mRNA m⁶A level and expression, inhibited cell proliferation and induced apoptotic cell death in tumors (Figures 7L, 7M and S8F-S8H). These data indicate that overexpressed FTO-IT1 is a viable therapeutic target of cancer.

Figure 7. FTO-IT1 regulation of PCa growth in vitro and in mice.

(A-C) Measurement of growth (A), size and weight (day 24) (B, C) of tumors derived from mock and FTO-IT1 KO C4–2R cells injected s.c. into SCID male mice.

(D) RT-qPCR analysis of FTO-IT1 in 22Rv1 cells transfected with control or FTO-IT1-specific antisense oligos (ASOs).

(E) Western blot analysis of indicated proteins in C4–2R and 22Rv1 cells transfected with control or FTO-IT1-specific ASOs.

(F, G) Colony formation (F) and MTS (G) assays using 22Rv1 cells transfected as in (E).

(H-K) Measurement of growth (H), weight and size (day 24) (I, J) of tumors derived from 22Rv1 cells injected s.c. into SCID male mice and treated with control ASO or FTO-IT1-specific ASOs and mouse body weight (K).

(L) RT-qPCR analysis of FTO-IT1 and indicated p53 target gene mRNAs in 22Rv1 xenografts harvested from the mice at 24 day after treated with control or FTO-IT1-specific ASOs as shown in (H).

(M) Dot blot detection of m⁶A modification on mRNAs in 22Rv1 xenograft samples.

A, C, H, I, K, Data shown as means ± SD (n = 8 biological replicates). D, F, L, Data shown as means ± SD (n = 3 biological replicates). G, Data shown as means ± SD (n = 5 biological replicates). The P values were calculated using an unpaired two-tailed Student’s t-test; *P < 0.05, **P < 0.01, ***P < 0.001, n.s., not significant. Experiments in E, M were repeated twice.

Discussion

The FTO gene locus appears to be a functionally dynamic, but not fully appreciated genomic region. FTO affects a large spectrum of biological processes by acting as a RNA m⁶A demethylase ^1,7. FTO also controls mTORC1 activities and amino acid sensing independently of the demethylase function ¹³. Additionally, enhancer activity in intron 1 of the FTO gene has been linked to obesity-associated SNPs (genomic variants) to the role of IRX3 expression in obesity ¹¹. In the present study we observed that FTO-IT1, a lncRNA transcribed from the last intron of the FTO gene is overexpressed during PCa progression and overexpressed FTO-IT1 induces downregulation of global and p53 target gene mRNA m⁶A levels in PCa cells by binding to and intervening the activities of the m⁶A methyltransferase complex. Thus, our study identifies a previously uncharacterized noncoding RNA as another key functional element in the FTO gene locus that restrains mRNA m⁶A modification and expression of p53 target gene such as SESN2, a key negative regulator of mTORC1 ^22,23 and p53 tumor suppression function (Figure 8).

Figure 8. Proposed working model.

A proposed working model for FTO-IT1 in regulating mRNA m⁶A methylation. Top, in cells where no/low FTO-IT1 is expressed, RBM15 in the MTase ‘writer’ complex mediates the m⁶A methylation of a subset of p53 target gene mRNAs and their stability and promotes the p53 tumor suppression functions. Bottom, overexpressed FTO-IT1 physically interacts with RBM15 and inhibits RBM15-mediated p53 target gene mRNA m⁶A methylation and stability, thereby attenuating p53 downstream tumor suppression signaling and tumor progression. M3, METTL3; M14, METTL14; W, WTAP.

It is well known that the tumor suppressor function of p53 is often inactivated by genetic alterations (gene mutation and/or deletion) or aberrant protein degradation ^25,60. We demonstrate that with no effect on TP53 gene mRNA and only marginal effect on p53 protein expression FTO-IT1 inhibits p53 tumor suppression signaling by largely decreasing the mRNA m⁶A levels of a few key p53 downstream target genes. Notably, depletion of each of these genes only partially reverses the activation of p53 tumor suppression function induced by FTO-IT1 KO, further supporting the notion that individual p53 target genes do not account for all of p53 function ^61,62. We further show that RBM15 directly binds to p53 protein and induces p53 target gene mRNA m⁶A methylation, thereby revealing a new connection of the m⁶A MTase ‘writer’ complex to p53 signaling, a mechanism in parallel to the direct binding of p53 by METTL3 reported recently ⁶³ (Figure 8, Top). However, this effect of RBM15 was abolished by FTO-IT1 binding of RBM15 (Figure 8, Bottom). Thus, our study identifies FTO-IT1 as a bona fide suppressor of the MTase ‘writer’ complex, a pivotal mechanism in suppression of mRNA methylation complementary to the exon junction complex discovered recently ⁶⁴. Our findings also reveal a previously unrecognized epitranscriptomic mechanism that circumvents the tumor suppressor activity of WT p53 by disrupting its downstream target gene signaling beyond TP53 gene itself, thereby phenocopying genomic inactivation of TP53 gene.

A previous study shows that RBM15 mainly binds to the U-rich motifs in the 3’-UTR of target genes in 293T cells (Patil et al., 2016). We found that RBM15 mainly bound to the CDS of genes with binding peaks at stop codon (similar to the m⁶A distribution) in PCa cells and “GACG” was the most enriched motif although a small percentage of U-rich motifs was also observed. This “inconsistency” is possibly due to different cell contexts given that different cell models were used in these studies. This phenomenon further supports the notion that the role of m⁶A modifiers is cell context-dependent. While the current study was ongoing, METTL3 was identified most recently by an independent group as a p53-interacting partner under genotoxic stress conditions ^63,65. Based on these studies, we envisage a model whereby p53 associates with the MTase ‘writer’ complex via directly interacting with RBM15 and/or METTL3 in the presence or absence of extracellular stress stimuli (Figure 8, Top).

Consistent with the recent report ^63,65, our data reveal that there is a m⁶A methyltransferase complex-augmented tumor suppressor action of p53 (Figure 8, Top). However, this role of the MTase complex appears to be conditional and it can be intrinsically impaired by FTO-IT1 lncRNA overexpression (Figure 8, Bottom). Importantly, due to the reversible nature of m⁶A modification ^1,66, we provide evidence that FTO-IT1 overexpression-induced inhibition of p53 target gene mRNA m⁶A modification, inactivation of p53 target gene networks and augmented tumor growth can be abolished by therapeutic targeting of FTO-IT1. We show that depletion of FTO-IT1 by ASOs not only restores the expression of p53 target genes, but also largely inhibits PCa cell growth in vitro and in mice. Thus, our findings stress that overexpressed FTO-IT1 could be a viable biomarker and therapeutic target of tumors, especially those p53-WT cases where mRNA m⁶A methylation-dependent p53 tumor suppression networks are inactivated by overexpressed FTO-IT1.

Limitations of the Study.

Our transcriptomic analysis shows that a group of p53 target genes are upregulated upon FTO-IT1 knockout; however, only a subset of them are affected by m⁶A modification. It is possible that FTO-IT1 may also regulate p53 tumor suppression signaling via the mechanism(s) independent of m⁶A methylation, which are unclear at present. While we show that FTO-IT1 directly binds to RBM15 and inhibits its effect on mRNA m⁶A modification, we noticed that not all RBM15-bound m⁶A-modified p53 target gene mRNAs are affected by FTO-IT1. The exact underlying mechanism remains to be determined.

STAR Methods

Resource availability

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Haojie Huang (Huang.Haojie@mayo.edu).

Materials Availability

All unique/stable reagents generated in this study are available from the Lead Contact with a completed Materials Transfer Agreement.

Data and Code Availability

All sequencing data generated in this study have been deposited in NCBI Gene Expression Omnibus (GEO): GSE189966, GSE189465, GSE212043, GSE229871. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD041053 and 10.6019/PXD041053. Raw images of gels have been deposited at Mendeley Data. All data are publicly available as of the date of publication. Accession numbers and DOI are listed in the key resources table.

Key Resources Table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti-m6A	Synaptic Systems	202 111
Rabbit monoclonal anti-FTO	abcam	ab126605
Rabbit polyclonal anti-METTL3	Proteintech	15073-1-AP
Rabbit polyclonal anti-METTL14	Sigma Aldrich	HPA038002
Rabbit polyclonal anti-METTL16	Proteintech	19924-1-AP
Rabbit polyclonal anti-ALKBH5	Proteintech	16837-1-AP
Rabbit polyclonal anti-WTAP	Proteintech	10200-1-AP
Rabbit polyclonal anti-RBM15	Proteintech	10587-1-AP
Rabbit polyclonal anti-RBM15B	Proteintech	22249-1-AP
Rabbit polyclonal anti-VIRMA	Proteintech	25712-1-AP
Rabbit polyclonal anti-HAKAI	Bethyl lab/ Fortis	A302-969A
Rabbit polyclonal anti-ZC3H13	Bethyl lab/ Fortis	A300-748A-T
Mouse monoclonal anti-FAS	Santa Cruz	SC-8009
Rabbit polyclonal anti-TP53INP1	Santa Cruz	SC-68919
Rabbit polyclonal anti-MDM2	abcam	ab260074
Rabbit polyclonal anti-SESN2	Proteintech	10795-1-AP
Mouse monoclonal anti-RB	BD Biosciences	554136
Rabbit polyclonal anti-P-RB T821	Invitrogen	44-582G
Mouse monoclonal anti-ERK2	Santa Cruz	SC-1647
Mouse monoclonal anti-p53	Santa Cruz	SC-126
Rabbit monoclonal anti-C-PARP1	Cell Signaling	5625S
Rabbit polyclonal anti-C-CASP3	Cell Signaling	9661S
Rabbit polyclonal anti-Ki67	abcam	ab15580
Rabbit monoclonal anti-IGF2BP1	Cell Signaling	8482S
Rabbit polyclonal anti-IGF2BP2	Proteintech	11601-1-AP
Rabbit polyclonal anti-IGF2BP3	Proteintech	14642-1-AP
Mouse monoclonal anti-BrdU	BD	555627
Goat Anti-Mouse IgG H&L (FITC)	ZENBIO	511101
Mouse monoclonal anti-Actin	Cell Signaling	3700S
Mouse monoclonal anti-MYC	Santa Cruz	SC-40
Chemicals, peptides, and recombinant proteins
Docetaxel (DTX)	Active Biochemicals	A-1917
Enzalutamide (MDV3100)	Selleckchem	S1250
Lipofectamine 2000	Life Technologies	11668-019
Polybrene	Santa Cruz Biotechnology	sc-134220
SYBR Green Mix	Bio-Rad	170-8885
Propidium Iodide (PI)	Sigma	P4170-100mg
BrdU	Yeasen	40204ES60
Annexin V-PE apoptosis detection kit I	BD	559763
Deposited data
RNA-seq in C4-2C and C4-2R	This paper	GEO: GSE189966
m6A-seq in 22Rv1 cells	This paper	GEO: GSE189465
RBM15 CLIP-seq in 22Rv1 cells	This paper	GEO: GSE212043
H3K4me1 and H3K27ac ChIP-seq in VCaP cells	This paper	GEO: GSE229871
AR ChIP-seq in Mib treated C4-2 cells	Zhao et al., Cell rep 2016	GEO: GSE55032
AR ChIP-seq in ENZ treated C4-2 cells	He et al., Nat Commun 2021	GEO: GSE136130
Mass Spectrometry proteomics data	This paper	PXD041053 and 10.6019/PXD041053
Western blot raw data have been deposited in Mendeley Data	This paper	DOI: 10.17632/vs3msdvsvw.1
IFC image raw data have been deposited in Mendeley Data	This paper	DOI: 10.17632/mcnbx2pf7z.1
Experimental models: Cell lines
HEK293T	ATCC	N/A
22Rv1	ATCC	N/A
C4-2	Uro Corporation	N/A
PC-3	ATCC	N/A
LNCaP	ATCC	N/A
LAPC4	ATCC	N/A
Breast cancer cell lines	Dr. John R. Hawse’s lab	See supplementary Table 12
Experimental models: Organisms/strains
SCID mice	Jackson Lab	N/A
Software and algorithms
m6A peaks analysis	Meng et al., Methods 2014	ExomePeak R package
CLIP-seq analysis	Drewe-Boss et al., Genome Biol 2018	omniCLIP
ChIP-seq analysis	Wang et al., Ann Oncol 2018	bowtie2 (version 2.2.9) and MACS2 (version 2.1.1)

This paper does not report original code.

Any additional information required to reanalyze the data reported in this paper is available from the Lead Contact upon request.

Experimental Model and Study Participant Details

22Rv1, LNCaP, PC-3, LAPC4, MDA-MB-468 and 293T cells were obtained from the American Type Culture Collection (ATCC). C4–2 cells were purchased from Uro Corporation. 22Rv1, C4–2, LNCaP and PC-3 cells were maintained in RPMI1640 supplemented with 10% FBS. 293T and LAPC4 cells were maintained in DMEM supplemented with 10% FBS. MDA-MB-468 cells were maintained in Leibovitz’s L15 supplemented with 10% FBS. C4–2R cells were generated by treating C4–2 cells with 10 μM enzalutamide for one month.

Method details

Antibodies and reagents

The antibodies used in this study are listed in key resources table.

Plasmids

GST-tagged RBM15 and Myc-tagged RBM15 were generated by cloning the corresponding cDNA into the pGEX-4T-1 and pCMV vector, respectively. HA-tagged p53 plasmids were previously generated in our lab. FTO-IT1 expression plasmid was generated by cloning the cDNA into the pCDNA-3.1 vector. The cDNA fragments were amplified by Phusion polymerase (NEB) using Phusion High-Fidelity PCR Master Mix. The primers used for plasmid construction and knock out test are listed in Table S18.

Transfection, and lentivirus infection

For transient transfection, cells were transfected with Lipofectamine 2000 (Thermo Fisher) according to the manufacturer’s instructions. For lentivirus production, the pLenti-CRISPR-V2 plasmid containing corresponding sgRNA sequence or pLKO-based gene shRNA knockdown plasmids or pTsin plasmid containing corresponding gene CDS were mixed with pMD2.G and psPAX2 and transfected into 293T cells. The virus-containing supernatant was harvested 48 h after transfection to infect PCa cells in the presence of 10 μg/ml polybrene. The successfully infected cells were selected with 1 μg/ml puromycin. The shRNA plasmids were purchased from Sigma-Aldrich. The shRNA/siRNA sequences targeting METTL3, SESN2 and RBM15, and the sgRNA sequences targeting FTO-IT1, IGF2BP1, IGF2BP2, IGF2BP3, FAS, TP53INP1 are listed in Table S19.

Generation of knockout cell lines

We generated coding gene knockout cell lines using the CRISPR/Cas9 approach as previously described ⁶⁷. To knock out a noncoding RNA or a genomic region in a gene desert region, in each case we designed a pair of sgRNAs targeting each end of the designated genomic region. The cells were co-infected with lentivirus for CRISPR/Cas9 and sgRNAs and selected with puromycin. The genomic DNA was isolated from stable clones for PCR amplification. Knockout of FTO-IT1 was also confirmed by RT-qPCR at the RNA level.

Antisense oligonucleotides (ASOs) design and screening

ASOs were designed based on the complementary sequence of FTO-IT1 with phosphorothioate (PS) backbone and MOE modification on the flanking six nucleotides (IDT). ASOs were transfected into 22Rv1 cells followed by RT-qPCR analysis and the highly efficient ASOs were used for further studies. ASO sequences are listed in Table S19.

RNA isolation from human prostate cancer specimens

Formalin-fixed paraffin-embedded (FFPE) hormone-naïve primary PCa and CRPC tissues were randomly selected from the Mayo Tissue Registry. RNAs were isolated using a RecoverAll Total Nucleic Acid Isolation Kit (Invitrogen). Informed consent was obtained from all human participants, and the studies were approved by the Institute Review Board (IRB) of the Mayo Clinic.

FTO-IT1 RNA copy number measurement

RNA copy number measurement was performed as previously described ⁴¹. Briefly, FTO-IT1 was cloned into the pcDNA3.1 backbone vector. The cDNA copy number and dilution calculation were performed using the method described in the web site shown below. (https://www.lifetechnologies.com/us/en/home/brands/thermo-scientific/molecular-biology/molecularbiology-learning-center/molecular-biology-resource-library/thermo-scientific-web-tools/dna-copy-numbercalculator.html.) 1×10⁵ cells were used for RNA extraction and the total RNA was diluted in 100 μl H₂O. One microliter of RNA was used for reverse transcription and 1% of the cDNA was used for qPCR. The final Ct value correspond to the copy number in 10 cells. A standard curve was used to correspond the Ct value with actual copy number. Copy number was calculated by the equation derived from standard curve.

RNA extraction from cultured cells and reverse transcription-quantitative PCR (RT-qPCR)

The total RNA was isolated using TRIzol reagent (Thermo Fisher Scientific) and reverse-transcribed to cDNA using superscript RT kit (Promega GoScript) according to manufacturer’s instruction. Quantitative PCR was performed using SYBR Green Master mix Kit (Bio-Rad) in Bio-Rad CFX manager 3.1. The quantification of indicated genes was normalized to that of endogenous control GAPDH. The primers for RT-qPCR are listed in Table S18.

m⁶A dot blot

Dot blot of m⁶A was carried out as previously reported ⁷. Briefly, mRNA was purified from total RNA using Dynabeads^™ mRNA Purification Kit (Thermo Fisher Scientific). The isolated mRNA was first denatured in 95 °C for 3 min and chilling on ice directly. Two-fold serial dilution of the mRNA were spotted on Biodyne B nylon membrane (PALL) and crosslinked by UV Stratalinker. The membrane was blocked by 5% non-fat milk and incubated with anti-m⁶A antibody (Synaptic Systems 1:2,000) overnight at 4 °C. Horseradish peroxidase (HRP)-conjugated secondary antibody was used to incubate with the membrane and ECL was used to visualize the signal. A copy of spotted membrane was stained with 0.02% methylene blue in 0.3 M sodium acetate (pH 5.2) to ensure that an equal amount of mRNA was loaded.

m⁶A RNA immunoprecipitation

Purified mRNA was partially digested with 1 unit of RNase T1 for 2 min and incubated with 2 μg of m⁶A antibody and protein A/G beads in IPP buffer (10 mM Tris pH7.5, 150 mM NaCl, 0.1% NP-40) supplemented with RNase inhibitor in 4 °C overnight. The beads were washed 6 times and the RNA was extracted using TRIzol reagent. RT-qPCR was performed to detect the enrichment of m⁶A modified RNA.

mRNA m⁶A and m⁶Am methylation level measured by LC-MS/MS

LC-QqQ-MS/MS measurements were performed as reported previously ⁶. In brief, total RNA was purified with TRIzol^® reagents (Thermo Fisher Scientific, #15596018) from fresh cells. mRNA was isolated by using Dynabeads^® mRNA DIRECT kit (Thermo Fisher Scientific, #61006) twice. After that, rRNA was further removed using RiboMinus Eukaryote kit (Thermo Fisher Scientific, A1083708). The purified mRNA was further digested into nucleotides with nuclease P1 (Sigma, N8630) in 20 ml of buffer containing 25 mM NaCl and 2.5 mM ZnCl2 for 1 h at 42°C, and then 1 unit of FastAP Thermosensitive Alkaline Phosphatase (1 U/μl, Thermo Fisher Scientific, EF0651) in FastAP buffer for another 4 h at 37°C. Samples were then filtered (0.22 mm, Millipore) and injected into a C18 reverse phase column coupled online to Agilent 6460 LC–MS/MS spectrometer in positive electrospray ionization mode. The nucleosides were quantified by using retention time and the nucleoside to base ion mass transitions (268-to-136 for A; 296-to-150 for m⁶Am, and 282-to-150 for m⁶A). Quantification was performed by comparing with the standard curve obtained from pure nucleoside standards running with the same batch of samples.

m⁶A-seq library preparation and data analysis

mRNA was purified from total RNA by using Dynabeads^® mRNA DIRECT kit (Thermo Fisher Scientific, #61006). 1 μg mRNA in 100 ul RNase free water was fragmented to ~200 nt using Bioruptor^® Pico Sonication Instrument with 30 cycles of 30s on/30s off mode. 5 μl of the fragmented mRNA was saved as input. The remaining fragmented mRNA was subjected to m⁶A IP by using the EpiMark^®N6-Methyladenosine Enrichment Kit (NEB, E1610S) following the manufactory protocol. RNA libraries were prepared for both input RNA and m⁶A-enriched mRNA after IP using TruSeq^® Stranded mRNA Library Prep (Illumina, 20020594) following the manufacturer’s protocol. Sequencing was performed at the University of Chicago Genomics Facility on an Illumina HiSeq 2000 machine in single-read mode with 50 bp per reading at around 25 M to 30 M sequencing depth. After obtaining the raw data, single-end reads were harvested and trimmed by Trim_Galore to remove adaptor sequences and low-quality nucleotides. High-quality reads were then aligned to UCSC hg19 reference genome by HISAT2 using default parameters, and only uniquely mapped reads were retained for all downstream analyses. FeatureCounts software was used to count reads mapped to RefSeq genes, and differentially expressed genes analysis was conducted by DESeq 2 Software. m⁶A peaks on RefSeq transcripts and differentially methylated m⁶A peaks were analyzed by ExomePeak R package. To visualize sequencing signals at specific genomic regions, we used Deeptools to normalize all libraries and imported into IGV.

In vitro transcribed biotin-labeled RNA pulldown

Full-length and fragments of FTO-IT1 were amplified by PCR using FTO-IT1 specific primers with the forward primer containing a T7 promoter. Biotin-labeled RNAs were in vitro transcribed using corresponding PCR products as template and Biotin RNA Labeling Mix (Roche) and T7 polymerase (New England Biolabs). Control biotin-labeled RNA was in vitro transcribed using empty pcDNA3.1-SFB vector as template (~ 600 nt) which contains a T7 promoter and the sequences of V5 epitope, polyhistidine, SFB tags (S tag, Flag tag and biotin binding protein (streptavidin) binding peptide) and some other sequences before the polyadenylation signal. The transcribed RNA products were treated with DNase I to eliminate the template DNA. 22Rv1 cells were lysed in modified binding buffer (50 mM Tris pH7.5, 150 mM NaCl, 1% NP-40, 0.1% SDS) supplemented with protease inhibitor and RNase inhibitor. For in vitro translated protein, plasmid containing RBM15 gene sequence and T7 promoter was incubated with TNT Quick Master Mix (PR-L1170) for 90 min. Cell lysates or in vitro translated protein were incubated with biotin-labeled RNAs and streptavidin beads at 4 °C for 12 h. The beads were washed with wash buffer (50 mM Tris pH7.4, 150 mM NaCl, 0.05% NP-40, 1mM MgCl2) for 4 times. The samples were resolved in SDS loading buffer and denatured in 95 °C. Western blot and mass spectrometry were used to analyze the interaction proteins. The primers for FTO-IT1 fragments PCR are listed in Table S18.

In vitro transcription and RNA pulldown by GST proteins

FTO-IT1 RNA was transcribed in vitro using T7 RNA polymerase (New England Biolabs) and NTP Mix (Thermo). The transcribed RNA products were treated with DNase I to eliminate the templated DNA. Plasmids of pGEX-4T-1 containing truncated GST-RBM15 proteins were transformed in E. coli (BL21) and induced by 0.1 mM IPTG at 16 °C for 12 h. The GST-RBM15 proteins were purified by glutathione Sepharose beads (GE Healthcare) as previously described (Wang et al., 2013). Purified GST-RBM15 proteins with glutathione Sepharose beads were incubated with in vitro transcribed FTO-IT1 RNA in RNA structure buffer (50 mM Tris pH7.4, 150 mM NaCl, 1 mM MgCl2) at 4 °C for 4 h. After 6 times of wash, the RNAs were purified by Trizol and detected using RT-qPCR.

GST pulldown of ectopically expressed protein and in vitro translated protein

Plasmids encoding p53 truncations were transfected into 293T cells. After 36 h, the cells were lysed with IP buffer (50 mM Tris pH7.5, 150 mM NaCl, 0.5% NP-40) and the cell lysates were incubated with purified GST-RBM15 protein as described previously. For in vitro translated protein, plasmid containing p53 gene sequence and T7 promoter was incubated with TNT Quick Master Mix (PR-L1170) for 90 min. The in vitro translated protein was then incubated with GST-RBM15 protein. After 4 h of incubation and 3 times of wash, the bound proteins were analyzed by Western blot.

RNA fluorescent in situ hybridization (FISH)

RNA FISH was performed as previously described ⁶⁸. Briefly, C4–2 cells were seeded on coverslips in 6 well plates. After the cells adhered on the coverslips, 4% paraformaldehyde in PBS was used to fix the cells for 15 min at RT. The fixing solution was removed, and samples were washed twice with PBS. Then 0.5% Triton X-100 in PBS at room temperature for 10 min was applied to permeabilize the cells. After washing with PBS for 3 times, the samples were rinsed with 2 × SSC (0.3 M NaCl, 0.03 M Na3 Citrate, pH 7.0) and hybridization was performed by incubating the samples with 10 nM of FAM-labeled control (antisense) and FTO-IT1 specific probe mix in hybridization solution (50% Formamide, 2 × SSC, 10% dextran sulfate, 1 mg/ml yeast t-RNA) in a humid box at 37 °C for 16 h. The samples were washed with 2 × SSC for 3 times and 1 × SSC for 3 times and mounted with VECTASHIELD mounting medium. Images were acquired using Zeiss LSM 780 confocal microscope. The intensity of fluorescence signals was quantified and normalized to DAPI using ImageJ. The sequences of probes used in FISH are listed in Table S19.

Cross-linking immunoprecipitation (CLIP)

6 × 150-mm dishes of 22Rv1 cells with 80% confluence were crosslinked by UV254 and harvested. The cells were then lysed in lysis buffer (150 mM NaCl, 0.5% NP-40, 50 mM Tris-HCl (pH 7.5), 2 mM EDTA) with complete, Mini, EDTA-free Protease Inhibitor Cocktail and SUPERNase in at 4°C for 1 hour. Subsequently, the lysis mixture was centrifuged at 17,000 g at 4°C for 30 min and the supernatant was carefully collected. The samples were then treated with RNase T1 (1 u/uL) at RT for 15 min and centrifuged to collect the supernatant, 10% of which was saved as input. Then RBM15 antibody conjugated protein G beads prepared by incubating antibody and beads at 4°C for 6 hours were added into the samples and incubate overnight. For CLIP-qPCR, the beads were washed for 4 times and subjected for Proteinase K de-crosslinking and RNA extraction. RT-qPCR was used to detect the interaction regions. For CLIP-seq, after washing beads three times, 10 U/μl RNase T1 was added and incubated at RT for 8 min. The beads were then resuspended in 50 μl of SDS-PAGE loading buffer and heated at 95°C for 5 min. The RNA was finally extracted by cutting and recovering the band of RBM15-RNA complex from the SDS-PAGE gel. The libraries for both input and IP samples are prepared using NEBNext^® Multiplex Small RNA Library Prep Set for Illumina^® (NEB, E7300S). CLIP samples were pooled and sequenced with NovaSeq6000. Raw reads from CLIP samples were first trimmed according to recommended settings ⁶⁹. Gene structure annotations were downloaded from UCSC hg19 RefSeq/Repeatmasker. For analysis of CLIP-seq data, we used omniCLIP ⁷⁰ for peak calling.

RNA stability assay

Cells were treated with 5 μg/ml Actinomycin D and collected at indicated time points. The total RNA was extracted, and mRNA level of each gene was analyzed by RT-qPCR. Linear regression was used to determine the trend line equation based on the changes of mRNA level at different time points. The half-life of each mRNA was calculated through the equation:

$$\text{dC}/\text{dt}=-{\text{K}}_{\text{decay}}\text{C}$$

Thus, the mRNA degradation rate ${\text{K}}_{\text{decay}}$ was estimated by:

$$\ln \left(\text{C}/{\text{C}}_0\right)=-{\text{K}}_{\text{decay}}\text{t}$$

To calculate the mRNA half-life ($t_{1/2}$), when 50% of the mRNA is decayed (that is, $C/C_0=1/2$), the equation was:

$$\ln (1/2)=-{\text{K}}_{\text{decay}}t1/2$$

From where:

$$\text{t}1/2=\ln 2/{\text{K}}_{\text{decay}}$$

Chromatin immunoprecipitation (ChIP) and ChIP-qPCR

For H3K27ac and H3K4me1 ChIP-seq, VCaP cells were fixed with formaldehyde and subjected to sonication by Bioruptor (Diagenode) as described previously ⁷¹. The supernatant was obtained and mixed with protein A/G beads and antibodies for H3K27ac and H3K4me1. After incubation overnight, beads were washed, and the complex containing DNA was eluted at 65°C. The elution was further treated with RNAase and proteinase K. Enriched DNA was extracted for high throughput sequencing. Sequencing libraries were prepared as previously described ⁷². The high-throughput sequencing was performed by Illumina HiSeq 4000 platform by the Mayo Clinic Genome Core Facilities. The raw reads were mapped to the human reference genome (GRCh37/hg38) using bowtie2 (version 2.2.9). MACS2 (version 2.1.1) was used for peak calling with a p value threshold of 1 × 10⁻⁵ as described ⁷³. BigWig files were generated for visualization using the UCSC Genome Browser. The assignment of peaks to potential target genes was performed by the Genomic Regions Enrichment of Annotations Tool (GREAT).

For ChIP-qPCR assays, 22Rv1 cells cultured in 2 × 150 mm dishes were cross-linked for 15 min at room temperature by adding 11% formaldehyde/PBS solution in cell culture medium. Cross-linked cells were scraped into tubes and sonicated with Bioruptor^® Pico Sonication Instrument with 10 cycles of 30s on/30s off mode. After centrifugation, the supernatant was incubated with antibody bound protein A/G agarose beads overnight at 4 °C. The beads were washed 4 times. The precipitated protein-DNA complexes were eluted and cross-linking was reversed at 65 °C for 12 h. After Proteinase K digestion, the chromatin was isolated and subjected to qPCR analysis.

Cell cycle analysis

1 × 10⁶ of 22Rv1 cells were suspended by trypsinization and washed with cold PBS. The suspended cells were fixed with 50% cold ethanol and kept in −20 °C overnight. After washing with cold PBS, the fixed cells were resuspended with 0.5 ml PBS and added with 0.2 mg/ml RNase A and incubated for 1 h at 37 °C. The cell suspension was added with 10 μg/ml PI (Sigma P4170) and analyzed on FACS by reading on cytometer at 488 nm. For 2-dimensional FACS, cells were treated with 30 μM BrdU for 30 min before harvested, and fixed with 70% cold ethanol and kept in −20 °C overnight. The cells were then treated with 2 N HCl for 30 min and incubated with BrdU antibody for 30 min followed by FITC labeled secondary antibody for another 30 min. The samples were treated with RNaseA and added with 10 μg/ml PI (Sigma P4170) and analyzed on FACS. The data was analyzed by FlowJo_V10.

Apoptosis analysis

1 × 10⁶ of 22Rv1 cells were suspended by trypsinization and washed twice with cold PBS. The cell pellet was resuspended in 1 ml 1X Binding Buffer (BD Pharmingen^™ BDB559763). 100 μl of cell suspension was transferred to a 5 ml culture tube and added with 5 μl PE Annexin V and 5 μl 7-AAD (BD Pharmingen^™ BDB559763). The cell suspension was gently vortexed and incubated for 15 min at RT in dark. 400 μl of 1X Binding Buffer was added to each tube and analyzed by flow cytometry. Unstained cells, single PE Annexin V and single 7-AAD stained cells were used for control. The data was analyzed by FlowJo_V10.

Protein co-immunoprecipitation (co-IP)

Immunoprecipitations were performed as described previously ⁷⁴. Briefly, cells were lysed with lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.1% Nonidet P-40, and freshly added protease inhibitor cocktails) and centrifuged to obtain supernatant. Protein A/G beads and indicated antibody were used to incubate with the supernatant at 4 °C overnight. Beads were washed 3 times with lysis buffer, re-suspended in SDS loading buffer prior to western blot analysis.

Western blot

Whole cell lysates or IP samples were subjected to SDS-PAGE. The proteins were transferred onto nitrocellulose membranes (GE Healthcare sciences). The transferred membranes were blocked using TBST with 5% w/v nonfat milk and incubated with indicated primary antibodies at 4 °C overnight. The antibodies used in this study were listed in the Key Resources Table. In the second day, the membranes were washed 3 times with TBST and followed by incubation with secondary antibodies at room temperature. After washing in TBST for three times, the membranes were visualized using Enhanced Chemiluminescence (ECL) system (Thermo Fisher Scientific).

Cell proliferation assay

Cells were seeded in 96-well plates in a concentration of 2,000 cells per well. The CellTiter 96 Aqueous One solution Cell Proliferation Assay (MTS) (Promega) was used to measure cell viability at indicated time points as described previously ⁷¹. The MTS was diluted at a ratio of 1:10 in PBS and added into the wells and incubated for 1 h at 37 °C in a cell incubator. Microplate reader was used to measure absorbance of 490 nm in each well.

Colony formation assay

The procedure was carried out as previously described ⁷⁴. Briefly, cells were seeded in 6-well plates in a concentration of 5,000 cells per well. Approximately 12 days later, the colonies were fixed with 4% paraformaldehyde for 15 min and stained with crystal violet (0.5% w/v) for 1 h. The colonies were gently washed with running tap water and counted for quantification.

Xenografts generation and drug treatment

The animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) at the Mayo Clinic. Six-week-old SCID male mice were housed in standard condition with a 12-h light /12-h dark cycle and randomly divided into different groups as indicated. 5 × 10⁶ of mock KO or FTO-IT1 KO C4–2R cells were mixed with Matrigel (50 μl of PBS plus 50 μl of Matrigel (BD Biosciences)) and injected subcutaneously into mice. For 22Rv1 xenografts, 5 × 10⁶ of cells were mixed with Matrigel (50 μl of PBS plus 50 μl of Matrigel (BD Biosciences)) and injected subcutaneously into mice. When xenografts reached a size of approximately 100 mm³, indicated vehicle (PBS with 0.3 mg/ml PEI) and drugs (FTO-IT1-specific ASOs 3 mg/kg in PBS with 0.3 mg/ml PEI) were administered by tail vein injection 4 days a week. Tumor growth was measured in a blinded fashion by a caliper. The volume of the tumors was calculated using the formula (L × W2)/2, where L stands for the length of the tumor and W stands for the width. Tumor volumes were compared, and P values were determined by unpaired two-tailed Student’s t-test. After 3-week injection, the tumors were dissected and photographed.

TCGA gene expression and survival analysis

IlluminaHiSeq (n=550) TCGA Hub level-3 data was downloaded from TCGA data coordination center. This dataset shows the gene-level transcription estimates, as in log2(x+1) transformed RSEM normalized count. Genes are mapped onto the human genome coordinates using UCSC Xena HUGO probeMap (see ID/Gene mapping link below for details). The TCGA reference method description is from the University of North Carolina Center for Genomic Characterization: DCC description. In order to make it easier to see differential expression between samples, we set the default view to center each gene or exon independently minus each gene with mean zero or exons. For survival analysis, the cohort was split into high-expression and low-expression group using a function of the X-tile software⁷⁵ as a method for selection of optimal cutpoint. The P values were calculated by logrank test.

WCDT dataset survival analysis

Gene expression and clinical information for the PCa West Coast Dream Team (WCDT) dataset were downloaded from previous publication ⁷⁶. For survival analysis, samples were median dichotomized into two groups according to FTO-IT1 expression. Logrank test from the R package Survival (v3.2.11) was used to calculate P value and the result was visualized using the R package BoutrosLab.plotting.general (v5.9.8) ⁷⁷.

Quantification and statistical analysis

All data are shown as means ± SD unless otherwise specified. The data was processed in Microsoft Excel version 2013. Difference between two groups were analyzed using unpaired two-tailed Student’s t-tests unless otherwise specified. A P value < 0.05 was considered statistically significant.

Supplementary Material

1

Table S9-S19 are included in Data S1

Table S9. Differential m⁶A peaks in FTO-IT1 KO versus mock KO 22Rv1 cells. Related to Figure 2B.

Table S10. Differentially expressed genes in FTO-IT1 KO versus mock KO 22Rv1 cells. Related to Figure 2D.

Table S11. Differentially expressed genes with differential m⁶A peaks in FTO-IT1 KO versus mock KO 22Rv1 cells. Related to Figure 2E.

Table S12. Source of breast cancer cell lines used in the study. Related to the cell line information in Key Resources Table.

Table S13. Proteins uniquely interacted with FTO-IT1 RNA. Related to Figure 4C.

Table S14. RBM15 binding sites by CLIP-seq. Related to Figure 5A.

Table S15. p53 signaling gene (WP_P53_transcriptional gene network, GSEA HALLMARK, KEGG P53 signaling pathway) and overlap with RBM15-ChIP binding genes (highlighted). Related to Figure 5A.

Table S16. RBM15 binding p53 target gene. Related to Figure 5A.

Table S17. Methylated RBM15-bound p53 pathway genes (highlighted genes are hypermethylated upon FTO-IT1 KO). Related to Figure 5A.

Table S18. Information of primer sequences. Related to Plasmids and reverse transcription-quantitative PCR (RT-qPCR) in STAR Methods.

Table S19. Information of shRNA/sgRNA/ASO sequences. Related to Transfection, and lentivirus infection, Antisense oligonucleotides (ASOs) design and screening, RNA fluorescent in situ hybridization (FISH) in STAR Methods.

3

Table S1. Differentially expressed ncRNAs in enzalutamide resistant versus control (mock-treated) C4–2 cells. Related to Figure 1A.

4

Table S2. ncRNAs upregulated (FC > 2 folds, p < 0.05) in enzalutamide resistant versus control C4–2 cells. Related to Figure 1A.

5

Table S3. ncRNAs transcribed from the gene loci of the major known m⁶A modifiers. Related to Figure 1A.

6

Table S4. m⁶A peaks form overlapping the two replicates in MOCK KO 22Rv1 cells. Related to Figure 2B.

7

Table S5. m⁶A peaks from overlapping the two replicates in FTO-IT1 KO 22Rv1 cells. Related to Figure 2B.

8

Table S6. m⁶A peaks identified in mock and FTO-IT1 KO 22Rv1 cells from overlapping the two replicates. Related to Figure 2B.

9

Table S7. m⁶A peaks form merging the two replicates in MOCK KO 22Rv1 cells. Related to Figure 2B.

10

Table S8. m⁶A peaks from merging the two replicates in FTO-IT1 KO 22Rv1 cells. Related to Figure 2B.

Highlights.

FTO-IT1 upregulation associates with poor prognosis of advanced prostate cancer

FTO-IT1 binds RBM15 and acts as an inhibitor of the m⁶A ‘writer’ complex

FTO-IT1 inhibits p53 target gene mRNA m⁶A level, phenocopying p53 inactivation

FTO-IT1 represents a viable therapeutic target of cancers