Epitranscriptic regulation of HRAS by N⁶-methyladenosine drives tumor progression

Significance

Here, we revealed the widespread N⁶-methyladenosine (m⁶A) modification of HRAS that mediates tumor progression. Mechanistically, three m⁶A modification sites of HRAS 3′ UTR, which is regulated by FTO and YTHDF1, increases its translational elongation. These m⁶A sites of HRAS targeted by dm⁶ACRISPR system can decrease cancer proliferation and metastasis. Clinically, up-regulated H-Ras expression is widely negatively or positively associated with FTO or YTHDF1 in various cancers. Collectively, our study shed light on a potentially therapeutic strategy for the undruggable Ras.

Abstract

Overexpression of Ras, in addition to the oncogenic mutations, occurs in various human cancers. However, the mechanisms for epitranscriptic regulation of RAS in tumorigenesis remain unclear. Here, we report that the widespread N⁶-methyladenosine (m⁶A) modification of HRAS, but not KRAS and NRAS, is higher in cancer tissues compared with the adjacent tissues, which results in the increased expression of H-Ras protein, thus promoting cancer cell proliferation and metastasis. Mechanistically, three m⁶A modification sites of HRAS 3′ UTR, which is regulated by FTO and bound by YTHDF1, but not YTHDF2 nor YTHDF3, promote its protein expression by the enhanced translational elongation. In addition, targeting HRAS m⁶A modification decreases cancer proliferation and metastasis. Clinically, up-regulated H-Ras expression correlates with down-regulated FTO and up-regulated YTHDF1 expression in various cancers. Collectively, our study reveals a linking between specific m⁶A modification sites of HRAS and tumor progression, which provides a new strategy to target oncogenic Ras signaling.

Rat sarcoma (Ras) proteins are a class of small GTPases family that play key roles in either physiological or pathological processes, and are encoded by Harvey RAS (HRAS), Kirsten RAS (KRAS) and Neuroblastoma RAS (NRAS) (1). Aberrant Ras function drives cancer cell proliferation, survival, and metastasis, which result from the well-established genomic mutations and posttranslational modifications of Ras proteins (2). Among which, KRAS is the most frequently mutated, while HRAS is the less in all cancer-associated RAS mutations (3, 4). However, whether RAS genes are regulated in posttranscriptional level remains unexplored.

Epitranscriptomics refers to the posttranscriptional modification of RNA bases, which is mediated by specific RNA modification enzymes, and plays critical roles in regulation of gene expression (5). N ⁶-methyladenosine (m⁶A) is the most prevalent posttranscriptional modification of eukaryote RNAs, which plays essential function in physiological processes, and its dysregulation is associated with various types of diseases, including cancers (6, 7). m⁶A methylation is reversibly regulated by the RNA methyltransferases, and demethylases, and exerts its functions mainly by recruiting reader proteins (8–10). However, whether RAS genes are modified in an m⁶A manner, and the function of this modification remain unknown.

Overexpression (OE) of Ras proteins occurs in various human cancers and is associated with poor prognosis in patients (11, 12), but the underlying mechanism is not fully elucidated. Here, we revealed that m⁶A modification occurs in HRAS, but not KRAS and NRAS. In mechanism, m⁶A modification of HRAS is mainly mediated by Fat mass and obesity-associated protein (FTO) and YTH N6-methyladenosine RNA binding protein 1 (YTHDF1) to regulate its protein expression, thus enhanced cell proliferation and metastasis. Collectively, our study shed light on the specific m⁶A modification of HRAS to regulate tumor progression, which will provide the potentially targeting Ras strategy for cancer therapeutics.

Results

The Identification of HRAS m⁶A Modification.

To explore the high m⁶A modification genes in cancer, we constructed a comprehensively differential m⁶A modification landscape across four distinct cancer types that contained 30 patient tumors and adjacent tissues from public meRIP-Seq datasets (SI Appendix, Table S1) (13–16). We identified thousands of differential m⁶A modification genes in ovarian cancer, endometrial cancer, hepatocellular carcinoma, and colon cancer tissues compared with adjacent tissues (Dataset S1), and there were only 446 overlapped genes among these cancer types (SI Appendix, Fig. S1A), supporting that m⁶A modification is markedly altered in different cellular contexts and disease states (17). Considering that genes may have tissue- and cancer-specific m⁶A peaks (18), we further identified genes that had the same trends for m⁶A peaks in these cancers. Analysis results showed that there were 36 genes with consistently high and 2 genes with consistently low m⁶A peaks in these cancer types (SI Appendix, Fig. S1B). To explore the function of these genes, we performed Kyoto Encyclopedia of Genes and Genomes pathway analysis and found that these genes were significantly enriched in 13 signaling pathways, including Mitogen-activated protein kinase (MAPK) and PI3K-AKT signaling pathways (SI Appendix, Fig. S1C). Notably, we noticed an important oncogenic gene HRAS that has significantly higher m⁶A modification in the four cancers, and was associated with nine of these signaling pathways (SI Appendix, Fig. S1 B and C), which was chosen for further study. Specially, compared with adjacent tissues, the 3′ UTR of HRAS was identified to have higher m⁶A level in tumor tissues, while the 5′ UTR and CDS had not obvious m⁶A modification (Fig. 1A). However, the KRAS and NRAS were not observed in obvious m⁶A modification (Fig. 1A). To validate these observations, we performed m⁶A-RIP-qPCR analysis and found that the HRAS was significantly modified by m⁶A in bladder cancer 5637 cell line, hepatocellular carcinoma HepG2 cell line (HRAS wild type) and bladder cancer T24 cell line (HRAS^G12V mutation) (Fig. 1B). Moreover, the m⁶A levels of HRAS were significantly higher in renal cell cancer tissues than those in adjacent tissues (Fig. 1C). In contrast, we did not detect m⁶A modification in both KRAS and NRAS genes (Fig. 1 B and C). Altogether, these data suggest that m⁶A modification exists in HRAS but not KRAS and NRAS.

Fig. 1.

The identification of HRAS m⁶A modification. (A) Distribution of m⁶A peaks across the HRAS, KRAS, and NRAS transcripts based on meRIP-Seq from the tumors and adjacent tissues of ovarian cancer (GSE119168), endometrial cancer (GSE93911), hepatocellular carcinoma (GSE120860), and colon cancer (GSE179042). (B and C) m⁶A RIP qPCR showing the enrichment of m⁶A modification in the HRAS, KRAS, and NRAS 3′ UTR in 5637, HepG2, and T24 cell lines (B), and in the tumors and adjacent tissues of patients with renal cancer (n = 3) (C). (D and E) m⁶A RIP qPCR showing the enrichment of m⁶A modification in HRAS 3' UTR in FTO or ALKBH5 KD 5637 and HepG2 cell lines (D), and the wild-type or mutants of FTO and ALKBH5 OE 5637 and HepG2 cell lines (E). HPRT was used as a negative control (n = 3). FTO.Mut, FTO mutant; ALKBH5.Mut, ALKBH5 mutant. (F) RIP qPCR showing the enrichment of FTO and ALKBH5 in the HRAS 3′ UTR (n = 3). The results are presented as mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001. ns, not significant.

To determine the major regulator of HRAS m⁶A modification, we performed the expression correlation analysis between HRAS and five major m⁶A regulators, including Methyltransferase-like protein 3/14/16 (METTL3, METTL14 and METTL16), FTO, and Alk B homolog 5 (ALKBH5) in TCGA Pan-Cancer and found that m⁶A demethylase FTO negatively associated with HRAS in almost all cancer types (SI Appendix, Fig. S1D), suggesting that FTO may be a majorly negative regulator of HRAS. m⁶A-RIP-qPCR analysis showed that both FTO and another demethylase ALKBH5 knockdown (KD) increased the m⁶A enrichment of HRAS but not of the m⁶A-negative HPRT transcript (19) (Fig. 1D and SI Appendix, Fig. S1F). In contrast, the OE of wild-type FTO and ALKBH5, but not catalytically inactive FTO and ALKBH5 mutants reduced the m⁶A enrichment of HRAS (Fig. 1E and SI Appendix, Fig. S1G). Consistent with our results, two published MeRIP-seq experiments (20, 21) also revealed that FTO negatively regulate the m⁶A level of HRAS (SI Appendix, Fig. S1E), confirming that demethylase decreases the m⁶A level of HRAS. Moreover, RIP-qPCR showed that FTO and ALKBH5 bind to HRAS (Fig. 1F). Together, these results confirm that the HRAS is a direct substrate of FTO and ALKBH5-catalyzed demethylation. Then, we further examined whether the methyltransferases METTL3/14 are responsible for catalyzing m⁶A modification of HRAS, and found the decreased m⁶A enrichment of HRAS in both METTL3 and METTL14 KD 5637 cells (SI Appendix, Fig. S1 H and I). In contrast, the OE of wild-type methyltransferases, but not catalytically inactive mutants (22) enhanced the m⁶A levels of HRAS (SI Appendix, Fig. S1 J and K), suggesting the m⁶A modification of the HRAS mRNA catalyzed by METTL3/14.

m⁶A Enhanced HRAS Translation.

Next, we examined the potential regulation function of m⁶A on HRAS expression. Luciferase reporter assays showed that there was no significant difference of promoter activities of HRAS between control and FTO or ALKBH5 KD cells (SI Appendix, Fig. S2A), which exclude the regulation of m⁶A on HRAS transcription. Cellular fractionation assays revealed that there was no significant difference of subcellular localization of HRAS mRNA between control and FTO or ALKBH5 KD cells (SI Appendix, Fig. S2 B and C). Additionally, we treated cells with actinomycin D to block transcription and found that FTO and ALKBH5 KD had no effect on the stability of the mRNA of HRAS (SI Appendix, Fig. S2 D and E). Furthermore, qRT-PCR showed that both FTO and ALKBH5 KD had no significant effect on the mRNA expression of HRAS, KRAS, and NRAS (Fig. 2 A and B and SI Appendix, Fig. S2F). These results indicated that m⁶A had no effect on the RNA level of HRAS. However, both FTO and ALKBH5 KD significantly increased H-Ras protein levels but not of K-Ras and N-Ras protein levels (Fig. 2 C and D and SI Appendix, Figs. S2G and S3 A and B). In contrast, the OE of wild-type FTO and ALKBH5, but not of catalytically inactive FTO and ALKBH5 mutants markedly decreased H-Ras protein levels (Fig. 2 E and F and SI Appendix, Figs. S2H and S3 C and D). Consistently, protein levels of exogenous WT and mutant HRAS fused with 3′ UTR (G12V and Q61L) were increased by FTO and ALKBH5 KD (Fig. 2G). However, METTL3/14 increased H-Ras protein level but not its transcriptional level in an m⁶A catalytic activity-dependent manner, as revealed by qRT-PCR and immunoblot (IB) assays from the METTL3 and METTL14 KD or OE cells (SI Appendix, Fig. S4 A–C). These results suggest that m⁶A enhanced the level of H-Ras protein.

Fig. 2.

m⁶A enhanced HRAS translation. (A and B) qRT-PCR analysis of HRAS, KRAS, and NRAS in FTO or ALKBH5 KD 5637 (A) and T24 (B) cell lines. (C–F) IB analysis of H-Ras, K-Ras, and N-Ras in FTO or ALKBH5 KD 5637 (C) and T24 (D) cell lines, and the wild type or mutants of FTO and ALKBH5 OE 5637 (E) and T24 (F) cell lines. (G) IB analysis of H-Ras wild type and mutants (G12V and Q61L) fused with HRAS 3′ UTR in FTO or ALKBH5 KD 5637 cells. (H) qRT-PCR showing the mRNA abundance of F-Luc and R-Luc (Left), dual-luciferase assay showing the relative luciferase activities of HRAS 3′ UTR (Middle), the translation outcome determined as a relative ration of luciferase intensities divided by relative mRNA level (Right) in FTO or ALKBH5 KD cells. (I) Polysome profiling of FTO or ALKBH5 KD 5637 cells (Top), and relative HRAS mRNA distribution in each ribosome fractions was analyzed by qRT-PCR (Bottom). *P < 0.05 and **P < 0.01. ns, not significant.

To investigate the potential mechanisms involved in m⁶A-regulated protein expression of HRAS, we treated 5637 cells with a protein synthesis inhibitor cycloheximide, and the half-lives of H-Ras protein was not significantly altered after FTO or ALKBH5 KD compared with control cells (SI Appendix, Fig. S4D), excluding the possibility of m⁶A-regulated the protein stability of H-Ras. Next, we constructed a pmirGLO-HRAS 3′ UTR luciferase reporter to explore whether m⁶A regulated the translation of HRAS. The dual-luciferase assay showed that luciferase activity was significantly enhanced after FTO and ALKBH5 KD, while the mRNA abundance of Firefly Luciferase (F-Luc) and Renilla Luciferase (R-Luc) were not affected (Fig. 2H), suggesting that the demethylases regulate translation efficiency of HRAS. Next, we performed ribosome purification by two methods (23, 24). qRT-PCR showed that FTO and ALKBH5 KD increased HRAS mRNA level either in translation-active polysomes (> 80S) by ribosome profiling (23) (Fig. 2I) or in ribosomal translationally active fraction by RiboLace (24) (SI Appendix, Fig. S4 E and F). Altogether, these data indicate that m⁶A modification is linked to translation of HRAS.

YTHDF1 Promotes HRAS Translation in an m⁶A-Dependent Manner.

It is reported that m⁶A modification may regulate protein translation through readers including YTHDF and IGF2BP families (25, 26). RIP-qPCR assay showed that YTHDF1, but not YTHDF2/3, or IGF2BP1/2/3 significantly bound to HRAS in 5637 and HepG2 cells (Fig. 3A). Consistent with our results, a published RIP-seq (27) also revealed that YTHDF1 is remarkably enriched in the 3′ UTR of HRAS (Fig. 3B), confirming that the 3′ UTR of HRAS is specifically bound by YTHDF1. To determine whether YTHDF1 binds to the HRAS transcript in an m⁶A-dependent manner, we first performed RIP-qPCR assay for YTHDF1 and found that the binding abilities of YTHDF1 to HRAS transcript were significantly increased after FTO and ALKBH5 KD (Fig. 3C). Next, a flag-tagged mutant YTHDF1 construct (YTHDF1-Mut) with two key amino acids mutations (K395A and Y397A) to abolish its m⁶A-binding pockets (28) was transfected into 5637 cells. Flag RIP-qPCR assay revealed that HRAS was significantly enhanced interaction with YTHDF1-WT, but the interaction between YTHDF1-Mut and HRAS was obviously impaired (Fig. 3D), suggesting that the binding of YTHDF1 to the HRAS transcript is dependent on m⁶A. We next examined the effects of YTHDF1 on HRAS expression and found that YTHDF1 KD led to a significant reduce in H-Ras protein level, but not its mRNA level (SI Appendix, Fig. S5 A and B). Conversely, YTHDF1-WT OE, but not YTHDF1-Mut increased H-Ras protein expression in 5637 and HepG2 cells (SI Appendix, Fig. S5C), which was supported by more cancer cell lines, including 769-P, AsPC-1, HCC827, Hela, MCF7, PC-3, SW480, and HGC27 cells (SI Appendix, Fig. S6 A and B). To further evaluate whether demethylases inhibit H-Ras expression via YTHDF1, we performed rescue experiments and showed that YTHDF1 KD prevented the protein level of H-Ras increased by FTO KD (Fig. 3E), suggesting that YTHDF1 is a m⁶A reader of HRAS and increases its protein expression. Furthermore, Ribosome profiling and RiboLace assays showed that YTHDF1 KD decreased HRAS mRNA level either in translation-active polysomes or in ribosomal translationally active fraction (Fig. 3 F and G). Consistently, the dual-luciferase assay showed that translation efficiency of HRAS was significantly decreased in YTHDF1 KD cells (Fig. 3H). YTHDF1 promotes translation of m⁶A-modified genes, which is associated with translation initiation and elongation factors, such as eIF3, eIF4, eEF1, and eEF2 (29, 30), thus we then evaluated the roles of these factors in the m⁶A-modified HRAS. RIP-qPCR assay showed that eEF1 and eIF4 significantly bound to the HRAS transcripts (Fig. 3I). Similar to the YTHDF1, the binding abilities of eEF1, but not of eIF4, to the HRAS transcripts were increased in FTO and ALKBH5 KD (Fig. 3 J and K), corroborating that m⁶A induced translation elongation of HRAS mRNA through YTHDF1 and eEF1. Collectively, these data demonstrate that YTHDF1 promotes HRAS translation efficiency in an m⁶A-dependent manner.

Fig. 3.

YTHDF1 promotes HRAS translation in an m⁶A-dependent manner. (A) RIP qRT-PCR detecting the enrichment of IGF2BP1/2/3 and YTHDF1/2/3 in the HRAS 3′ UTR in 5637 and HepG2 cell lines. (B) Distribution of peaks across the HRAS transcript based on YTHDF1 RIP-seq data (GSE136433, n = 2). (C) RIP qPCR showing the enrichment of YTHDF1 in HRAS 3′ UTR in FTO or ALKBH5 KD 5637 and HepG2 cell lines. (D) IB detecting flag proteins in Flag-tagged YTHDF1 RIP analysis (Left), and RIP qPCR analysis showing the enrichment of YTHDF1 in HRAS 3′ UTR in the wild type or mutants of YTHDF1 OE 293T cells (Right). (E) IB detecting the expression of H-Ras in FTO KD 5637 cells with YTHDF1 KD. (F) Polysome profiling of YTHDF1 KD 5637 cells (Left), and relative HRAS mRNA distribution in each ribosome fractions was analyzed by qRT-PCR (Right). (G) qRT-PCR analysis the relative enrichment of HRAS in YTHDF1 KD 5637 cells in ribosomal translationally active fraction by RiboLace. (H) qRT-PCR showing the mRNA abundance of F-Luc and R-Luc (Left), dual-luciferase assay showing the luciferase activities of HRAS 3′ UTR (Middle), the translation outcome determined as a relative ration of luciferase intensities divided by relative mRNA level (Right) in YTHDF1 KD cells. (I) RIP qRT-PCR detecting the enrichment of eEF1/2 and eIF3/4 in HRAS 3′ UTR in 5637 cells. (J and K) RIP qRT-PCR detecting the enrichment of eEF1 (J) and eIF4 (K) in HRAS 3′ UTR in FTO or ALKBH5 KD 5637 cells. *P < 0.05, **P < 0.01, and ***P < 0.001. ns, not significant.

The Identification of HRAS m⁶A Sites.

To further explore the exact effects of m⁶A on HRAS translation, we determined the m⁶A sites on the HRAS 3′ UTR. According to the consensus sequence (RRACH) of m⁶A motif (31, 32), we identified nine potential m⁶A sites in the 3′ UTR of HRAS (SI Appendix, Fig. S7A). We determined the methylation sites in vivo by performing SELECT method (33) with the accuracy of single-base resolution, and found that m⁶A levels of three sites (A923, A930, and A955) in the 3′ UTR of HRAS were significantly increased upon FTO and ALKBH5 KD among 5637, HepG2, and T24 cell lines, while other six sites were unchanged (Fig. 4A and SI Appendix, Fig. S7B). Moreover, RIP-qPCR assay showed that FTO, ALKBH5, and YTHDF1 were more effectively bound to the WT HRAS 3′ UTR compared with the mutated m⁶A construct with simultaneously three m⁶A point mutations (Fig. 4 B–D). Multiple published single-nucleotide resolution crosslinking-immunoprecipitation and high-throughput sequencing datasets (34–36) further confirmed that the three m⁶A sites in the 3′ UTR of HRAS are modified by m⁶A (SI Appendix, Fig. S8A). To further verify the binding of YTHDF1 to the three m⁶A sites, we performed RNA pulldown using the biotinylated and m⁶A-modified RNA probes. The results showed that YTHDF1 had a specific enrichment in m⁶A probe pulldown fraction, but not in non-m⁶A and mutant probes (Fig. 4E). In contrast, YTHDF2 and YTHDF3 had weak binding capacity, but IGF2BP family proteins did not bind to this probe (Fig. 4E), which may be explained by that YTHDF1/2/3 bind m⁶A sites using their high homologous YTH domains (37). Furthermore, YTHDF1 did interact with individual m⁶A-modified probe for each site (SI Appendix, Fig. S8B), suggesting that the m⁶A sites of A923, A930, and A955 are responsible for the binding of YTHDF1 to the HRAS 3′ UTR. To next define the effects of these three m⁶A sites on HRAS expression, we designed reporters containing individual and simultaneous point mutations at these m⁶A sites (Fig. 4F). Luciferase reporter assays showed that individual point mutation can significantly reduce the luciferase activities, and simultaneous point mutations suppressed more remarkably the luciferase activities (Fig. 4F). Additionally, we constructed HA-tagged HRAS expression vector (HRAS-WT) and its mutants with individual m⁶A or simultaneous mutations (HRAS-Mut). IB analysis revealed that YTHDF1-WT but not YTHDF1-Mut could enhance the expression of HRAS-WT, while YTHDF1-WT had less effects on the expression of HRAS-Mut, and HRAS-Mut expression was decreased compared with HRAS-WT in 5637 cells (Fig. 4G and SI Appendix, Fig. S8C). Together, these data suggest that HRAS 3' UTR contains three m⁶A sites, which are required for YTHDF1 binding and for the HRAS translation.

Fig. 4.

The identification of HRAS m⁶A sites. (A) The threshold cycle (Ct) of qPCR showing SELECT results of A923, A930 and A955 sites of HRAS in FTO or ALKBH5 KD 5637, HepG2 and T24 cell lines. (B–D) RIP qRT-PCR detecting the enrichment of FTO (B), ALKBH5 (C) and YTHDF1 (D) in HRAS 3′ UTR in wild type or mutants (A923, A930 and A955 simultaneous point mutations) of HRAS 3′ UTR OE cells. (E) IB detecting the enrichment of YTHDF1/2/3 and IGF2BP1/2/3 in RNA pulldown results using non-m⁶A, m⁶A or mutant HRAS 3′ UTR probe in 5637 and T24 cell lines. (F) Schematic representation of mutation sites of pmirGLO-HRAS 3’ UTR luciferase reporter (Top) and dual-luciferase assay showing the luciferase activities of these constructs. (G) IB analysis of HA and flag in HA-tagged wild-type or mutant HRAS in 5637 cells cotransfected with wild-type or mutant Flag-tagged YTHDF1. *P < 0.05, **P < 0.01, and ***P < 0.001.

Targeting m⁶A of HRAS by dm⁶ACRISPR-Inhibited Tumor Growth.

To functionally determine the effects of m⁶A-modified HRAS on tumor growth, we first specifically demethylated the m⁶A of HRAS using dm⁶ACRISPR system, fusing the catalytically dead Type VI-B Cas13 enzyme with the m⁶A demethylase ALKBH5 (dCas13b-ALKBH5) (38) and FTO (dCas13b-FTO) (39), respectively (SI Appendix, Fig. S9A). Two gRNAs targeting distinct sequence around the three m⁶A sites were designed to target HRAS (SI Appendix, Fig. S9A). To test the efficiency of gRNAs, we analyzed the mRNA levels of HRAS in cells transfected with gRNAs and wild-type Cas13b. qRT-PCR analysis showed that the two gRNAs cotransfected with wild-type Cas13b, but not transfection of gRNAs alone or gRNAs combined with dCas13b respectively can significantly decreased the mRNA level of HRAS (SI Appendix, Fig. S9B), suggesting that the two gRNAs can efficiently recognize HRAS. Next, we verified the demethylation effects of dm⁶ACRISPR system on HRAS. m⁶A-RIP-qPCR analysis showed that both dCas13b-ALKBH5 and dCas13b-FTO combined with gRNAs indeed decreased the m⁶A level of HRAS (Fig. 5A), and RIP-qPCR analysis showed that the binding abilities of YTHDF1 to HRAS was significantly decreased after the m⁶A of HRAS were demethylated by dm⁶ACRISPR systems (SI Appendix, Fig. S9C). The dm⁶ACRISPR-induced demethylation of A923, A930, and A955 sites was confirmed by the SELECT method (Fig. 5B), further supporting that the m⁶A sites at A923, A930, and A955 of the HRAS 3′ UTR exist and are reversibly modified. We then tested the effects of HRAS 3′ UTR demethylation on its expression and found that dm⁶ACRISPR targeting HRAS led to significant downregulation of H-Ras protein in three cell lines (Fig. 5C and SI Appendix, Fig. S9 D and E), while HRAS mRNA level had no significant change (Fig. 5D and SI Appendix, Fig. S9 F and G), indicating that dm⁶ACRISPR decreases H-Ras protein expression through demethylation of m⁶A sites at 3′ UTR of HRAS.

Fig. 5.

Targeting m⁶A of HRAS by dm⁶ACRISPR-inhibited tumor growth. (A) m⁶A RIP qPCR showing the enrichment of m⁶A modification in HRAS 3′ UTR in 5637 cells transfected with dm⁶ACRISPR systems of FTO and ALKBH5. (B) Ct of qPCR showing SELECT results of A923, A930 and A955 sites of HRAS in 5637 cells transfected with dm⁶ACRISPR systems. (C and D) IB (C) and qRT-PCR (D) analysis of H-Ras in 5637 cells transfected with dm⁶ACRISPR systems of FTO and ALKBH5. (E and F) Cell proliferation assay (Left) assessing 5637 cells ectopically expressed dm⁶ACRISPR systems of FTO (Top) and ALKBH5 (down). Representative micrographs (Middle) and quantification (Right) of the abovementioned cells in the Matrigel-coated or noncoated Transwell assays (n = 3). (G and H) Effects of 5637 cells ectopically expressing dm⁶ACRISPR systems of FTO (G) and ALKBH5 (H) on tumor growth (Left), tumor size (Middle) and mass (Right) in nude mice (n = 6). (I and J) Representative images of brightfield (Left) and H&E-stained (Middle) samples from the tumor foci of the lungs obtained from nude mice, the lateral tail vein of which were injected by 5637 cells ectopically expressing dm⁶ACRISPR systems of FTO (I) and ALKBH5 (J), and the number of metastatic lesions in mice (Right) as determined by H&E staining (n = 8). (K) IB analysis of H-Ras, p-ERK/ERK, p-MEK/MEK in 5637 cells ectopically expressing dm⁶ACRISPR systems of FTO (Left) and ALKBH5 (Right). (L–O) Representative images (Left) and quantification (Right) of H-Ras (L and N) and p-ERK expressions (M and O) in tumors from G and H at the experiment end point by immunohistochemistry (IHC) analysis (n = 6). (P) The significantly enriched GO terms in the biological process category for down-regulated genes from 5637 cells with ectopically expressed dCas13b-FTO combined with gRNA control or gRNA1/2. (Scale bars, 20 µm.) *P < 0.05, **P < 0.01, and ***P < 0.001. ns, no significant.

Based on the above dm⁶ACRISPR systems, we next determine the role of m⁶A-modified HRAS on tumor growth in vitro and in vivo. Cell proliferation and transwell assays showed that both dCas13b-ALKBH5 and dCas13b-FTO combined with gRNAs significantly decreased proliferation, migration, and invasion of the 5637, HepG2, and T24 cells, compared with gRNA control (Fig. 5 E and F and SI Appendix, Fig. S10 A–L). Similar results were also observed in HRAS KD cells (SI Appendix, Fig. S11 A–H). Furthermore, the subcutaneous tumorigenesis mouse model showed that both dCas13b-ALKBH5 and dCas13b-FTO combined with gRNA inhibited in vivo xenograft tumor growth and mass of 5637 cells (Fig. 5 G and H). Moreover, hematoxylin–eosin (H&E) staining showed that number of metastatic nodules in the lungs were significantly decreased in the dm⁶ACRISPR targeting HRAS group (Fig. 5 I and J), suggesting that targeting m⁶A of HRAS 3′ UTR by dm⁶ACRISPR can suppress the tumor progression in cancers.

MAPK pathway is a major downstream of Ras, which plays an essential role in cancer (4). As expected, dm⁶ACRISPR targeting HRAS reduced expression of p-ERK and p-MEK (Fig. 5K). Consistently, dm⁶ACRISPR targeting HRAS also significantly reduced the H-Ras and p-ERK expressions of tumor tissues in mouse xenograft models (Fig. 5 L–O). Furthermore, RNA-sequencing analysis revealed that dCas13b-FTO combined with gRNAs targeting HRAS had no significant difference of FTO and HRAS expression, but led to upregulation of 447 genes and downregulation of 569 genes (SI Appendix, Fig. S12 A and B and Dataset S2), while the down-regulated genes were preferentially enriched in ERK1 and ERK2 cascade and MAPK signaling pathway (Fig. 5P and SI Appendix, Fig. S12C). Collectively, these data indicated that HRAS translation by m⁶A modification in its 3′ UTR promotes tumor growth via the MAPK signaling pathway.

FTO/YTHDF1/H-Ras Axis Regulates Cancer Growth.

To further examine whether demethylases and YTHDF1 regulate cancer growth through H-Ras, we first evaluated the effects of FTO and ALKBH5 on bladder cancer cell lines. The results showed that FTO KD increased, but ALKBH5 KD reduced the cell proliferation and invasion abilities of 5637 and T24 cell lines (SI Appendix, Fig. S13 A–F), suggesting that FTO is a tumor suppressor gene, whereas ALKBH5 as an oncogene promotes tumor growth not through H-Ras in bladder cancer. Indeed, the rescue experiments showed that H-Ras KD prevented the increased cell proliferation and migration abilities induced by FTO KD (Fig. 6 A–C), H-Ras OE reversed the inhibition of YTHDF1 KD on the cell proliferation and migration abilities in 5637 cell line (Fig. 6 D–F). Additionally, based on the public RNA sequencing datasets (40, 41), we found that the up-regulated genes from FTO KD or the down-regulated genes from YTHDF1 KD were indeed enriched for Ras-mediated signaling pathway, including Ras signaling pathway, PI3K-AKT signaling pathway, and MAPK signaling pathway (SI Appendix, Fig. S13 G and H). These data suggest that FTO and YTHDF1 regulate cancer growth through H-Ras. Consequently, we further examined whether FTO and YTHDF1 regulate resistance of cancer cells to selumetinib, a kinase inhibitor targeting H-Ras downstream MEK1/2 (42). Cell proliferation assay showed that H-Ras KD prevented the acquired resistance to selumetinib by FTO KD, and H-Ras OE reversed the reduction of YTHDF1 KD on the resistance to selumetinib (Fig. 6 G and H). All these data reveal that H-Ras is a major target of FTO/YTHDF1 axis to regulate cancer growth.

Fig. 6.

FTO/YTHDF1/H-Ras axis regulate cancer cell growth. (A–F) Cell proliferation assay (A and D) assessing FTO KD 5637 cells with H-Ras KD, or YTHDF1 KD 5637 cells ectopically expressing H-Ras. Representative micrographs (B and E) and quantification (C and F) of the above-mentioned cells in the Matrigel-noncoated Transwell assays (n = 3). (G and H) Cell proliferation assay assessing the half-maximal inhibitory concentration (IC₅₀) in 5637 cells transfected with indicated vectors following treatment with selumetinib for 2 d. (I and J) The log₂ Flod changes in FTO, HRAS, and YTHDF1 expression in cancer tissues relative to matched normal tissues (I) and the correlation between HRAS and FTO or YTHDF1 (J) across 21 cancer types in TCGA datasets. (K–N) IHC analysis of FTO, H-Ras, and YTHDF1 in bladder (K) and breast (M) tumor tissue microarray. Representative images of tissue sections are shown (L and N). Spearman correlation analysis between HRAS and FTO or YTHDF1 (O and P). The tissue microarrays containing 63 cancer tissues and 16 adjacent tissues for bladder tumor, 119 cancer tissues and 40 adjacent tissues for breast tumor. (Q) Proposed model for the FTO/YTHDF1/H-Ras axis regulate cancer cell growth. (Scale bars, 20 µm.) *P < 0.05, **P < 0.01, and ***P < 0.001.

To further determine the clinical relevance of FTO/YTHDF1/H-Ras axis, we first explored the TCGA datasets, and found that FTO expression was down-regulated, while the HRAS and YTHDF1 were up-regulated in most cancer types (Fig. 6I). More importantly, the expression of HRAS was negatively correlated with FTO, but positively correlated with YTHDF1 expression in most cancer types (Fig. 6J). Furthermore, we performed immunohistochemistry (IHC) staining in bladder and breast cancer. Both IHC results from bladder and breast tumor tissue microarray showed that FTO expression significantly reduced, while H-Ras and YTHDF1 expression significantly increased in the cancer tissues versus normal tissues (Fig. 6 K–N). In particular, a strong positive correlation was observed between YTHDF1 and H-Ras expression in bladder (r = 0.870, p < 0.01) and weak positive correlation in breast (r = 0.343, p < 0.01) cancers, while FTO and H-Ras expression had no significant relevance in bladder cancer (r = 0.206, p > 0.05), but significant negatively correlated in breast cancer (r = -0.357, p < 0.01) (Fig. 6 O and P). These results supported that H-Ras is clinically associated with FTO and YTHDF1 in cancers.

Discussion

Ras is the most frequently mutated gene family in cancers (3). The “hot-spot” mutations lead to Ras retain in GTP-bound state, and directly activate several downstream signaling pathways including the MAPK and PI3K pathways, and is involved in the progression of human cancers (43). Additionally, emerging evidence suggests that posttranslational modification regulated the stabilization of Ras proteins and associated with tumorigenesis (11, 44, 45). Although microRNA-mediated regulation of RAS family genes has well been studied (46), posttranscriptional modification and regulation of the RAS genes remain poorly understood. In this study, we identified that HRAS mRNA, but not KRAS and NRAS is modified by m⁶A and this methylation increases HRAS protein expression. However, we are not able to exclude the m⁶A modification of KRAS and NRAS in other different conditions, while we used the limited tissues and cell lines for the test of KRAS and NRAS.

The m⁶A modification of HRAS was revealed in human lung cancer cells, and mouse scleroderma model, based on meRIP-seq (47) and epitranscriptomic microarray (48), receptively. However, the roles and mechanisms of m⁶A in regulating the fate of HRAS have not yet been elucidated. Our study revealed that higher m⁶A modification of HRAS in tumors increased its protein expression. This regulation occurs mainly during translation elongation in cancer cells that we studied, i.e., FTO-mediated m⁶A modification of HRAS enhances translation elongation and YTHDF1, but not YTHDF2 nor YTHDF3, likely mediates this enhancement by recruiting eEF1. Hence, our study provides a mechanism for H-Ras protein regulation, which differs from the well-studied of Ras protein stability regulated by Wnt/β-catenin signaling and distinct ubiquitin ligases (11, 49, 50). HRAS is the less frequently mutated RAS gene in cancers (4) and frequently overexpressed in some types of tumors (11), which highlight that the gain of H-Ras protein contributes toward H-Ras-driven tumorigenesis. Despite multiple attempts to develop Ras blockers, including targeting Ras directly or indirectly, such as Ras pathways (43), efficacy of these strategies is limited (51). Our identified specific m⁶A modification sites of HRAS associated with tumor progression could be a promising approach for therapeutic target of Ras-associated malignancies.

m⁶A regulatory proteins play distinct roles in regulating tumor development for various cancer types. m⁶A reader YTHDF1 is a well-known oncogene in ovarian (52), gastric (53), lung (27) and breast cancers (54). Whereas demethylase FTO and ALKBH5 function as an oncogene or tumor suppressor in a cancer-dependent manner (55–58). H-Ras is a mainly downstream effector for FTO-, but not ALKBH5-mediated function based on the functional assays in our study, which may be further demonstrated by the TCGA results that FTO, but not ALKBH5, is negatively associated with HRAS in almost all cancer types. In addition, we speculate that ALKBH5 regulates cancer progression through other targets, such as PD-L1 (59) and FOXM1 (60). Given the known roles of FTO and YTHDF1 in cancers, we speculate that HRAS may be a major effector of YTHDF1 but not FTO in various cancers. Thus, we provide a regulator of Ras effector signaling, which will be important to develop Ras-targeted therapeutic agents focusing on the function of YTHDF1. The function and molecular mechanism of m⁶A-modified HRAS in other pathological conditions need to be further investigated.

In summary, we provide compelling evidence that HRAS harbors three m⁶A sites in its 3′ UTR region, which enhance its protein expression by FTO and YTHDF1 in an m⁶A-dependent manner, thus enhancing cell proliferation and metastasis (Fig. 6Q). Our study highlights the functional importance of the specific m⁶A modification sites of oncogene in cancer, and provides profound insights into the molecular mechanisms underlying tumorigenesis by revealing the mechanism of these m⁶A modification sites in cancer.

Materials and Methods

MeRIP-Seq Data Analyses.

MeRIP-Seq data were collected from Gene Expression Omnibus (GEO), and were analyzed according to the protocol in previous publications (61). Briefly, TopHat2 (version 2.2.1) with Bowtie1 support (62, 63) was run to align the sequence reads to reference genome and transcriptome (hg38). Then the exomePeak R/Bioconductor package (version 3.7) (61) was used to find the differential m⁶A peaks between the tumor and adjacent samples. Significant peaks with P value less than 0.05 were annotated to the RefSeq database (hg38). The mapping reads were converted to normalized BigWig files for genomic visualization using Deeptools (version 3.4.1) (64), and normalized by CPM with reads coverage for each window of 100 bp (Counts per million mapped reads). Integrative Genomics Viewer was used to visualize the m⁶A peaks.

Cell Lines and Culture.

Human cancer 5637, HepG2, T24, 769-P, AsPC-1, HCC827, Hela, MCF7, PC-3, SW480, and HGC27 cells were purchased from the Shanghai Cell Bank Type Culture Collection Committee (Shanghai, China) and cultured in RPMI1640 medium (Gibco) with 10% fetal bovine serum and 1% penicillin/streptomycin (Invitrogen), while Hela and MCF7 were cultured in Dulbecco’s modified Eagle medium (Gibco). These cell lines were verified by short tandem repeat assays for their identification and tested negative for mycoplasma contamination, and cultured at 37 °C in a humidified incubator with 5% CO₂.

In vivo Animal Assays.

A total of 5637 cells with ectopic expression of dm⁶ACRISPR systems were injected subcutaneously (1 × 10⁷ cells/inoculum) into the flanks of 5-wk-old nude mice (Shanghai Model Organisms Center). Tumor formation/growth was assessed until the experimental endpoint, and tumor volume was calculated by the formula: (width)² × length/2. For tail vein injection, 5637 cells with ectopic expression of dm⁶ACRISPR systems (5 × 10⁶ cells/0.1 mL PBS) were injected into the lateral tail vein of 5-wk-old nude mice. Mice were euthanized after 4 to 5 wk. The number of metastatic foci in the lung was determined using the H&E staining (Beyotime Biotechnology) in tissue sections under a binocular microscope (Leica). All protocols involving animals were previously approved by the Ethics Committee for the Use of Experimental Animals of the Suzhou Institute of Biomedical Engineering and Technology, Chinese Academy of Sciences (Suzhou, Jiangsu, China).

Statistical Analysis.

Data were presented as mean ± SEM or SD. Two-tailed Student’s t tests were performed to assess the statistical significance of differences between groups. Spearman correlation was performed to analyze the correlation. P < 0.05 was considered statistically significant. All statistical analyses were performed using GraphPad Prism 8.0 or R software (version 3.6.1). *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant.

Additional methods can be found in SI Appendix, Material and Methods.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Data, Materials, and Software Availability

RNAseq data have been deposited in NCBI GEO database (GSE211698) (65). All other data are included in the manuscript and/or supporting information.