N6-methyladenosine facilitates mitochondrial fusion of colorectal cancer cells via induction of GSH synthesis and stabilization of OPA1 mRNA

ABSTRACT

Mitochondria undergo fission and fusion that are critical for cell survival and cancer development, while the regulatory factors for mitochondrial dynamics remain elusive. Herein we found that RNA m⁶A accelerated mitochondria fusion of colorectal cancer (CRC) cells. Metabolomics analysis and function studies indicated that m⁶A triggered the generation of glutathione (GSH) via the upregulation of RRM2B—a p53-inducible ribonucleotide reductase subunit with anti-reactive oxygen species potential. This in turn resulted in the mitochondria fusion of CRC cells. Mechanistically, m⁶A methylation of A1240 at 3′UTR of RRM2B increased its mRNA stability via binding with IGF2BP2. Similarly, m⁶A methylation of A2212 at the coding sequence (CDS) of OPA1—an essential GTPase protein for mitochondrial inner membrane fusion—also increased mRNA stability and triggered mitochondria fusion. Targeting m⁶A through the methyltransferase inhibitor STM2457 or the dm⁶ACRISPR system significantly suppressed mitochondria fusion. In vivo and clinical data confirmed the positive roles of the m⁶A/mitochondrial dynamics in tumor growth and CRC progression. Collectively, m⁶A promoted mitochondria fusion via induction of GSH synthesis and OPA1 expression, which facilitated cancer cell growth and CRC development.

INTRODUCTION

Mitochondrial dynamics, which comprises mitochondrial fission and fusion, is a variation of mitochondrial morphology in response to different physiological conditions [1]. The fusion/fission process is regulated by homo- and heterotypic interactions between mitofusin 1 (MFN1) and MFN2 at the outer mitochondrial membrane and optic atrophy 1 (OPA1) at the inner mitochondrial membrane [2]. The function of mitochondrial fusion is critical for maintaining a healthy mitochondrial population [3]. For example, OPA1 is crucial for inner mitochondrial membrane (IMM) fusion and the arrangement of electron transport chain (ETC) super complexes, whilst loss of OPA1 causes the collapse of the mitochondrial network, impairs mitochondrial morphology and promotes apoptosis [4]. In addition, mitochondrial dynamics and bioenergetics reciprocally influence each other [5]. Increased oxidative phosphorylation (OXPHOS) activity can stimulate mitochondrial fusion to cause elongation [6].

Studies have revealed that dysfunction of mitochondrial dynamics is critical for cancer progression by disturbing cellular apoptosis, energy production and signal transduction [7,8]. With regard to colorectal cancer (CRC)—the third most prevalent cancer and the second most lethal malignancy in the world [9]—mitochondrial dynamics contributes to the pathogenesis of intestinal inflammation, colorectal tumorigenesis and CRC transformation [10]. For example, fatty acid-induced mitochondrial fission potentiated Wnt signaling in colon cancer to regulate in vivo formation of tumor organoids and growth of a xenograft tumor [11]. Hypoxia-induced activation of the OMA1–OPA1 axis increased mitochondrial reactive oxygen species (ROS) and promoted glycolysis in CRC cells [12]. Given that mitochondria dynamics plays a significant role in maintaining cellular structure and function, investigations into its regulatory factors will be greatly helpful for identification of therapy targets for cancer treatment.

N ⁶-methyladenosine (m⁶A) is the most abundant mRNA modification in eukaryotes that has been identified since the 1970s [13,14]. Dynamic m⁶A modification in a specific transcript is governed by m⁶A methyltransferase complexes composited by methyltransferase-like 3 (METTL3), METTL14, Wilms' tumor 1-associating protein (WTAP) and m⁶A demethylases fat mass and obesity-associated protein (FTO), and AlkB homolog 5 (ALKBH5) [15]. m⁶A modification of mRNA was recognized using m⁶A readers including YTH-domain family proteins (YTHDF1/2/3) and the insulin-like growth factor 2 (IGF2) mRNA-binding protein family (IGF2BP1/2/3) to regulate mRNA translation, cellular location and degradation, and modulate protein production [14,16]. As a fundamental regulatory factor for gene expression, m⁶A plays a role in various biological processes, including cell differentiation, embryonic development and cancer development [14].

Recent studies have identified a connection between mitochondrial dynamics and a number of epigenetic regulatory factors [17]. It has been reported that SIRT3-mediated deacetylation of OPA1 at K931, as well as a novel site at K834, is important in mediating mitochondrial elongation [18] while DNMT1 maintains the metabolic fitness of adipocytes through acting as an epigenetic safeguard of mitochondrial dynamics [19]. Little evidence has indicated that m⁶A may be critical for mitochondrial fission/fusion and metabolism. For example, FTO promotes growth and metastasis of gastric cancer via metabolic regulation of mitochondrial dynamics [20]. However, whether m⁶A modification could directly regulate mitochondrial fission/fusion and affect the metabolism of cancer cells remains unknown. In the present study, we investigated the potential effects of m⁶A on mitochondrial dynamics of CRC cells. Our data revealed that m⁶A can positively promote mitochondria fusion of CRC cells via induction of glutathione (GSH) synthesis and stabilization of OPA1 mRNA.

RESULTS

m⁶A facilitated mitochondria fusion of CRC cells

To quantitatively assess the effects of m⁶A on mitochondrial homeostasis, we generated sh-control and sh-METTL3 RKO (cell line name) and HCT-116 cells with lentiviruses (Fig. S1A). Staining with MitoTracker Green showed that sh-control RKO cells contained an extended network of elongated mitochondria uniformly distributed in the perinuclear region, while sh-METTL3 cells had mostly smaller and shorter rounded mitochondria (Fig. 1A). Mitochondrial length was measured using ImageJ-MiNA and categorized as elongated (˃3 μm), intermediate (0.5–3 μm) or fragmented (˂0.5 μm) types [21]. Results showed that the fragmented mitochondria were increased significantly in sh-METTL3 RKO cells (Fig. 1A). This phenomenon was similarly observed in HCT-116 cells (Fig. S1B). Furthermore, electron microscopy (EM) showed that sh-control RKO (Fig. 1B) and HCT-116 (Fig. 1C) cells displayed exaggerated mitochondrial elongation and swollen mitochondria in contrast to the sh-METTL3 cells (P < 0.001). Statistical results indicated that sh-METTL3 RKO or HCT-116 cells resulted in a 45% or 31% decrease in mitochondrial size in the corresponding control cells, which was associated with an increase in the circularity index (from 0.79 ± 0.02 to 0.87 ± 0.02, 0.64 ± 0.02 to 0.83 ± 0.02, respectively) (Fig. 1B and C). In addition, knock-down of METTL3 in RKO cells decreased mitochondrial cristae and caused a relative increase in abnormal mitochondrial cristae that were crooked and disordered (Fig. S1C). These data indicated that knock-down of METTL3 modulated the mitochondrial morphology in CRC cells.

Figure 1.

m⁶A facilitated mitochondria fusion of CRC cells. (A) Representative confocal images of mitochondrial morphology in sh-control (sh-NC) and sh-METTL3 (sh-M3) RKO cells (left) and the percentages of mitochondrial types were measured (right). Elong, elongated type (˃3 μm); Inter, intermediate type (0.5–3 μm); Frag, fragmented type (˂0.5 μm) (scale bar, 20 μm). (B and C) Representative transmission electron microscopy images of the mitochondria in sh-control and sh-METTL3 RKO (B) or HCT-116 (C) cells. The average mitochondrial area and circularity indexes were quantified via transmission electron microscope (TEM) analysis (scale bar, 200 nm). (D) Mitochondrial membrane potential (Δψm) was measured in sh-control and sh-METTL3 RKO cells using flow cytometry (left) and median fluorescence intensity (MFI) was measured (right). (E) Intracellular ATP levels were determined via luciferin/luciferase-based assays in sh-control and sh-METTL3 RKO and HCT-116 cells. (F) The percentages of apoptosis cells in sh-control and sh-METTL3 RKO cells treated with 3 μM H₂O₂ for 4 h. (G) MitoTracker Deep Red staining was performed on sh-control and sh-METTL3 RKO cells to stain the mitochondria for flow cytometric analysis (left) and MFI was measured (right). (H) Representative confocal images of the mitochondrial morphology in sh-control and sh-METTL3 RKO cells transfected with vector control, METTL3 WT plasmid, METTL3 DA mutant plasmid for 24 h (scale bar, 20 μm). (I) Mitochondrial membrane potential in sh-control and sh-METTL3 RKO cells transfected with vector control, METTL3 WT plasmid, METTL3 DA mutant plasmid for 24 h. (J) Intracellular ATP levels in sh-control and sh-METTL3 RKO cells transfected with vector control, METTL3 WT plasmid, METTL3 DA mutant plasmid for 24 h. Data are presented as mean ± SD from three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significant, using Student's t-test between two groups and using one-way ANOVA followed by Bonferroni test for multiple comparisons.

Mitochondrial fusion is critical for mitochondrial function and is associated with increased oxidative phosphorylation [22]. Flow cytometer results indicated that sh-METTL3 reduced the mitochondrial membrane potential (Δψm) in CRC cells (Fig. 1D and Fig. S1D). Moreover, knock-down of METTL3 resulted in a 29% and 34% decrease in adenosine triphosphate (ATP) levels in sh-METTL3 RKO and HCT-116 cells, respectively (Fig. 1E). Further, the sh-METTL3 RKO cells were significantly more sensitive to H₂O₂-induced cell apoptosis than the sh-control cells (Fig. 1F). This indicated that sh-METTL3 induced significant changes in mitochondrial function.

We subsequently aimed to determine whether these changes were attributable to mitochondrial biogenesis by the use of MitoTracker Deep Red assay. The flow cytometry results showed that sh-METTL3 elicited no significant changes in mitochondrial mass in the corresponding control cells (Fig. 1G and Fig. S1E). Consistently, mitochondrial DNA contents were comparable between sh-control and sh-METTL3 CRC cells (Fig. S1F). This suggested that METTL3 had no effect on mitochondrial biogenesis.

To confirm the essential roles of m⁶A in mitochondrial dynamics, sh-METTL3 RKO cells were transfected with wild-type (WT) METTL3 and catalytically inactive METTL3 mutant DA (D395A) [23] (Fig. S1G). Results showed that the overexpression of METTL3, rather than METTL3 DA mutant, reversed the mitochondrial size (Fig. 1H), membrane potential (Fig. 1I) and ATP levels (Fig. 1J) in RKO cells. Consistently, overexpression of the m⁶A demethylase ALKBH5 [24] (Fig. S1H) had similar effects on sh-METTL3, which resulted in a decrease in mitochondrial size (Fig. S1I), membrane potential (Fig. S1J) and ATP (Fig. S1K) in RKO cells. All these data suggested that m⁶A facilitated mitochondria fusion of CRC cells.

Glutathione was involved in m⁶A-induced mitochondrial fusion

Considering that mitochondrial dysfunction may alter cellular metabolites [25], we performed metabolomics analysis in sh-METTL3 RKO cells by using gas chromatography–mass spectrometry (GC–MS) and carrying out multivariate statistical analysis using Principal Component Analysis (PCA), Partial Least Squares Discrimination Analysis (PLS-DA) and Orthogonal PLS-DA (OPLS-DA) to evaluate the metabolic pattern changes (Fig. S2A). By analysing using PCA, PLS-DA and OPLS-DA, we could markedly distinguish sh-METTL3 cells from sh-control cells (Fig. S2A). In total, we identified 110 metabolites with significant changes in the METTL3 knock-down RKO cells (Fig. S2B and Table S1). Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of these metabolites attributed 155 metabolic-related pathways in sh-METTL3 RKO cells (Fig. 2A and Table S2). Among the identified pathways, glutathione metabolism, GABAergic synapse, and alanine, aspartate and glutamate metabolism were identified as the major target pathways in the METTL3-mediating metabolism of RKO cells via overlapping analysis of the top 20 P-values, impacts and hits (Fig. 2B). Constantly, metabolites in the glutathione metabolism pathway such as GSH, oxydized glutathione (GSSG), L-glutamic acid and ascorbate decreased, while spermidine and gamma-Glutamylalanine increased, which suggested reduced GSH generation in sh-METTL3 RKO cells (Fig. 2C and Fig. S2C). GSH and GSSG Assay Kit analysis confirmed that the knock-down of METTL3 resulted in a significant decrease in GSH and GSSG levels in both RKO and HCT-116 cells (Fig. 2D).

Figure 2.

Glutathione was involved in m⁶A-induced mitochondrial fusion. (A) KEGG enrichment analysis of these 110 metabolites variated in sh-METTL3 RKO cells. (B) The overlap among ‘Top 20 P-value’, ‘Top 20 impact’ and ‘Top 20 hits’ of the KEGG enrichment pathways. (C) The GSH and GSSG levels in sh-control and sh-METTL3 RKO cells using the metabolomics analysis. (D) The GSH and GSSG levels in sh-control and sh-METTL3 RKO or HCT-116 cells. (E) The total GSH (T-GSH, GSH + GSSG) levels in sh-control and sh-METTL3 RKO cells transfected with vector control, METTL3 WT plasmid, METTL3 DA mutant plasmid for 24 h. (F) The T-GSH levels in sh-control or sh-METTL3 RKO cells transfected with vector control or OPA1 plasmid for 24 h. (G and H) Representative confocal images of the mitochondrial morphology (G) and intracellular ATP levels (H) in RKO cells treated with 100 μM BSO for 24 h (scale bar, 20 μm). (I–K) Representative confocal images of the mitochondrial morphology (I), mitochondrial membrane potential (J) and intracellular ATP levels (K) in sh-METTL3 RKO cells that were treated with 100 μM GSH for 24 h (scale bar, 20 μm). Data are presented as mean ± SD from three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significant, using Student's t-test between two groups and using one-way ANOVA followed by Bonferroni test for multiple comparisons.

Further, sh-METTL3 RKO cells were transfected with WT METTL3 and catalytically inactive METTL3 mutant DA (D395A). Results showed that the overexpression of METTL3, rather than METTL3 DA mutant, reversed the total GSH (T-GSH, GSSG + GSH), GSH, GSSG levels in sh-METTL3 RKO cells (Fig. 2E and Fig. S2D) while overexpression of ALKBH5 suppressed the levels of T-GSH in RKO cells (Fig. S2E). This indicated that m⁶A positively regulated the generation of GSH in CRC cells.

Mitochondria is affected by GSH redox changes, as mitochondria cannot produce GSH themselves and an accumulation of GSSG can cause widespread oxidation of proteins including essential mitochondrial proteins of the ETC [26,27]. At the same time, mitochondrial networks are readily responsive to redox imbalances [28], which in turn may reduce the production of ROS or protect mitochondria from degradation by mitophagy [29]. In order to evaluate the cause relationship between GSH generation and mitochondria fusion, both sh-control and sh-METTL3 RKO cells were transfected with OPA1 (Fig. S2F). OPA1 regulates mitochondrial fusion and cristae structure in the IMM [30]. Overexpression of OPA1 affected GSH redox change (Fig. S2G), but had no significant effect on levels of T-GSH in either sh-control or sh-METTL3 RKO cells (Fig. 2F), suggesting that mitochondrial dynamics had no effect on GSH generation.

Recent research demonstrated that GSSG strongly induces mitochondrial fusion by the generation of disulphide-mediated mitofusin oligomers, with Guanosine triphosphate (GTP) hydrolysis [31]. The addition of GSSG consistently reversed sh-METTL3-suppressed mitochondria fusion (Fig. S2H). To test whether reduced GSH generation leads to a decrease in GSSG production, ultimately inducing mitochondrial fission in CRC cells, we used L-buthionine-sulfoximine (BSO) to block GSH synthesis [32]. Our data showed that BSO suppressed mitochondria fusion (Fig. 2G) and ATP generation (Fig. 2H) in RKO cells. Further, the addition of GSH reversed sh-METTL3-suppressed mitochondria fusion (Fig. 2I), membrane potential (Fig. 2J) and ATP generation (Fig. 2K) in RKO cells. All these data indicated that m⁶A triggers GSH generation to induce mitochondria fusion.

m⁶A-regulated GSH synthesis through stabilization of RRM2B mRNA

To identify potential targets involved in m⁶A-regulated GSH synthesis and fusion of mitochondria, we performed mRNA-seq in sh-control and sh-METTL3 RKO cells. Expression levels of 849 genes were found to be significantly changed with the upregulation of 97 and downregulation of 752 genes in sh-METTL3 RKO cells (Fig. S3A and Table S3). The genes varied in sh-METTL3 RKO cells and those involved in glutathione metabolism (Table S4, summarized using Gene Set Enrichment Analysis (GSEA)) [33] were overlap analysed. Overlap analysis showed that three candidates (down: RRM2B (ribonucleotide reductase regulatory TP53 inducible subunit M2B) and RRM1 (ribonucleotide reductase catalytic subunit M1), up: GPX3 (glutathione peroxidase 3)) were overlapping between the gene set and variated genes in sh-METTL3 RKO cells (Fig. 3A). qRT–PCR consistently showed that a reduction in METTL3 can increase the mRNA of GPX3 and decrease the mRNA of RRM2B and RRM1 in both RKO (Fig. 3B) and HCT-116 (Fig. S3B) cells. However, Western blot analysis showed that knock-down of METTL3 decreased the protein expression of RRM2B, while it had no consistent effect on GPX3 or RRM1, in RKO (Fig. 3C) and HCT-116 (Fig. S3C) cells.

Figure 3.

m⁶A-regulated GSH synthesis through stabilization of RRM2B mRNA. (A) The analysis process diagram to identify RRM2B as METTL3-regulated targets that were involved in glutathione metabolism pathways in CRC cells. (B and C) The mRNA (B) and protein (C) levels of RRM2B, RRM1 and GPX3 in sh-control and sh-METTL3 RKO cells. (D and E) The T-GSH levels (D) and representative confocal images of the mitochondrial morphology (E) in sh-control and sh-METTL3 RKO cells transfected with vector control or RRM2B plasmid for 24 h. (F) Representative transmission electron microscopy images of the mitochondria in sh-control, sh-METTL3 and RRM2B stable overexpression sh-METTL3 RKO cells. The average mitochondrial area and circularity indexes were quantified via TEM analysis (scale bar, 200 nm). (G) m⁶A peaks were enriched in RRM2B mRNA from m⁶A RIP-seq data. (H) m⁶A RIP–qPCR analysis of RRM2B in sh-control and sh-METTL3 RKO cells. (I) The protein expression of RRM2B in RKO cells transfected with vector control, METTL3 WT plasmid or METTL3 DA mutant plasmid for 24 h. (J and K) The protein (J) and mRNA (K) expression of RRM2B in RKO cells transfected with vector control or ALKBH5 plasmid for 24 h. (L and M) After treatment with Act-D for the indicated times, the mRNA levels of RRM2B were checked in sh-control and sh-METTL3 RKO (L) or HCT-116 (M) cells. Data are presented as mean ± SD from three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significant, using Student's t-test between two groups and using one-way ANOVA followed by Bonferroni test for multiple comparisons.

RRM2B (also known as p53R2) is a p53-inducible ribonucleotide reductase subunit that exhibits anti-ROS potential [34]. We therefore investigated its roles in m⁶A-regulated GSH generation and mitochondria fusion of CRC cells. Results showed that overexpression of RRM2B (Fig. S3D) can reverse the downregulation of T-GSH levels (Fig. 3D) and mitochondria fusion (Fig. 3E and F) in sh-METTL3 RKO cells. It indicated the essential roles of RRM2B in m⁶A-regulated GSH generation and mitochondria fusion of CRC cells.

We further investigated whether m⁶A regulated the expression of RRM2B via direct methylation of mRNA. m⁶A-RIP-seq data showed that the 3′UTR region of RRM2B mRNA was modified by m⁶A (Fig. 3G). m⁶A-RIP–qPCR confirmed that a 4-fold enrichment of m⁶A antibody was observed in RRM2B mRNA in RKO cells and this enrichment significantly decreased in sh-METTL3 RKO cells (Fig. 3H). Similar results were also observed in HCT-116 cells (Fig. S3E). Furthermore, the overexpression of WT METTL3, but not the METTL3 DA mutant, reversed the levels of RRM2B in RKO cells (Fig. 3I) while the overexpression of ALKBH5 suppressed the protein expression of RRM2B in RKO cells (Fig. 3J). This indicated that RRM2B was modified by m⁶A and that m⁶A positively regulates the expression of RRM2B in CRC cells.

We further investigated the potential mechanisms involved in the m⁶A-regulated expression of RRM2B. To determine whether METTL3 can modulate the transcription of RRM2B, we performed a luciferase reporter assay by transfecting the promoter reporter gene plasmid pGL3-Basic-RRM2B-luc into RKO cells. There was no significant difference in the luciferase activity of the RRM2B promoter between sh-control and sh-METTL3 cells (Fig. S3F), suggesting that METTL3 does not affect the transcription of RRM2B. This was further confirmed using qRT–PCR analysis, which showed comparable levels of the precursor mRNA of RRM2B between sh-control and sh-METTL3 CRC cells (Fig. S3G). Additionally, fractionation assay results indicated that there was no difference in the subcellular localization of RRM2B mRNA in sh-control and sh-METTL3 RKO cells (Fig. S3H). This indicated that m⁶A had no effect on the subcellular localization of RRM2B mRNA.

However, knock-down of METTL3 (Fig. 3B) or overexpression of ALKBH5 (Fig. 3K) significantly decreased the mRNA expression of RRM2B in CRC cells. Since m⁶A had no effect on promoter activity but regulated the mRNA expression of RRM2B, we then tested its effect on mRNA stability. Our results showed that knock-down of METTL3 significantly decreased the mRNA stability of RRM2B in both RKO (Fig. 3L) and HCT-116 (Fig. 3M) cells. Therefore, the m⁶A-regulated expression of RRM2B should be down to the positive effects of m⁶A in the mRNA stability of RRM2B mRNA.

Regarding the translation efficiency of the endogenous RRM2B mRNA, which is defined as the ratio of protein production (RRM2B/GAPDH) to mRNA abundance [35], knock-down of METTL3 had no effect on the translation efficiency of RRM2B in RKO cells (Fig. S3I). Further, polysome profiling analysis (Fig. S3J) confirmed that the relative ribosome occupancy of RRM2B in monosome and polysome fractions (levels of RRM2B mRNA/total RNA) were comparable between sh-control and sh-METTL3 RKO cells (Fig. S3K). To assess whether m6A can post-translationally regulate the expression of RRM2B, both sh-control and sh-METTL3 RKO cells were further treated with cycloheximide (CHX) to inhibit translation. Our data revealed that the half-life of RRM2B protein had no significant difference between sh-Control and sh-METTL3 cells (Fig. S3L). All these data indicated that METTL3 positively regulates the mRNA stability of RRM2B, without affecting its transcription, nuclear export, translation efficiency or protein stability.

m⁶A stabilized RRM2B mRNA via methylation of A1240 at 3′UTR

We further investigated the methylation site and reader protein responsible for m⁶A-stablized RRM2B mRNA. m⁶A-RIP–PCR using fragmented poly⁺ RNA indicated that the relative enrichment of 3′UTR of RRM2B mRNA was much greater than that of coding sequence (CDS) in RKO cells (Fig. 4A), which is consistent with m⁶A-seq results (Fig. 3G). To investigate whether m⁶A-methylated 3′UTR was involved in the m⁶A-regulated mRNA stability of RRM2B mRNA, we constructed 3′UTR reporters containing wild-type RRM2B 3′UTR behind the firefly luciferase reporter gene by use of the pmiR-GLO vector (Fig. 4B). The luciferase assay illustrated that the levels of F-Luc in sh-METTL3 RKO cells were significantly decreased, which was due to the downregulation of F-Luc mRNA and not the translation efficiency (Fig. 4C). Further, the addition of RRM2B 3′UTR decreased the half-life of F-Luc mRNA in sh-METTL3 RKO cells (Fig. 4D). It further confirmed that m⁶A-methylated 3′UTR mediated the METTL3-regulated mRNA stability of RRM2B.

Figure 4.

m⁶A stabilized RRM2B mRNA via methylation of A1240 at 3′UTR. (A) The m⁶A in 5′UTR, CDS or 3′UTR of RRM2B in RKO cells were analysed via m⁶A-RIP–qPCR using fragmented RNA. (B) Schematic representation of pmirGLO-RRM2B 3′UTR reporter. (C) The mRNA abundance, luciferase activity and translation efficiency of F-Luc in sh-control and sh-METTL3 RKO cells transfected with pmirGLO-RRM2B 3′UTR reporter for 24 h. (D) After treatment with Act-D for the indicated times, the mRNA levels of F-Luc were checked in sh-control and sh-METTL3 RKO cells transfected with pmirGLO-RRM2B 3′UTR reporter for 24 h. (E) RIP–qPCR analysis of RRM2B mRNA in RKO cells by use of antibody of YTHDF2, YTHDF3 and IGF2BP1-3. (F) IGF2BP2 RIP–qPCR analysis of RRM2B mRNA in sh-control and sh-METTL3 RKO cells. (G and H) The mRNA (G) and protein (H) expression of RRM2B in RKO and HCT-116 cells transfected with si-NC or si-IGF2BP2 for 24 h. (I) After treatment with Act-D for the indicated times, the mRNA levels of RRM2B were checked in RKO cells transfected with si-NC or si-IGF2BP2 for 24 h. (J) The protein expression of RRM2B in sh-control and sh-METTL3 RKO cells transfected with vector control or IGF2BP2 plasmid for 24 h. (K) Schematic representation of positions of m⁶A motifs within RRM2B mRNA. (L) The threshold cycle (Ct) of qPCR showing SELECT results for detecting m⁶A site in the potential m⁶A site of RRM2B in sh-control and sh-METTL3 RKO and HC-T116 cells. (M) m⁶A-Rol-LAMP products of methylation modified site (A1240) and negative site (A1235) of RRM2B mRNA in sh-control or sh-METTL3 RKO cells were detected using SYBR Green I (the trade name of a DNA dye). (N) Schematic representation of mutation in 3′UTR to investigate the m⁶A roles on RRM2B expression. (O) The relative luciferase activity of F-Luc/R-Luc of pmirGLO-RRM2B 3′UTR WT, Mut1 and Mut2 reporter in sh-control and sh-METTL3 RKO cells. (P) After treatment with Act-D for the indicated times, the mRNA levels of F-Luc were checked in sh-control and sh-METTL3 RKO cells transfected with pmirGLO-RRM2B 3′UTR WT, Mut1 and Mut2 reporter for 24 h. Data are presented as mean ± SD from three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significant, using Student's t-test between two groups and using one-way ANOVA followed by Bonferroni test for multiple comparisons.

The m⁶A binding proteins insulin-like growth factor 2 mRNA-binding proteins (IGF2BPs; including IGF2BP1/2/3), but not YTH-domain family proteins YTHDF2/3, recognize and stabilize m⁶A-modified cellular RNAs [36]. RIP–qPCR showed that IGF2BP2 (not IGF2BP1/3 or YTHDF2/3) can bind with RRM2B mRNA in RKO cells (Fig. 4E). Further, the binding between IGF2BP2 and RRM2B mRNA was decreased in sh-METTL3 RKO cells (Fig. 4F). Knock-down of IGF2BP2 was able to suppress the mRNA (Fig. 4G) and protein (Fig. 4H) expression of RRM2B in both RKO and HCT-116 cells. It was due to that knock-down of IGF2BP2 that there was suppression of the mRNA stability of RRM2B (Fig. 4I). Furthermore, overexpression of IGF2BP2 reversed the knock-down of METTL3-suppressed mRNA (Fig. S4A) and protein (Fig. 4J) expression of RRM2B. All these data suggested that IGF2BP2 mediated m⁶A-regulated expression of RRM2B.

We further investigated the methylation site for m⁶A-regulated mRNA stability of RRM2B. One GGAC motif with A1240 in RRM2B 3′UTR was identified (Fig. 4K), which was consistent with the positions and numbers of peaks identified by using m⁶A RIP-seq. In order to confirm the methylation of A1240 within RRM2B 3′UTR, its position was checked using the ‘SELECT′ method [37] in RKO and HCT-116 cells. SELECT showed that knock-down of METTL3 decreased the methylation levels of A1240 (Fig. 4L) while the nearby nucleotide A1235 without m⁶A modification had a significantly lower Ct value than that of A1240 (Fig. S4B). The methylation of A1240 was further confirmed using our recently developed m⁶A-Rol-LAMP based on rolling circle amplification and loop-mediated isothermal amplification (LAMP) (Fig. 4M and Fig. S4C).

We further mutated A1240 within RRM2B 3′UTR to investigate its roles in the m⁶A-regulated mRNA stability of RRM2B (Fig. 4N). The mutation of A1240, while not at the control site (A1235) in 3′UTR, resulted in a downregulation of luciferase activity of F-Luc (Fig. 4O) whereas the mutation-induced downregulation of luciferase activity of F-Luc was reversed in sh-METTL3 cells. Further, the mutation of 3′UTR A1240 decreased the mRNA stability of F-Luc (Fig. 4P). All data confirmed that A1240 within RRM2B 3′UTR mediated m⁶A-regulated mRNA stability of RRM2B.

m⁶A-stablized OPA1 mRNA regulated mitochondrial fusion

Previous studies indicated that key proteins involved in mitochondrial dynamics such as mitochondrial biogenesis (PPARGC1A, TFAM), fusion (MFN1, MFN2, OPA1) and fission (DRP1) may be m⁶A-methylated [1,38]. We further investigated whether these key proteins (PPARGC1A, TFAM, MFN1, MFN2, OPA1 and DRP1) participate in m⁶A-regulated mitochondrial fusion. Overlap analysis showed that two candidates (down: MFN1 and OPA1) overlap among the key proteins and variated genes in sh-METTL3 RKO cells (Fig. 5A). qRT–PCR showed that consistent knock-down of METTL3 decreased the mRNA expression of MFN1 and OPA1 in both RKO (Fig. 5B) and HCT-116 (Fig. S5A) cells. Western blot analysis showed that knock-down of METTL3 only decreased the protein expression of OPA1, while it had no effect on MFN1, in both RKO and HCT-116 cells (Fig. 5C).

Figure 5.

m⁶A-stablized OPA1 mRNA regulated mitochondrial fusion. (A) The analysis process diagram to identify OPA1 as METTL3-regulated targets that were involved in mitochondrial dynamics in CRC cells. (B) The mRNA expression of PGC-1α, TFAM, Mfn1, Mfn2, OPA1 and Drp1 in sh-control and sh-METTL3 RKO cells. (C) The protein expression of TOM20, MFN1, MFN2, OPA1 and DRP1 in sh-control and sh-METTL3 CRC cells. (D) Representative confocal images of the mitochondrial morphology in sh-control and sh-METTL3 RKO cells transfected with vector control, OPA1 plasmid for 24 h (scale bar, 20 μm). (E) Representative transmission electron microscopy images of the mitochondria in sh-control, sh-METTL3 and OPA1 stable overexpression sh-METTL3 HCT116 cells. The average mitochondrial area and circularity indexes were quantified via TEM analysis (scale bar, 200 nm). (F and G) Mitochondrial membrane potential (F) and intracellular ATP levels (G) in sh-control and sh-METTL3 RKO cells transfected with vector control, OPA1 plasmid for 24 h (scale bar, 20 μm). (H) m⁶A RIP–qPCR analysis of OPA1 in sh-control and sh-METTL3 RKO cells. (I) The protein expression of OPA1, MFN1 and MFN2 in RKO cells transfected with vector control, METTL3 WT plasmid or METTL3 DA mutant plasmid for 24 h. (J) The mRNA levels of OPA1 in RKO cells transfected with vector control, ALKBH5 plasmid for 24 h. (K) Cells were transfected with pGL3-Basic-OPA1-luc reporter and pRL-TK plasmid for 24 h. The promoter activities were presented as the ratios between the reporter and pRL-TK plasmid. (L and M) After treatment with Act-D for the indicated times, the mature mRNA levels of OPA1 were checked in sh-control and sh-METTL3 RKO (L) or HCT-116 (M) cells. Data are presented as mean ± SD from three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significant, using Student's t-test between two groups and using one-way ANOVA followed by Bonferroni test for multiple comparisons.

Although the promotional roles of OPA1 in mitochondria have been well illustrated [39], we further checked its roles in m⁶A-regulated mitochondria fusion of CRC cells. Results showed that overexpression of OPA1 (Fig. S2B) can reverse the reduced mitochondria fusion (Fig. 5D and E), membrane potential (Fig. 5F) and ATP generation (Fig. 5G) of sh-METTL3 RKO cells. This confirmed the essential role of OPA1 in m⁶A-regulated mitochondria fusion of CRC cells.

OPA1 is cleaved by mitochondrial peptidases OMA1 and YME1L in a stress-dependent manner [40]. Knock-down of METTL3 had no significant effect on the mRNA expression of either OMA1 or YME1L in RKO cells (Fig. S5B). We further investigated whether m⁶A regulated the expression of OPA1 via directly methylating its mRNA. m⁶A-seq data confirmed that mRNA of OPA1 was m⁶A-methylated (Fig. S5C). m⁶A-RIP–qPCR confirmed a 2-fold m⁶A antibody-enriched OPA1 mRNA in RKO cells, while this enrichment significantly decreased in sh-METTL3 RKO cells (Fig. 5H). Similar results were also observed in HCT-116 cells (Fig. S5D). This indicated that OPA1 mRNA was m⁶A-methylated.

Further, the transfections of WT METTL3, but not catalytically inactive METTL3 mutant DA (D395A), increased the expression of OPA1 in RKO cells (Fig. 5I) while neither WT-METTL3 nor inactive METTL3 mutant had an effect on the protein expression of MFN1 or MFN2 in RKO cells (Fig. 5I). Further, overexpression of ALKBH5 decreased the mRNA (Fig. 5J) and protein (Fig. S5E) expression of OPA1 in RKO cells. This indicated that METTL3 can positively regulate the expression of OPA1 in CRC cells in an m⁶A-dependent manner.

We further investigated the mechanisms for m⁶A-regulated expression of OPA1. Similarly to RRM2B, the luciferase reporter assay using pGL3-Basic-OPA1-luc showed that there was no significant difference in the luciferase activity of the OPA1 promoter between sh-control and sh-METTL3 cells (Fig. 5K). qRT–PCR analysis consistently showed that the levels of precursor mRNA of OPA1 were comparable between sh-control and sh-METTL3 cells (Fig. S5F). This indicated that m⁶A had no significant effect on the transcription of OPA1. Similarly, the subcellular localization of OPA1 mRNA (Fig. S5G) and half-life of OPA1 protein (Fig. S5H) in sh-control and sh-METTL3 cells were comparable. In addition, knock-down of METTL3 had no effect on the translation efficiency of endogenous OPA1 mRNA in RKO cells (Fig. S5I). Further, ribosome profiling analysis confirmed that there is no significant difference between the ribosome occupancy of OPA1 in monosome and polysome fractions of sh-control and sh-METTL3 RKO cells (Fig. S5J). We then tested its effect on mRNA stability. The results showed that knock-down of METTL3 significantly decreased the mRNA stability of OPA1 in both RKO (Fig. 5L) and HCT-116 (Fig. 5M) cells. All these data indicated that METTL3 positively regulated the mRNA stability of OPA1, while it had no effect on its transcription, nuclear export, translation efficiency or protein stability.

m⁶A methylation at A2212 stabilized OPA1 mRNA via binding with IGF2BP2

We further investigated the methylation site and reader protein responsible for m⁶A-stablized OPA1 mRNA. m⁶A-RIP–PCR using fragmented poly + RNA demonstrated a significantly greater relative enrichment of the CDS of OPA1 mRNA compared with the 5′UTR in RKO cells (Fig. 6A), which was consistent with the prediction results of OPA1 mRNA by use of the m⁶A sites predictor SRAMP (http://www.cuilab.cn/sramp) (Fig. S6A). To investigate whether m⁶A-methylated CDS were involved in m⁶A-regulated stability of OPA1 mRNA, we constructed CDS reporters by inserting the wild-type OPA1 CDS downstream of the firefly luciferase reporter gene by use of the pmiR-GLO vector (Fig. 6B). The luciferase assay revealed a significant decrease in the luciferase expression of F-Luc in sh-METTL3 RKO cells, which was attributed to the downregulation of F-Luc mRNA rather than a decrease in the translation efficiency (Fig. 6C). Further, the addition of OPA1 CDS decreased the half-life of F-Luc mRNA in sh-METTL3 RKO cells (Fig. 6D). This further confirmed that m⁶A-methylated CDS mediates the METTL3-regulated mRNA stability of OPA1.

Figure 6.

m⁶A methylation at A2212 stabilized OPA1 mRNA via binding with IGF2BP2. (A) The m⁶A in 5′UTR, CDS or 3′UTR of RRM2B in RKO cells were analysed via m⁶A-RIP–qPCR using fragmented RNA. (B) Schematic representation of pmirGLO-OPA1 CDS reporter. (C) The mRNA abundance, luciferase activity and translation efficiency of F-Luc in sh-control and sh-METTL3 RKO cells transfected with pmirGLO-OPA1 CDS reporter for 24 h. (D) After treatment with Act-D for the indicated times, the mRNA levels of F-Luc were checked in sh-control and sh-METTL3 RKO cells transfected with pmirGLO-OPA1 CDS reporter for 24 h. (E) RIP–qPCR analysis of OPA1 mRNA in RKO cells by use of antibody of YTHDF2, YTHDF3 and IGF2BP1∼3. (F) IGF2BP2 RIP–qPCR analysis of OPA1 mRNA in sh-control and sh-METTL3 RKO cells. (G and H) The mRNA (G) and protein (H) expression of OPA1 in RKO and HCT-116 cells transfected with si-NC or si-IGF2BP2 for 24 h. (I) After treatment with Act-D for the indicated times, the mRNA levels of OPA1 were checked in RKO cells transfected with si-NC or si-IGF2BP2 for 24 h. (J) The protein expression of OPA1 in sh-control and sh-METTL3 RKO cells transfected with vector control or ALKBH5 plasmid for 24 h. (K) Schematic representation of positions of m⁶A motifs within OPA1 mRNA. (L) The threshold cycle (Ct) of qPCR showing SELECT results for detecting m⁶A site in the potential m⁶A site of OPA1 in sh-control and sh-METTL3 RKO and HCT116 cells. (M) m⁶A-Rol-LAMP products of methylation modified sites (A2212) of OPA1 mRNA in sh-control or sh-METTL3 RKO and HC-T116 cells were detected using SYBR Green I. (N) Schematic representation of mutation in CDS to investigate the m⁶A roles on OPA1 expression. (O) The relative luciferase activity of F-Luc/R-Luc of pmirGLO-OPA1 CDS WT, Mut1 and Mut2 reporter in sh-control and sh-METTL3 RKO cells. (P) After treatment with Act-D for the indicated times, the mRNA levels of F-Luc were checked in sh-control and sh-METTL3 RKO cells transfected with pmirGLO-OPA1 CDS WT, Mut1 and Mut2 reporter for 24 h. Data are presented as mean ± SD from three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significant, using Student's t-test between two groups and using one-way ANOVA followed by Bonferroni test for multiple comparisons.

RIP–qPCR showed that IGF2BP2, and not IGF2BP1/3 or YTHDF2/3, can bind with OPA1 mRNA (Fig. 6E). Further, the binding between IGF2BP2 and OPA1 mRNA was decreased in sh-METTL3 RKO cells (Fig. 6F). Knock-down of IGF2BP2 suppressed the mRNA (Fig. 6G) and protein (Fig. 6H) expression of OPA1 in both RKO and HCT-116 cells. That knock-down of IGF2BP2 decreased the mRNA stability of OPA1 (Fig. 6I). Furthermore, overexpression of IGF2BP2 can reverse the knock-down of METTL3-suppressed mRNA (Fig. S6B) and protein (Fig. 6J) expression of OPA1. All data suggested that IGF2BP2 mediated m⁶A-regulated expression of OPA1.

As shown in Fig. 6K, one GGAC motif at exon 18 in OPA1 CDS was identified, which is consistent with the positions and numbers of peaks identified using the m⁶A sites predictor (Fig. S6A). The m⁶A methylation of A2212 within the GGAC motif at exon 18 in OPA1 CDS was confirmed using the ‘SELECT’ method in RKO and HCT-116 cells, whilst knock-down of METTL3 can decrease the methylation levels of A2212 (Fig. 6L), whereas the nearby nucleotide A2207 without m⁶A modification had a significantly lower Ct value than that of A2212 (Fig. S6C). The methylation of A2212 was further confirmed using our recently developed m⁶A-Rol-LAMP assay (Fig. 6M and Fig. S6D).

We further mutated A2212 within OPA1 CDS to investigate its roles in the m⁶A-regulated mRNA stability of OPA1 (Fig. 6N and Fig. S6E). The mutation of A2212 but not the control site (A2207) in CDS resulted in a decrease in F-Luc activity (Fig. 6O), which was abolished in sh-METTL3 cells. Further, the mutation of CDS A2212 decreased the mRNA stability of F-Luc (Fig. 6P). All these data confirmed that the methylation of A2212 within OPA1 CDS was involved in the m⁶A-regulated mRNA stability of OPA1.

Generally or specifically targeting m⁶A suppressed mitochondria fusion

Firstly, we evaluated the potential roles of STM2457, a highly potent and selective inhibitor of METTL3/14 [41], on mitochondria dynamics and expression of RRM2B and OPA1. Results showed that STM2457 significantly decreased the m⁶A enrichment of RRM2B and OPA1 in RKO cells (Fig. 7A). Further, STM2457 suppressed the mRNA (Fig. 7B) and protein (Fig. 7C) expression of RRM2B and OPA1 in RKO cells. In addition, STM2457 decreased the levels of T-GSH in RKO cells via a concentration-dependent manner (Fig. 7D). Staining with MitoTracker Green showed that treatment with STM2457 resulted in smaller and shorter rounded mitochondria (Fig. 7E). It confirmed that targeting m⁶A via the METTL3 inhibitor can suppress mitochondria fusion of CRC cells.

Figure 7.

Generally or specifically targeting m⁶A-regulated mitochondria fusion. (A) m⁶A RIP–qPCR analysis of RRM2B or OPA1 in RKO cells treated with METTL3 inhibitor STM2457 for 48 h. (B and C) The mRNA (B) or protein (C) levels of RRM2B or OPA1 in RKO cells treated with STM2457 for 48 h. (D) The T-GSH levels in RKO cells treated with STM2457 for 48 h. (E) Representative confocal images of the mitochondrial morphology in RKO cells treated with STM2457 for 48 h. (F) Schematic representation of positions of m⁶A site within RRM2B or OPA1 mRNA and the regions targeted by three gRNAs, respectively. (G) The threshold cycle of qPCR showing SELECT results for detecting m⁶A site in RRM2B or OPA1 in RKO cells transfected with dCas13b-ALKBH5 combined with gRNA negative control or gRNA1/2/3, respectively, for 24 h. (H) m⁶A-RIP–PCR analysis of RRM2B or OPA1 mRNA in RKO cells transfected with dCas13b-ALKBH5 combined with gRNA negative control or gRNA1/2/3, respectively, for 24 h. (I and J) The mRNA (I) and protein (J) levels of RRM2B or OPA1 in RKO cells transfected with dCas13b-ALKBH5 combined with gRNA negative control or gRNA1/2/3, respectively, for 24 h. (K and L) RKO cells were transfected with gRNA negative control, dCas13b-ALKBH5, with gRNA1 for RRM2B or gRNA3 for OPA1 for 24 h and then further treated with Act-D for the indicated times. The mRNA level of RRM2B (K) and OPA1 (L) was checked using qRT–PCR. (M) The T-GSH levels in RKO cells transfected with gRNA negative control, gRNA1/2/3 for RRM2B and dCas13b-ALKBH5 for 24 h. (N) The T-GSH levels in RKO cells transfected with gRNA negative control, gRNA1 for RRM2B and dCas13b-ALKBH5 or dCas13b-ALKBH5-Mut for 24 h. (O) The representative confocal images of the mitochondrial morphology and the percentages of mitochondrial types in RKO cells transfected with gRNA negative control, gRNA1 for RRM2B or gRNA3 for OPA1 and dCas13b-ALKBH5 for 24 h (scale bar, 20 μm). (P) Representative transmission electron microscopy images of the mitochondria in RKO cells transfected with gRNA negative control, gRNA1 for RRM2B or/and gRNA3 for OPA1 and dCas13b-ALKBH5; the average mitochondrial area was quantified via TEM analysis (scale bar, 200 nm). Data are presented as mean ± SD from three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significant, using Student's t-test between two groups and using one-way ANOVA followed by Bonferroni test for multiple comparisons.

We further specifically demethylated m⁶A of RRM2B and OPA1 via development of the dm⁶ACRISPR, which fuses the catalytically inactive Type VI-B Cas13 enzyme with the m⁶A demethylase ALKBH5, as previously described [42] (Fig. S7A). Three gRNAs were designed to target the 3′UTR of RRM2B mRNA or the CDS of OPA1 mRNA, respectively (Fig. 7F). Our data showed that co-transfection of wild-type Cas13b with three gRNAs can significantly reduce the mRNA levels of RRM2B and OPA1 (Fig. S7B), indicating that all gRNAs functioned efficiently.

SELECT–qPCR analysis showed a significant decrease in m⁶A levels at the targeted sites of RRM2B and OPA1 in RKO cells following transfection with gRNAs and dCas13b-ALKBH5 (Fig. 7G). Additionally, m⁶A-RIP–PCR confirmed that gRNAs and dCas13b-ALKBH5 can effectively decrease the m⁶A methylation of RRM2B and OPA1 in RKO cells (Fig. 7H). Results showed that dm⁶ACRISPR targeting RRM2B or OPA1 led to a significant downregulation of mRNA (Fig. 7I) and protein (Fig. 7J) expression of RRM2B and OPA1 in RKO cells. It was due to dm⁶ACRISPR with specific gRNA that there was a decrease in the mRNA stability of RRM2B (Fig. 7K) and OPA1 (Fig. 7L) in RKO cells.

We further investigated whether dm⁶ACRISPR targeting RRM2B and OPA1 can modulate mitochondria fusion. Our data showed that dm⁶ACRISPR targeting RRM2B significantly decreases the levels of T-GSH in RKO cells (Fig. 7M). However, dm⁶ACRISPR targeting OPA1 had no similar effects on the levels of T-GSH (Fig. S7C), but it affected GSH redox change (Fig. S7D). In addition, gRNA1 for RRM2B combined with dCas13b-ALKBH5 mutant did not produce a significant effect (Fig. 7N). Further, dm⁶ACRISPR targeting RRM2B and OPA1, respectively, resulted in smaller and shorter rounded mitochondria in RKO cells (Fig. 7O and P). The combination of gRNA for RRM2B and OPA1 showed a synergistic effect on mitochondria fusion (Fig. 7O and P). These data suggested that generally or specifically targeting m⁶A can regulate mitochondria fusion in CRC cells.

Oncogenic roles of m⁶A-regulated mitochondria dynamics in CRC development

We further checked the in vivo effect of m⁶A-regulated mitochondria dynamics in CRC progression. Cell viability assay showed that overexpression of RRM2B and OPA1 can reverse the suppressed growth effect of sh-METTL3 RKO cells (Fig. 8A). Further, dCas13b-ALKBH5 with gRNAs of RRM2B or OPA1 decreased the proliferation in both RKO (Fig. S7E) and HCT-116 (Fig. S7F) cells. Furthermore, our data suggested that knock-down of METTL3 enhanced the sensitivity of RKO cells to doxorubicin (Dox), yet overexpression of RRM2B and OPA1 was able to mitigate these effects and decrease Dox sensitivity (Fig. 8B). This suggested that mitochondria dynamics was involved in METTL3-regulated growth and chemosensitivity of CRC cells.

Figure 8.

Oncogenic roles of m⁶A-regulated mitochondria dynamics in CRC development. (A) The relative cell proliferation of sh-control, sh-METTL3 and RRM2B or OPA1 stable overexpression sh-METTL3 RKO cells. (B) sh-control, sh-METTL3 and RRM2B or OPA1 stable overexpression sh-METTL3 RKO cells were treated with increasing concentrations of Dox for 24 h and the cell proliferation was tested. (C) The tumor growth curves of sh-control, sh-METTL3 and RRM2B or OPA1 stable overexpression sh-METTL3 RKO xenografts. (D and E) The tumor volume (D) and tumor weight (E) of sh-control, sh-METTL3 and RRM2B or OPA1 stable overexpression sh-METTL3 RKO xenografts at the end of the experiment. (F) IHC (METTL3, RRM2B, OPA1 and Ki67)-stained paraffin-embedded sections obtained from sh-control, sh-METTL3 RKO cells. The scale bar is 50 μm. (G–I) Expression of METTL3 (G), RRM2B (H) or OPA1 (I) in CRC tumor tissues and adjacent normal mucosa tissues from GENT database. (J and K) Correlation between METTL3 and RRM2B (J) or OPA1 (K) in CRC patients from Gene Expression Profiling Interactive Analysis (GEPIA) database. (L–N) The Kaplan–Meier survival curves of OS based on METTL3 (L), RRM2B (M) or OPA1 (N) in CRC cancer patients from The Cancer Genome Atlas (TCGA) database. Data are presented as mean ± SD from three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significant, using Student's t-test between two groups and using one-way ANOVA followed by Bonferroni test for multiple comparisons.

RKO sh-control, sh-METTL3 and RRM2B or OPA1 stable overexpression in sh-METTL3 cells was used to establish xenografts. Results showed that knock-down of METTL3 significantly inhibited the growth of RKO xenografts, while the overexpression of RRM2B or OPA1 can rescue this effect (Fig. 8C). At the end of the experiment, the tumor volumes (Fig. 8D) and weights (Fig. 8E) in the sh-METTL3 group were significantly lower than those measured in the sh-control group, and these effects were reversed by overexpression of RRM2B or OPA1. No significant change in body weight was observed among all the groups (Fig. S8A). Further, IHC showed that the expression of RRM2B, OPA1 and Ki67-positive staining, which recognizes a nuclear antigen expressed in proliferating cells, in the sh-METTL3 group was decreased in vivo (Fig. 8F). This suggested that the knock-down of METTL3 can suppress the mitochondrial fusion and tumor growth of CRC cells.

At this point, we investigated the potential connection between m⁶A methylation, mitochondria dynamics and CRC development using clinical data from databases. The data form GENT indicated an upregulation in the expression of METTL3 (Fig. 8G), RRM2B (Fig. 8H), OPA1 (Fig. 8I) and IGF2BP2 (Fig. S8B) in CRC tumor tissues compared with normal tissues. Furthermore, data from Gene Expression Profiling Interactive Analysis (GEPIA) demonstrated a positive correlation between METTL3 and RRM2B (Fig. 8J), as well as OPA1 (Fig. 8K) in CRC cancer patients. Consistently, the protein expression of METTL3 was positively correlated with the RRM2B (Fig. S8E) and OPA1 (Fig. S8F) in CRC cancer patients from the Clinical Proteomic Tumor Analysis Consortium (CPTAC) database. Moreover, the expression of IGF2BP2 exhibited a positive correlation with the RRM2B (Fig. S8C) and OPA1 (Fig. S8D) in CRC cancer patients. In addition, the expression of RRM2B (Fig. S8E) and OPA1 (Fig. S8F) was notably lower in Grade I CRC tissues compared with Grade II/III tissues. Further, CRC patients with increased expression of METTL3 (Fig. 8L), RRM2B (Fig. 8M), OPA1 (Fig. 8N) and IGF2BP2 (Fig. S8G) showed reduced overall survival (OS). Together, these data suggested that the m⁶A/mitochondria dynamics axis regulated CRC progression.

DISCUSSION

Dysregulated mitochondrial dynamics are critical for cancer progression such as metastasis, drug resistance and cancer stem cell survival [7,8]. Increasing evidence has confirmed that targeting mitochondrial dynamics and its regulatory factors is a potential therapeutic strategy for cancer therapy [10,43]. Our present study revealed that m⁶A methyltransferase accelerated mitochondria fusion of CRC cells via regulation of the RRM2B/GSH axis and induced the expression of OPA1 (Fig. 9). Mechanically, IGF2BP2 bound with the A1240 at 3′UTR of RRM2B and A2212 at exon 18 in OPA1 CDS increased their mRNA stability, respectively. General or specifically targeting m⁶A significantly suppressed mitochondria fusion of CRC cells. In vivo and clinical data confirmed the positive roles of m⁶A/mitochondrial dynamics in tumor growth and the progression of CRC cancer. Consistently, recent studies have indicated that m⁶A and METTL3 can promote CRC progression, such as by inducing growth and drug resistance, and inhibiting antitumor immunity [44–46]. Our results describe the potential roles of m⁶A in mitochondrial dynamics and also create the possibility of developing therapeutic strategies against CRC progression by targeting m⁶A/mitochondria pathways.

Figure 9.

Working model of m⁶A-facilitated mitochondrial fusion of CRC cells via induction of GSH and stabilization of OPA1. In this model, METTL3 methylates the m⁶A modification of RRM2B and OPA1 mRNAs, which in turn increases their mRNA stability by IGF2BP2, thereby increasing their protein expression, leading to promotion of GSH synthesis and mitochondrial fusion. Loss of METTL3 decreases the m⁶A levels of RRM2B and OPA1 mRNAs, which are specifically recognized and stabilized by IGF2BP2, resulting in the decreased protein expression, thereby inhibiting GSH synthesis and mitochondrial fusion. The METTL3 knock-down cell with a depolarized mitochondrial membrane potential shows lower cell viability and higher sensitivity to antitumor drugs.

Recent studies indicate that m⁶A is critical for mitochondrial biogenesis and physiological functions [47]. For example, our previous study revealed that m⁶A can positively regulate the glycolysis of cancer cells via regulation of PDK4 [48]. Herein, we found that METTL3 can positively regulate the mitochondrial fusion via an m⁶A enzyme activity-dependent manner. One recent study consistently showed that the deletion of m⁶A demethylase FTO depletion induces mitochondrial fission in gastric cancer cells via caveolin-1 [20]. Further, FTO reduced the mRNA m⁶A of Drp1 and impaired the Drp1-mediated mitochondrial fragmentation [49]. Our data and previous studies confirmed the promotional effects of m⁶A on mitochondria dynamics and oncogenic roles of m⁶A/mitochondria in CRC development.

We identified that RRM2B/GSH is involved in m⁶A-regulated mitochondria fusion of CRC cells. Metabolomics analysis indicated that knock-down of METTL3 suppressed GSH levels in CRC cells, while the addition of GSH can reverse sh-METTL3-suppressed mitochondria fusion. RRM2B is a ribonucleotide reductase that protects glutathione synthetase (GSS) from proteasome degradation, thus maintaining the GSH concentration to prevent damage from lipid peroxide [50]. It participates in the regulation and modification of proteins and is also considered a vital component in tumor progression [51,52]. Since RRM2B is significantly upregulated in CRC tissues and associated with poor prognosis [53], our data suggest that m⁶A/RRM2B-regulated GSH generation and mitochondria fusion are potential targets for CRC therapy.

We further identified that m⁶A methylation of A2212 at exon 18 in OPA1 increase its mRNA stability and is involved in m⁶A-regualted mitochondria fusion of CRC cells. As the key regulator for cristae junction formation and respiratory chain super complexes, the IMM regulator OPA1 emerges as an intriguing candidate for targeted mitochondrial cancer therapy [54,55]. It is frequently amplified across pan-cancer genomic data sets and is associated with a poor prognosis and a heightened resistance to chemotherapy [56,57]. Our data revealed that METTL3 positively regulated the expression of OPA1, while overexpression of OPA1 reversed sh-METTL3-suppressed mitochondria fusion of CRC cells. Previous studies indicated mitochondrial peptidases OMA1 and YME1L cleaved OPA1 to regulate mitochondria dynamics [40]. Our present study revealed another layer of regulation factors for OPA1 expression and mitochondria fusion.

Developing inhibitors/activators of m⁶A-related proteins has become a hot spot in the field of anticancer epigenetic drugs [41]. Mitochondrial dysfunction and metabolic alterations have long been proposed to play a critical role in the pathogenesis of various cancers [7,58]. We therefore believe that novel, specific, effective and promising methods targeting m⁶A and mitochondrial dynamics will be developed, leading to a new generation in both cancer diagnosis and targeted therapy. Further, mitochondrial RNAs that have been discovered cover methylation, pseudouridylation and other modifications, which are involved in the biogenesis, stability and function of all mtRNA species [59,60]. Thus, the role of RNA modification in mtRNAs in mitochondria dynamics makes plenty of sense and remains to be further investigation.

Overall, our study sheds light on a novel relationship between mitochondria dynamics and m⁶A methylation. Specifically, m⁶A accelerated the mitochondria fusion of CRC cells via activation of the RRM2B/GSH axis and upregulation of OPA1. Given the numerous genes involved in mitochondrial dynamics, it is plausible that m⁶A modification indirectly regulates CRC mitochondrial dynamics by affecting other genes. Our study suggested that m⁶A regulates the mitochondrial dynamics of CRC cells, which has expanded our understanding of such interplays that are essential for therapeutic application.

MATERIALS AND METHODS

Mitochondrial morphology

To assess mitochondrial morphology, cells were seeded into confocal dishes and cultured with Mito-Tracker Green probe (Beyotime, China) and Hoechst 33342 staining solution (Beyotime, China) for 30 min at 37°C. They were then washed using phosphate buffer saline (PBS) three times for analysis using an Olympus Microscope (Olympus, Japan). Mitochondrial length was measured using ImageJ-MiNA [21] and categorized as elongated (˃3 μm), intermediate (0.5–3 μm) or fragmented (˂0.5 μm) types. The number of mitochondria in each category was counted from three independent experiments (>1000 mitochondria were analysed) and presented as mean of type percentage (± SD).

EM

Following the indicated treatments, cells were dissociated, centrifuged and fixed in 4% glutaraldehyde (Solarbio, China) for 2 h at 4°C. They were then post-fixed in 1% osmium tetroxide for an additional hour at 4°C. The samples were dehydrated, embedded in epoxy resin and ultimately sectioned into 70-nm-thick slices for microscopic examination. These sections were post-stained using 5% uranyl acetate and examined under a transmission electron microscope. Images were randomly captured to measure the mitochondrial area and circularity using ImageJ software.

Luciferase reporter assay

To assess the impact of m⁶A on the transcription of the target gene, we cloned the promoter region of the gene (–1000 to +100) into pGL3-Basic-Vector (Promega, USA) for promoter activity measurement. Cells were co-transfected with pGL3-basic-promoter and TK-Rluc reporter in a six-well plate for 24 h and the Dual-Luciferase Reporter Gene Assay Kit (Beyotime, China) was used to measure transcriptional activity. Renilla Luciferase (R-Luc) served as an internal control to normalize firefly luciferase (F-Luc) activity.

The CDS or 3′UTR sequence was subcloned into the dual-luciferase vector pmiGLO (Promega, USA). Mutagenesis of m⁶A sites (A to G) was conducted using a site-directed mutagenesis kit (Thermo Fisher, USA). The F-Luc activity values were normalized to the R-Luc activity values to reflect expression efficiency. The translation outcome was determined by comparing the F-Luc/R-Luc signal to mRNA abundance, while translation efficiency was defined as the quotient of reporter protein production divided by mRNA abundance[35]. All experiments were performed three times with consistent results.

Statistical analyses

Data were presented as mean ± SD from a minimum of three independent experiments. Statistical analysis was performed using a two-tailed unpaired Student's t-test for comparing two groups, while one-way or two-way Analysis of Variance (ANOVA) followed by the Bonferroni test was used for multiple comparisons. All statistical tests were two-sided. Analysis was performed using SPSS 16.0 for Windows. A P-value of <0.05 was considered to be statistically significant. *P < 0.05, **P < 0.01, ***P < 0.001; ns, no significant.

FUNDING

This research was supported by the National Key Research and Development Program of China (2022YFC2601800), the National Natural Science Foundation of China (32161143017, 82173833, 82372743, 82173126, 82373893 and 82341053), the Guangdong Basic and Applied Basic Research Foundation (2023B1515040006), the Key-Area Research and Development Program of Guangdong Province (2023B1111020007), the Open Program of Shenzhen Bay Laboratory (SZBL202009051006), the Guangdong Provincial Key Laboratory of Construction Foundation (2023B1212060022) and the Shenzhen Bay Scholars Program.

AUTHOR CONTRIBUTIONS

H.S. Wang, Z.J. Chen, Z.G. Li, Q. Peng, X. Deng, and J.X. Li are responsible for conceptualization, methodology and project administration; J.W. Zhou, K. Zhong, H.S. Zhang, L.J. Tao, Y. Lin, G.Y. Xie, Y.H. Tan, Y. Wu, and Y.Q. Lu are responsible for data curation and investigation; J.W. Zhou and H.S. Wang wrote the original draft.

Conflict of interest statement. None declared.