METTL3-mediated m⁶A mRNA modification of FBXW7 suppresses lung adenocarcinoma

This study was approved by the Ethics Committee of the Fourth Military Medical University. All animal experiments were carried out in strict accordance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health.

The human LUAD cell lines HCC827 and PC9 have been fully described. All cell lines were obtained between 2016 and 2019 and validated by short tandem repeat analysis, tested for mycoplasma contamination within the last six months, and used at passage numbers of < 10. HCC827 and PC9 cells were cultured in RPMI 1640 (Corning, Inc., Corning, NY, USA).

All compounds were purchased from Selleck Chemicals (Houston, TX, USA), except for D-luciferin (Promega, Madison, WI, USA), cycloheximide (Sigma, St. Louis, MO, USA), and puromycin (Sigma), and were dissolved in dimethyl sulfoxide for cell culture.

LUAD tissues and surrounding tissues were purchased from Shanghai Outdo Biotech (Shanghai, China) and collected from Tangdu Hospital (Xian, China). For IHC staining, the sections were deparaffinized in xylene, rehydrated using graded ethanol, and incubated with 0.1% sodium citrate buffer (pH 6.0) for 20 min at 95 °C for antigen retrieval. After endogenous peroxidase activity had been quenched with 3% H₂O₂·dH₂O and nonspecific binding had been blocked with 1% bovine serum albumin buffer, the sections were incubated overnight at 4 °C with anti-METTL3 (ab195352, 1:1000, Abcam, Cambridge, UK) or anti-FBW7α (A301–720, 1:1000, Bethyl Laboratories, Montgomery, TX, USA) antibodies. Following several washes, the sections were treated with horseradish peroxidase-conjugated secondary antibodies for 30 min at room temperature and stained with 0.05% 3, 3- diaminobenzidine tetrahydrochloride. The slides were then photographed using a virtual slide microscope (Olympus VS120, Tokyo, Japan), and the images were analyzed with Image-Pro Plus 6.0 software (Media Cybernetics, Silver Spring, MD, USA). IHC sections were analyzed by two independent investigators.

FBXW7 and METTL3 constructs were prepared using the pcDNA 3.1 vector (Invitrogen, Carlsbad, CA, USA).

Total RNA was isolated using RNeasy reagents (Qiagen, Hilden, Germany), and real-time PCR (qRT-PCR) was performed using an iCycler Real-Time System (Bio-Rad Laboratories, Hercules, CA, USA) with a SYBR Premix EX Tag Mastermix kit (TaKaRa, Shiga, Japan) according to the manufacturer’s instructions. Glyceraldehyde-3-phosphate dehydrogenase was used as an internal reference for gene expression. The relative mRNA expression of target genes was expressed using the 2^-ΔΔCt method. Experiments were repeated three times [30]. The same method was used for mRNA detection in subsequent cell experiments. The primers used are listed in Additional file 2: Table S1.

Lentiviral vectors containing METTL3 shRNA were purchased from Genchem (Shanghai, China). Cells were seeded in a 6-cm dish at a density of 2.5 × 10⁶ cells, allowed to attach for 12 h, and transfected with overexpression plasmid (2 μg) or siRNA (1.5 μg) targeting each gene using Lipofectamine® 2000 reagent (Invitrogen) according to the manufacturer’s instructions. The knockdown siRNA sequences are listed in Additional file 2: Table S2.

Cells were washed with ice-cold phosphate-buffered saline, scraped from the culture dishes, centrifuged at 12,000 rpm at 4 °C for 15 min, and then lysed in radioimmunoprecipitation assay lysis buffer (Thermo Fisher Scientific, Waltham, MA, USA) before being centrifuged at 15,000×g for 20 min at 4 °C. The protein concentration was determined using a BCA protein assay kit (Thermo Fisher Scientific). Proteins were then separated by SDS-PAGE, transferred to a polyvinylidene fluoride membrane (Millipore, Billerica, MA, USA), and blocked with 5% non-fat milk. Next, the membranes were incubated with antibodies or conjugated to horseradish peroxidase for 2 h at room temperature as a control. Proteins were visualized using electrochemiluminescence reagents (GE Healthcare Life Sciences, Little Chalfont, UK) and detected using a LAS-4000 imaging system (Fujifilm, Tokyo, Japan). Image density was quantified using ImageJ analysis software (NIH, Bethesda, MD, USA). The antibodies used are listed in Additional file 2: Table S3.

After gene manipulation, 20 μg/mL of cycloheximide (CHX; Sigma, final concentration 100 μg/mL) was added to the cell medium, and the cells were harvested at the indicated time points, lysed, and FBXW7 expression was measured by western blot analysis.

RIP assays were performed using a Magna RIPTM RNA-Binding Protein Immunoprecipitation Kit (Millipore) according to the manufacturer’s protocol. Briefly, the cells were collected and lysed in a complete radioimmunoprecipitation assay buffer containing a protease inhibitor cocktail and RNase inhibitor. Antibodies (5 μg) were pre-bound to Protein A/G magnetic beads in immunoprecipitation buffer (20 mM Tris-HCl pH 7.5, 140 mM NaCl, 0.05% TritonX-100) for 2 h and then incubated with 100 μL of cell lysate overnight at 4 °C with rotation. RNA was eluted from the beads by incubation with 400 μL of elution buffer for 2 h, precipitated with ethanol, and dissolved in RNase-free water. The enrichment of certain fragments was determined by real-time PCR.

MeRIP assays were performed using a Magna MeRIP m⁶A Kit (Millipore) according to the manufacturer’s instructions. Briefly, RNAs were chemically fragmented to ~ 100 nucleotides and incubated with magnetic beads conjugated to m⁶A antibodies (Abcam) for immunoprecipitation. The enrichment of m⁶A-containing mRNA was analyzed by qRT-PCR and normalized to the input.

Cells were seeded into each well of a 24-well plate and co-transfected with vectors according to the Lipofectamine®2000 (Invitrogen) protocol. After 48 h, luciferase activity was measured using a Dual-Glo Luciferase Assay system (Promega) according to the manufacturer’s instructions. Relative luciferase activity was assessed using a SYNERGY microplate reader (BioTek, Winooski, VT, USA). Firefly luciferase (F-luc) activity was normalized using Renilla luciferase to evaluate reporter translation efficiency.

Cells were incubated with Annexin V and propidium iodide (PI) according to the manufacturer’s instructions to analyze apoptosis. After 30 min, the cells were analyzed by fluorescence-activated cell sorting using a flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA). Annexin V-FITC-staining indicated early apoptosis, whereas cells with double-positive Annexin V-FITC and PI signals were pooled for analysis.

Cell proliferation was assessed using a Cell Counting Kit-8 (CCK-8; Dojindo Laboratories, Kumamoto, Japan) according to the manufacturer’s instructions. Briefly, the cells were seeded in 96-well plates (5 × 10³ cells/well), and then the medium was discarded. Next, a diluted CCK-8 solution (10% of total volume) was added at the indicated time points and incubated for a further 1.5 h. Absorbance was measured at 570 nm with a spectrophotometer (Tecan, Männedorf, Switzerland). Experiments were conducted in triplicate.

A total of 20 BALB/c nude mice (4–6 weeks old; weighing 17–20 g) were purchased from Cavens Laboratory Animal Co., Ltd. (Changzhou, China) and housed in a specific pathogen-free environment. The mice were randomly divided into groups before subcutaneous injection with cells (10⁷) suspended in 200 μL of phosphate-buffered saline. Tumors were measured on day 7 and then once every 7 days until 28 days after injection. Tumor volume was calculated using the following formula: V (mm³) = 0.5 × length (mm) × width² (mm²). At 4 weeks post-injection, the mice were sacrificed, and their tumors were isolated and weighed.

Substantial evidence has confirmed that FBXW7 plays a negative role in the pathogenesis of lung cancers. To explore the function of m⁶A methylation in the epigenetic regulation of FBXW7, we examined whether FBXW7 mRNA was m⁶A-methylated. MeRIP-qPCR indicated that FBXW7 mRNA contained abundant m⁶A modifications in HCC827 and PC9 cells (Fig. 1a, Additional file 1: Fig. S1A). To confirm the catalytic proteins that may be involved in m⁶A modification of FBXW7 mRNA, we compared and analyzed the correlations between FBXW7 and demethylases or methyltransferases from TCGA (https://​portal.​gdc.​cancer.​gov/​) [31]. Interestingly, FXBW7 expression exhibited a positive correlation with all m⁶A modification system-related proteins except for METTL16 in LUAD (Fig. 1b); therefore, we predicted that the loss of FBXW7 was related to deregulation of m⁶A-related enzymes in LUAD, particularly the aberrant expression of METTL3, which is a key “writer” that installs m⁶A on RNAs.

Simultaneously, we constructed METTL3 overexpression or knockdown systems using lentiviruses to study the function of METTL3 in LUAD (Additional file 1: Fig. S1B–D) and its effect on FBXW7 protein expression. We found that METTL3 overexpression in HCC827 and PC9 cells upregulated FBXW7 protein levels (Fig. 1c upper panels, Additional file 1: Fig. S1D, left panels), whereas METTL3 depletion significantly reduced FBXW7 protein expression (Fig. 1c, bottom panels, Additional file 1: Fig. S1D, right panels).

We also analyzed FBXW7 and METTL3 protein expression in twenty-eight paired LUAD and normal tissues, with IHC staining revealing that METTL3 and FBXW7 protein levels were higher in normal adjacent lung tissues than in LUAD tissues. Consistently, FBXW7 was expressed at lower levels in LUAD samples with low METTL3 expression (Fig. 1d). After standardization, the relationship between FBXW7 and METTL3 was plotted as a scatter diagram (Fig. 1e), which revealed that decreased FBXW7 protein expression was accompanied by decreased METTL3 expression and indicated a strong correlation between FBW7 and the m⁶A “writer.” Together, these results demonstrate that METTL3 mediates m⁶A modification in FBXW7 mRNA.

Recent studies have suggested that the m⁶A modification of many proteins plays a critical role in the development of various cancers; however, the role of METTL3 in LUAD remained unclear. We analyzed METTL3 protein expression in eight paired LUAD and normal tissues and found that METTL3 was remarkably downregulated in LUAD tissues (Fig. 2a). To identify direct targets of METTL3, we investigated changes in protein expression following METTL3 overexpression or knockdown in HCC827 and PC9 cells. We found that METTL3 overexpression increased the protein levels of apoptosis molecules (cleaved caspase 3 and Bax) and decreased the protein levels of Mcl-1 and c-Myc, which are classical FBXW7 target genes. In contrast, METTL3 depletion in HCC827 and PC9 cells had the opposite effects (Fig. 2b, Additional file 1: Fig. S2A).

To investigate the contribution of m⁶A toward LUAD development and progression, we focused on apoptosis and proliferation, which play key roles in the progression of various cancers. Although METTL3 depletion reduced apoptosis and promoted HCC827 and PC9 cell proliferation (Fig. 2c–d, Additional file 1: Fig. S2B–C, upper panels), its overexpression strongly enhanced apoptosis and impeded proliferation compared to in control cells (Fig. 2c–d, Additional file 1: Fig. S2B-C, bottom panels). In addition, we evaluated the clinical relevance of METTL3 protein levels in LUAD using IHC staining in eight paired patient samples. METTL3 expression was weak or undetectable in most LUAD samples but was moderate or high in most para-tumor controls (Fig. 2e), suggesting that METTL3 reduces LUAD tumorigenesis.

To verify whether the impaired tumorigenesis phenotype in METTL3-depleted cells depended on FBXW7 reduction, we overexpressed FBXW7 in both HCC827 and PC9 subclones with METTL3 knockdown. Western blotting revealed that FBXW7 overexpression abrogated the increased expression of its downstream targets (Mcl-1 and c-Myc) and decreased the expression of apoptosis molecules (cleaved caspase 3 and Bax) in cells with METTL3 knockdown (Fig. 3a, Additional file 1: Fig. S3A).

To validate whether FBXW7 contributes to the anti-tumor phenotype regulated by METTL3, we generated stable cells expressing the indicated genes (Fig. 3b, Additional file 1: Fig. S3B) to determine whether FBXW7 overexpression could rescue the effects of METTL3 knockdown on the biological behavior of HCC827 or PC9 cells. As shown in Fig. 3c–d, FBXW7 overexpression partially reversed the effects of METTL3 depletion on the apoptosis and proliferation of HCC827 cells and reversed the phenotype caused by METTL3 deficiency in PC9 cells (Additional file 1: Fig. S3C–D). Collectively, these results indicate that FBXW7 overexpression rescues the impaired anti-tumor phenotype caused by METTL3 knockdown.

In addition, quantitative RIP assays suggested that FBXW7 mRNA interacts with METTL3 in HCC827 and PC9 cells (Fig. 3e, Additional file 1: Fig. S3E). To verify that m⁶A-modified FBXW7 is regulated by METTL3, we performed MeRIP-qPCR experiments in HCC827 and PC9 cells with METTL3 overexpression or knockdown. METTL3 overexpression significantly promoted m⁶A modification in FBXW7 mRNA. However, METTL3 depletion had the opposite effect in HCC827 and PC9 cells (Fig. 3f, Additional file 1: Fig. S3F).

Next, we investigated the mechanisms via which m⁶A modification affects FBXW7 expression in LUAD. First, evaluation of FBXW7 mRNA levels revealed no major differences in shMETTL3 or overexpressing cells (Fig. 4a). To verify whether METTL3 affected the subcellular FBXW7 mRNA distribution, we isolated and analyzed cytoplasmic and nuclear RNAs (Additional file 1: Fig. S4A); however, no notable discrepancy was observed in the subcellular localization of FBXW7 mRNA between METTL3 knockdown and control HCC827 cells (Fig. 4b). Similarly, no dramatic differences in the half-life of FBXW7 mRNA were detected in METTL3 knockdown and control HCC827 cells, as measured using actinomycin D (Fig. 4c). Consequently, protein stability or translation efficiency may be responsible for the reduced FBXW7 protein expression in shMETTL3 HCC827 cells.

To test this assumption and validate whether METTL3 is engaged in translational control, we explored whether METTL3 was localized in the cytoplasmic fractions where the translation is initiated and occurs. Apart from nuclear localization where m⁶A modification occurs, METTL3 was expressed in the cell cytoplasm; however, other known METTL3-interacting proteins, including the m⁶A methyltransferase METTL14 and cofactor WTAP, displayed pronounced nuclear localization (Fig. 4d). To assess the potential effects of METTL3 on FBXW7 protein stability, we blocked translation using CHX and measured FBXW7 degradation in METTL3 knockdown and control HCC827 cells. The western blot analysis results revealed that the half-life of FBXW7 proteins was similar in METTL3 knockdown and control HCC827 cells (Fig. 4e), excluding the possibility that m⁶A -induced FBXW7 expression was associated with protein stability. HCC827 cells were also pretreated with the proteasome inhibitor MG132 or translation inhibitor CHX and then transfected with METTL3 overexpression plasmids. Interestingly, impairing METTL3 induced FBXW7 expression in HCC827 cells in the presence of CHX but not MG-132 (Fig. 4f). The PmirGLO-FBXW7-CDS luciferase reporter was generated by ligating the FBXW7 CDS to a multiple cloning site (Fig. 4g upper panel), and the subsequent assay indicated that the translational efficiency of FBXW7 was significantly lower in shMETTL3 cells than in the control group (Fig. 4g, bottom panel). Together, these results suggest that m⁶A mRNA modification regulates FBXW7 translation.

Next, we attempted to elucidate the molecular mechanisms via which METTL3 promotes the translational efficiency of FBXW7. Fragmented RNA isolated from HCC827 cells was immunoprecipitated using m⁶A antibodies to investigate the distribution of m⁶A methylation in FBXW7 mRNA. The highest levels of m⁶A methylation were observed in the CDS, followed by the 3′ untranslated region (UTR) and 5′UTR (Fig. 5a). Accordingly, decreased m⁶A enrichment was observed in the FBXW7 CDS in shMETTL3 cells, indicating that m⁶A modification is more dynamic in the CDS than in the 3′UTR region. Moreover, we predicted possible m⁶A modification sites in the FBXW7 sequence using the sequence-based RNA adenosine methylation site predictor (SRAMP) [32], which valued each identified m⁶A site by assigning a predictive score (Additional file 1: Fig. S5, Additional file 2: Table S4). The FBXW7 genome contained 26 potential m⁶A sites predicted by SRAMPS, of which four had very high confidence at positions 2066, 2111, 2264, and 3190. To verify these predicted m⁶A sites, we designed specific primers to amplify 15 sites, which were merged into one region of less than 200 base pairs (bp), and MeRIP-qPCR was performed. After normalizing m⁶A-precipitated signals to the input, only three m⁶A sites (sites 8#, 10#, and 14#) showed elevated levels, indicating that the m⁶A sequence exists in these sites (Fig. 5b). Subsequently, we identified sites #11, #16, and #23 in cells transfected with the indicated mutant plasmids (RRACH → RRCCH) using MeRIP (Additional file 1: Fig. S4B).

To explore the potential roles of m⁶A methylation in the FBXW7 3′UTR region, which demonstrated m⁶A methylation at site #23, we performed a luciferase assay in HCC827 cells using reporters containing FBXW7–3′UTR-WT or -MUT (site #23: AGAC to AGCC.). Compared to FBXW7–3′UTR-WT, the translational activity of FBXW7–3′UTR-MUT was similar between the control and shMETTL3 group, and METTL3 overexpression did not alter the luciferase activity of FBXW7–3′UTR-WT or -MUT in HCC827 cells (Fig. 5c). Together, these results indicate that the m⁶A modification, which promotes FBXW7 translation, does not correlate with m⁶A methylation levels at sites in the 3′UTR.

Therefore, we examined whether m⁶A methylation in the CDS promotes the translation of FBXW7. First, we constructed an FBXW7 CDS expression plasmid and mutated the m⁶A motif in the coding region of FBXW7, as follows: FBXW7-CDS-MUTs*: FBXW7-CDS-MUT1* and FBXW7-CDS-MUT2* (containing only one predicted m⁶A site that was mutated to create FBXW7-CDS-MUT1/2); FBXW7-CDS-MUTs series: FBXW7-CDS-MUT1, FBXW7-CDS-MUT2 (containing three putative m⁶A sites of which only one was mutated to create FBXW7-CDS-MUT1/2), and FBXW7-CDS-MUT3 (containing two potential m⁶A sites, which were both mutated to form FBXW7-CDS-MUT3; Fig. 5d). FBXW7 protein expression was lower in cells co-transfected with METTL3 and FBXW7-CDS-MUTs than in the METTL3 and FBXW7-CDS-WT group (Fig. 5e). We also performed MeRIP-qPCR to examine the level of m⁶A modification, which showed that the FBXW7-CDS-WT had high levels of m⁶A modification, whereas FBXW7 MUTs (MUT1, MUT2, and MUT3) contained little or no m⁶A in HCC827 cells with endogenous METTL3 expression (Fig. 5f). METTL3 overexpression significantly increased the abundance of m⁶A in FBXW7-CDS-WT/MUTs compared to the control vector of METTL3; however, this effect was inhibited in FBXW7 MUTs compared to in the FBXW7-CDS-WT because of the mutated m⁶A site. Remarkably, the m 6 A modification of FBXW7 MUT3, containing two m 6 A site mutation, was not elevated in METTL3 overexpression and control HCC827 cells (Fig. 5g). Compared to the control group, m⁶A modification was not enhanced in FBXW7-MUT1/2 in METTL3-overexpressing cells. Notably, the levels of transfected FBXW7 MUTs* were unchanged in METTL3-overexpressing compared to the control METTL3 vector due to only having one mutated m⁶A site (Fig. 5h), suggesting that the putative sites in FBXW7 mRNA are m⁶A methylated by METTL3. Furthermore, FBXW7-CDS-WT and FBXW7 MUT mRNAs interacted with METTL3, revealing that the mutated sites did not interrupt direct interaction between METTL3 and FBXW7 MUT mRNAs (Fig. 5i). Therefore, these results exclude the possibility that mutation interfered with the interaction between FBXW7 MUTS and METTL3. Together, our data indicate that METTL3 site-specific m⁶A modifications in the FBXW7 CDS promote FBXW7 translation.

To validate the effect of METTL3 on tumor growth in vivo, we conducted a xenograft tumor assay by subcutaneously injecting nude mice with HCC827 Ctrl, shMETTL3, OE-FBXW7, and shMETTL3/OE-FBXW7 cells. Significantly more tumors were derived from shMETTL3 HCC827 cells than from HCC827 Ctrl cells; however, FBXW7 overexpression attenuated the tumor-promoting effect of shMETTL3 HCC827 cells in vivo (Fig. 6b). We found that the shMETTL3/OE-FBXW7 groups displayed smaller tumors (Fig. 6c) and slower tumor growth (Fig. 6d) than the shMETTL3 group (p < 0.01), whereas peritumoral tissues displayed high FBXW7 levels and a remarkable reduction in tumor size. Additionally, METTL3 exhibited a similar expression pattern to FBXW7 (Fig. 6a), indicating a positive correlation between FBXW7 and METTL3 at the protein level.

Low METTL3 expression in LUAD was validated by Gene Expression Profiling Interactive Analysis (GEPIA, http://​gepia.​cancer-pku.​cn/​index.​html) [33] (Fig. 6e) and the prognostic correlation between METTL3 mRNA expression and the survival of patients with LUAD was evaluated using publicly available datasets. Kaplan-Meier curves revealed that patients with LUAD with higher METTL3 expression had better overall survival (n = 1927, p = 4.2 × 10^− 13), post-progression survival (n = 344, p = 0.004), and progression-free survival (n = 982, p = 0.006; Fig. 6f, upper panels). Interestingly, this effect was markedly more significant in patients with LUAD than in those with lung squamous cell carcinoma (LUSC) (Fig. 6f, lower panels), consistent with the overall survival analysis of FBXW7 (Fig. 6g). Taken together, these data suggest that the METTL3-FBXW7 axis contributes to inhibition of tumor growth in vivo and that targeting m⁶A-modified FBXW7 is a promising strategy for overcoming tumorigenesis in LUAD.