The N6-methyladenosine METTL3 regulates tumorigenesis and glycolysis by mediating m6A methylation of the tumor suppressor LATS1 in breast cancer

Background Posttranscriptional modification of tumor-associated factors plays a pivotal role in breast cancer progression. However, the underlying mechanism remains unknown. M6A modifications in cancer cells are dynamic and reversible and have been found to impact tumor initiation and progression through various mechanisms. In this study, we explored the regulatory mechanism of breast cancer cell proliferation and metabolism through m6A methylation in the Hippo pathway. Methods A combination of MeRIP-seq, RNA-seq and metabolomics-seq was utilized to reveal a map of m6A modifications in breast cancer tissues and cells. We conducted RNA pull-down assays, RIP-qPCR, MeRIP-qPCR, and RNA stability analysis to identify the relationship between m6A proteins and LATS1 in m6A regulation in breast cancer cells. The expression and biological functions of m6A proteins were confirmed in breast cancer cells in vitro and in vivo. Furthermore, we investigated the phosphorylation levels and localization of YAP/TAZ to reveal that the activity of the Hippo pathway was affected by m6A regulation of LATS1 in breast cancer cells. Results We demonstrated that m6A regulation plays an important role in proliferation and glycolytic metabolism in breast cancer through the Hippo pathway factor, LATS1. METTL3 was identified as the m6A writer, with YTHDF2 as the reader protein of LATS1 mRNA, which plays a positive role in promoting both tumorigenesis and glycolysis in breast cancer. High levels of m6A modification were induced by METTL3 in LATS1 mRNA. YTHDF2 identified m6A sites in LATS1 mRNA and reduced its stability. Knockout of the protein expression of METTL3 or YTHDF2 increased the expression of LATS1 mRNA and suppressed breast cancer tumorigenesis by activating YAP/TAZ in the Hippo pathway. Conclusions In summary, we discovered that the METTL3-LATS1-YTHDF2 pathway plays an important role in the progression of breast cancer by activating YAP/TAZ in the Hippo pathway. Supplementary Information The online version contains supplementary material available at 10.1186/s13046-022-02581-1.


Background
The latest global burden of cancer released by the International Agency for Research on Cancer (IARC) in 2020 declared breast cancer to be the world's leading cancer [1]. Owing to the high heterogeneity of breast cancer, there are considerable differences in the diagnosis, treatment, and prognosis of different breast cancer types. It is widely recognized that the axillary lymph node metastasis rate in patients with luminal B breast cancer is higher than that in patients with luminal A disease, whereas the axillary lymph node metastasis rate in patients with triple-negative breast cancer (TNBC) is lower. Patients with the luminal A type generally have a better survival rate than those with other types [2]. Endocrine therapy is usually the preferred choice of treatment for patients with luminal A and luminal B tumors. Patients with HER-2 overexpression are sensitive to targeted therapy, while patients with TNBC experience rapid clinical progression [3]. The heterogeneity of breast cancer is also internally reflected in differences in the genome, transcriptome, proteome, and metabolome [4].
Nonmutational epigenetic reprogramming is one of the 14 features of cancer listed in the third edition of Hallmarks of Cancer [5]. As the most abundant internal modification of mammalian mRNA, N6-methyladenosine (m6A) modification is involved in multiple aspects of RNA metabolism, including RNA stability, translation, splicing, transport, and localization, which have been discovered to have an impact on tumor initiation and progression through various mechanisms [6]. The m6A protein affects tumor proliferation and metastasis in breast cancer and is closely related to the prognosis of patients [7][8][9]. However, other potential mechanisms remain unclear. The fate of m6A modification of mRNA varies according to different reader proteins. Therefore, it is necessary to conduct systematic sequencing and bioinformatics analysis of the levels of m6A modification in key pathways to examine the development of breast cancer.
The Hippo pathway plays an important role in organ development and serves as a tumorigenesis suppressor. Following stimulation with extracellular growth inhibition signals, a series of kinase cascades, such as LATS1/ LATS2, are activated, resulting in phosphorylation of the effector factor Yes-associated protein (YAP) and transcriptional coactivator with PDZ-binding motif (TAZ). Notably, recent findings have implicated YAP and TAZ in the metabolism of cancer cells [10]. The Hippo pathway controls cell proliferation and organ size by binding to cytoskeletal proteins and remaining in the intracytoplasmic space as well as by reducing nuclear activity. LATS1 has also been reported to be a tumor suppressor [11]. We previously found that HERC4 is an E3 ligase for the tumor suppressor LATS1 and destabilizes LATS1 by promoting its ubiquitination of LATS1 [12].
Whether m6A modification occurs in the Hippo pathway remains unclear. Recent studies have shown that METTL3 is highly expressed in breast cancer [9] and can be regulated by multiple noncoding RNAs such as miR-483-3p [13], let-7 g [7] and pri-miR-221-3p [8]. Moreover, m6A-mediated lincRNAs, such as LINC00958 [14], were found to regulate breast cancer tumorigenesis. Here, we demonstrated that LATS1 mRNA m6A modification mediated by METTL3 and recognized by YTHDF2, plays a positive role in promoting both tumorigenesis and glycolysis in breast cancer.

Breast cancer patient tumor samples
After obtaining adequate informed consent, breast cancer tissue and adjacent normal tissue were obtained from 8 patients who underwent curative resection for breast cancer at Nanfang Hospital of Southern Medical University, between November 2019 and May 2020. All patients satisfied the following inclusion criteria: the surgical margins were confirmed to contain no residual carcinoma tissue; clinicopathological information on age, sex, clinical stage, neoadjuvant therapy, ER, PR, HER2, Ki67, histological subtype and histological grade was available and is listed in Supplementary Table 1. All patient-related studies were approved by the institutional review boards of the Seventh Affiliated Hospital of Southern Medical University and Nanfang Hospital of Southern Medical University.

Western blot analysis
Lysis buffer was used for total protein analysis of extracted cells. Samples were separated on an 8-15% gel by SDS-PAGE. Nitrocellulose membranes were blocked with blocking buffer and incubated with the appropriate primary antibody. The membranes were washed with blocking buffer three times, probed with the appropriate secondary antibody and developed using SuperSignal West Pico or Dura (Thermo Fisher Scientific).

Quantitative PCR analysis
Real-time PCR analysis was performed by using the Bio-

Cell proliferation, apoptosis, migration and invasion assays
For the EdU cell proliferation assay, the cells were trypsinized and seeded onto 6-well plates at a density of 1 × 10 6 cells per well. After incubation for 24 h at 37 °C, 1 ml medium was added to each well and incubated for 2 h after discarding the old medium. The cells were collected in flow tubes and centrifuged at 1,500 rpm for 5 min. The supernatant was discarded and the cells were resuspended in PBS and centrifuged at 1,500 rpm for 5 min. After discarding the supernatant, the cells were fixed with 4% paraformaldehyde for 15-30 min, neutralized with glycine for 5 min, and resuspended in PBS. The cells were then incubated with 0.5% Triton X-100 penetrant at room temperature for 10 min. One hundred microliters of 1⨯ Apollo staining reaction solution was added to each tube and the cells were resuspended again. After incubation for 10 min at room temperature, the staining reaction solution was discarded by centrifugation at 1,500 rpm for 5 min. Next, 0.5% Triton X-100 penetrant was used to clean the cells 1-3 times at room temperature and the cells were resuspended in PBS. Flow cytometry was performed immediately after staining.

Human cancer cell xenograft model
All animal experiments were approved by the Institutional Animal Care and Use Committee of Southern Medical University. Breast cancer cells (5 × 10 6 ) were implanted into the subcutaneous axilla of the forelimb of 3-4-week-old BALB/c nude mice. Seven days after transplantation, the diameter of the tumors was measured, and the tumors were removed after three weeks.

Coimmunoprecipitation assay
Immunoprecipitation assays were performed as previously described [16]. Cells were lysed in radioimmunoprecipitation assay buffer containing protease and phosphatase inhibitors. The cells were then centrifuged at 12,000 × g for 10 min at 4 °C and collected. The supernatants were immunoprecipitated with antibodies and magnetic protein A/G beads (Pierce) were used for incubation for 2 h at 4 °C. The above immune complexes were washed with PBS and resuspended in SDS-PAGE buffer, followed by western blotting analysis.

MeRIP-seq
Total RNA was isolated from cells and tissues using Magzol Reagent (Magen, China), according to the manufacturer's protocol. The quantity and integrity of the RNA yield were assessed using a K5500 microspectrophotometer (Beijing Kaiao, China) and the Agilent 2200 TapeStation system (Agilent Technologies, USA), respectively. m6A antibody-immunoprecipitated RNA was quality controlled using the Qubit (Thermo Fisher Scientific, USA) and Agilent 2200 TapeStation (Agilent Technologies, USA) systems. Briefly, RNA was fragmented into molecules approximately 200 bp in length. The RNA fragments were then subjected to first-strand and secondstrand cDNA synthesis, followed by adaptor ligation and enrichment with a low cycle according to the instructions of the NEBNext ® Ultra RNA LibraryPrep Kit for Illumina (NEB, USA). The final library product was assessed using the Agilent 2200 TapeStation and Qubit ® system (Life Technologies, USA), and then sequenced on an Illumina platform (Illumina, USA) with a paired-end length of 150 bp at RiboBio Co., Ltd. (Ribobio, China). Adaptor and low-quality bases were trimmed using Trimmomatic tools (version: 0.36), and the clean reads were subjected to rRNA deletion through RNAcentral to obtain effective reads. Genomic alignment was performed using TopHat (version 2.0.13) to obtain uniquely mapped reads. Effective reads from the input sample were used for RNAseq analysis and the read count value of each transcript was calculated using HTSeq (version 0.6.0). Differentially expressed genes were identified using the DEseq/ DESeq2/edgeR/DEGseq R package according to the following criteria: |log2 (fold change)|≥ 1 and P value < 0.05.

Sequence-based RNA adenosine methylation site predictor (SRAMP)
SRAMP is a site prediction tool based on a random forest machine-learning framework that can predict m6A sites based on sequence-derived features. The m6A locus was predicted using the target sequence [17].

Metabolomic analysis
A high-resolution mass spectrometer (QEXactive, Thermo Fisher Scientific, USA) was used to collect positive ion (POS) and negative ion (NEG) data from 18 cell samples for untargeted metabolomics detection using LC-MS/MS technology to explore the metabolomic composition and biological function of the samples. Compound Discoverer 3.1.0 (Thermo Fisher Scientific, USA) software was used for data processing, including peak extraction, peak alignment, fill gaps, and compound identification. An in-house metabolome information analysis process was used to carry out metabolite annotation, classification (Kyoto Encyclopedia of Genes and Genomes [KEGG], Human Metabolome Database [HMDB]), and enrichment analysis for the identified substances, and to explain the physical and chemical properties and biological functions of the metabolites. The R software package metaX [18] was used for data preprocessing, statistical analysis (univariate and multivariate analyses), and screening of metabolites with significant differences. Discriminant analysis was performed using partial least squares discriminant analysis (PLS-DA). Differentially expressed metabolites were screened by the VIP values of the first two principal components of the model [19] and Student's t-test was used to analyze the results of the univariate analysis (fold change).

Transcriptome sequencing
Total RNA was extracted using TRIzol reagent (Thermo Fisher, 15,596,018) as previously reported [12]. Total RNA quantity and purity were analyzed using a Bioanalyzer 2100 and RNA 6000 Nano LabChip Kit (Agilent, CA, USA, 5067-1511). Additionally, mRNA was purified from total RNA (5 μg) using Dynabeads Oligo (dT) (Thermo Fisher Scientific, CA, USA) with two rounds of purification and fragmented into short fragments using divalent cations at an elevated temperature (Magnesium RNA Fragmentation Module (NEB, cat. e6150, USA) at 94 °C for 5-7 min). Then, the cleaved RNA fragments were reverse-transcribed to create cDNA with SuperScript ™ II Reverse Transcriptase (Invitrogen, cat.1896649, USA). Dual-index adapters were used and size selection was performed by using AMPureXP beads. The U-labeled second-stranded DNAs were treated with the heat-labile UDG enzyme (NEB, cat.m0280, USA). PCR was used to amplify to the ligated products under the following conditions: 95 °C for 3 min; 8 cycles of denaturation at 98 °C for 15 s, 60 °C for 15 s, and 72 °C for 30 s; and 72 °C for 5 min. Finally, 2 × 150 bp pairedend sequencing (PE150) was performed on an Illumina Novaseq 6000 (LC-Bio Technology Co., Ltd., Hangzhou, China).

RNA stability analysis
Cells in the treatment and control groups were collected after 24 h. RNA was extracted using TRIzol reagent, and mRNA expression was determined by qPCR.

Seahorse assay
On the same day that the cells were seeded, 180 μl of hydration solution was added to the lower part of the XF96 Extracellular Flux Assay Kit. Cells were hydrated overnight in an incubator without CO 2 at 37 °C. The required drugs and Seahorse XF basic culture medium were prepared simultaneously, and the pH of the medium was adjusted to 7.4. The cells were then placed in a 37 °C water bath for 1 h before use. On the second day, the cells were washed twice with Seahorse XF basic culture medium in a water bath and 175 μl Seahorse XF basic culture medium was added to each well. The cells were then cultured in a CO 2 -free incubator at 37 °C for 1 h. Each drug was diluted to the required concentration, and 25 μl was added to each well of an XF96 extracellular flux assay kit. Thirty minutes later, the lower part of the XF96

RIP-qPCR
Endogenous RNA was retrieved after trapping in the nucleus or cytoplasm with an antibody or epitope marker. The RNA-binding protein was separated from the bound RNA by immunoprecipitation. The cells were crosslinked with 1% formaldehyde and treated with 150 mM NaCl RIPA buffer containing RNase and protease inhibitors. Cells were lysed in 0.5% sodium deoxycholate, 0.1% SDS, 1% NP40, 1 mM EDTA, and 50 mM Tris (pH 8.0) for 30 min and centrifuged for precipitation. The supernatant was incubated four times with METTL3, YTHDF1, or YTHDF2 primary antibody or the corresponding IgG antibody for 4 h, and protein A/G glycosylated beads were added and incubated for 2 h with shaking. After washing three times with RIPA buffer, RNA was extracted after crosslinking. Quantitative RT-PCR was performed to measure RNA.

RNA pulldown assays
RNA pulldown assays were performed in accordance with the Pierce ™ Magnetic RNA-Protein Pull-down Kit protocol. RNA was prepared by in vitro transcription and biotin was used. Poly (A) 25 RNA was used as the negative control. Total protein was extracted from each group, and the required protein concentration was greater than 2 mg/ml. RNA pull-down was performed according to a standard protocol, and the expression level was confirmed by western blotting.

Statistical analysis
Data were analyzed using SPSS 20.0. With a two-tailed independent Student's t-test, p < 0.05 was considered significant. Two patient cohorts were compared using a Kaplan--Meier survival plot, and log-rank p value were calculated.

Aberrant m6A methylation levels of Hippo pathway proteins were found in breast cancer
To investigate m6A modification levels in breast cancer tissue samples, MeRIP-seq was conducted for paired breast ductal cancer tissues and adjacent normal breast tissue samples. The corresponding sequencing data have been uploaded to the GEO database (GSE217977, www. ncbi. nlm. nih. gov/ geo/). By analyzing the sequencing data, we found that large amounts of differentially methylated or modified mRNAs existed in cancer tissue (Fig. 1a). The differential enrichment peaks of each pair revealed that more than half of the m6A-specific sites were enriched in tumor tissue and appeared near the classic mRNA-modified regions, namely exons, 3'UTRs, and stop codons ( Fig. 1b and Supplementary Fig. 1a). Combined analysis of differentially expressed m6A genes and mRNAs helped us discover the influence of m6A modification on mRNA expression, mRNA abundance, and differentially expressed genes. Figure 1c and Supplementary Fig. 1b show the results of the combined analysis of the differentially expressed m6A mRNAs and differentially expressed genes. Functional enrichment analysis of these differentially expressed genes (Fig. 1d, right panel and Supplementary Fig. 1c-d) further revealed that m6A modification plays an important role in regulating multiple cancers and pathways, such as the MAPK, Hippo, and PI3K-Akt signaling pathways.
Notably, genes in the Hippo pathway were found to be differentially modified and expressed in breast cancer tissue (Fig. 1d, left panel and Supplementary Fig. 1e). The Hippo pathway is a significant proliferation-regulating pathway involved in cancer development. The results showed that m6A modification of the tumor suppressor LATS1 may be involved in the downregulation of mRNA expression ( Supplementary Fig. 1f-g) and negative regulation of cellular metabolic processes (Fig. 1d, left panel). MeRIP-seq results revealed that the m6A modification level of LATS1 mRNA was significantly upregulated at more than two sites (Fig. 1e, upper panel, Supplementary Table 2 and Supplementary Fig. 1h). To determine the mechanism of m6A modification of LATS1 in breast cancer, we used SRAMP to predict the m6A sites in LATS1 mRNA (Fig. 1e, lower panel). m6A2Target (http:// m6a2t arget. cance romics. org/#/), a database for predicting the target genes of m6A writers, erasers, and readers, was used to search for proteins related to m6A modification of LATS1 mRNA. As shown in Fig. 1f, the m6A writer protein mettl3 was predicted to be the binding protein of LATS1 mRNA during m6A modification in MDA-MB-231 breast cancer cells.

Expression of METTL3 could be an independent factor affecting the prognosis of breast cancer patients and promoting tumorigenesis of breast cancer cells
To investigate the expression of METTL3 in breast cancer, we examined its protein levels in breast cancer tissues and paired adjacent normal tissues. As shown in Fig. 2a, Mettl3 was highly expressed in breast ductal carcinoma tissues. Bioinformatics analysis was applied to the expression data of breast tumor tissues and tumor-adjacent normal tissues in the Gene Expression Omnibus (GEO) database (GSE70951, 195 breast adenocarcinomas and matched adjacent normal breast tissue samples). We found that the expression levels of ER/PR/HER2 were not correlated with those of Mettl3 ( Supplementary Fig. 2a). However, the immunohistochemistry (IHC) results conducted in breast cancer tissue microarrays (129 breast cancer tissue samples and three normal breast tissue samples) revealed that Mettl3 was expressed mostly in the cytoplasm and nuclei of cancer tissues, with a few expression in the cell membrane (Fig. 2b). The relationship between Mettl3 and patient clinicopathological features is shown in Supplementary Table 3, and the results of univariate and multivariate analyses of the factors that correlated with the overall survival of cancer patients are shown in Table 1.
Kaplan-Meier survival analysis revealed that high METTL3 protein expression was significantly associated with poor prognosis in invasive ductal carcinoma and luminal breast cancer tissues ( Fig. 2c and Supplementary Fig. 2b). Moreover, we examined the protein level of METTL3 in six different breast cancer cell types. The results are shown in Fig. 2d. Mettl3 showed the highest expression in T47D cells, followed by MCF-7 cells, and the lowest expression in MDA-MB-231 cells. These data indicate that METTL3 promotes tumorigenesis of invasive ductal carcinoma of the breast. Therefore, we examined the role of METTL3 in breast tumorigenesis by knocking out or overexpressing METTL3 in MCF-7 and T47D cells ( Supplementary Fig. 2c-d). Tumor cell proliferation and survival were significantly suppressed after METTL3 deletion (Fig. 3a-b and Supplementary Fig. 3ab). Trans-well assays showed that METTL3 deletion inhibited the migration and invasion of MCF-7 and T47D cells (Fig. 3c-d and Supplementary Fig. 3c-d). Knockout of METTL3 in MCF-7 and T47D cells inhibited the growth of tumors formed by the corresponding cells in immunodeficient mice, whereas overexpression of METTL3 significantly promoted the growth of transplanted tumors ( Fig. 3e and Supplementary Fig. 3e). Based on these data, we confirmed the important role of METTL3 in breast cancer tumorigenesis.

METTL3 affects breast cancer cell metabolism by regulating m6A modification of LATS1 mRNA
RNA-seq was used to determine the influence of METTL3 on transcriptome expression in the MCF-7cells. METTL3 expression markedly increased 2845 transcripts and reduced the expression of 2044 transcripts (Fig. 4a). KEGG analysis revealed that the Hippo pathway and glycolysis were significantly associated with METTL3 expression (Fig. 4b and Supplementary Table 4). Additionally, pathway enrichment analysis of the differentially expressed genes indicated that METTL3 was significantly related to breast cancer and the Hippo pathway (Fig. 4c). Since RNA-seq revealed that METTL3 might be related to glycolysis in breast tumor cells, we conducted metabolomic-seq analysis in METTL3 KO MCF-7 breast cancer cells. Differential metabolite KEGG analysis further demonstrated that the expression of METTL3 is involved in  central carbon metabolism in cancer, especially in glycolysis and gluconeogenesis (Fig. 4d). Multiple metabolites involved in central carbon metabolism in cancer and glycolysis/gluconeogenesis were identified ( Supplementary  Fig. 4a-b and Supplementary Table 5). To understand the mechanism by which METTL3 promotes glycolysis in breast cancer, we evaluated the effect of METTL3 on glycolysis in MCF-7 and T47D cells. Notably, we found that deletion of METTL3 decreased glycolytic activity in breast cancer cell lines (Fig. 5a-b and Supplementary Fig. 5a-b). Since it is well known that the Hippo pathway is involved in cancer development, we predicted that METTL3 could mediate m6A modification of LATS1 and further affect the Hippo pathway. Using an RNA pull-down assay with western blotting, we confirmed the interaction between LATS1 mRNA and METTL3 (Fig. 5c). In addition, the protein levels of METTL3 were inversely correlated with the mRNA levels of LATS1 in breast cancer cells (Supplementary Fig. 5c). MeRIP-qPCR results indicated that the presence of METTL3 directly increased the m6A level of LATS1 mRNA (Fig. 5d and Supplementary Fig. 5d). To identify the region of LATS1 mRNA that interacted with METTL3, we examined the interaction between METTL3 and LATS1 mRNA. The mutation sites were based on the results of MeRIP-seq and SRAMP prediction. The results of MeRIP-qPCR further confirmed that the regulation of METTL3 mediates m6A methylation of LATS1 mRNA ( Fig. 5e and Supplementary Fig. 5e). To investigate the The cell number was determined with EdU cellular proliferation assay with flow detection, n = 3, *** p < 0.001. b An apoptosis assay was performed after knocking out or overexpressing METTL3 in MCF-7 cells, and the apoptotic cell number was counted with a fluorescein assay, n = 3, ** p < 0.01, *** p < 0.001. c Blocking METTL3 inhibited the migration of MCF-7 cells, as detected by wound healing tests at 0 h, 12 h, 24 h and 36 h. d The invasion ability of MCF-7 cells, as revealed by the trans-well assay, was significantly suppressed by knocking out METTL3, n = 3, * p < 0.05, ** p < 0.01. e The tumor growth of MCF-7 cells in nude mice was delayed by METTL3 knockout, n = 7, * p < 0.05, *** p < 0.001 impact of m6A regulation on proliferation and glycolysis in breast cancer cells, a colony formation assay was performed, which revealed that the proliferation of breast cancer cells could be rescued by the inhibition of LATS1 after METTL3 knockout ( Fig. 5f and Supplementary Fig. 5f ). Moreover, the seahorse assay revealed that glycolysis was upregulated when LATS1 expression was suppressed after METTL3 expression was deleted in breast cancer cells ( Fig. 5g and Supplementary Fig. 5g).

M6A regulation of LATS1 mRNA was identified by ythdf2 in breast cancer cells
LATS1 acts as a tumor suppressor in various human cancers [20]. Our previous study showed that an E3 ligase of LATS1 could destabilize the protein level of LATS1 by inducing its ubiquitination [12]. Since a high m6A modification level of LATS1 mRNA was found by MeRIP-seq, we aimed to determine which m6A readers directly recognize m6A modification sites and regulate the expression of LATS1 in breast cancer cells. A stability assay of LATS1 mRNA showed that METTL3 deletion promoted the stability of mRNA (Fig. 6a), which was also enhanced by mutating the m6A sites in LATS1 mRNA (Fig. 6b). Using an RNA pull-down assay and western blotting, we found that YTHDF2 was an m6A reader of LATS1 mRNA (Fig. 6c). It has been reported that YTHDF2 promoted the degradation of its target gene mRNA by recognizing m6A modifications [21]. We found that the expression of METTL3 had no influence on the expression of the m6A eraser FTO or the m6A reader YTHDF2 (Fig. 6d). However, the protein level of LATS1 was inversely correlated with the expression of YTHDF2 in breast cancer cells (Fig. 6e). To determine the impact of YTHDF2 in breast cancer cells and understand its role in the m6A methylation of LATS1 mRNA in tumorigenesis and glycolysis, we altered the expression of YTHDF2 in breast cancer cells (Fig. 6e). We found that m6A methylation of LATS1 downregulated LATS1 expression in breast cancer cells (Figs. 5d, 6e and Supplementary Fig. 5c-d) and activated YAP/TAZ by inhibiting its phosphorylation (Fig. 6f ). High levels of YAP/TAZ were found in the nucleus of breast cancer cells after inducing the expression of METTL3, while deleting the expression of YTHDF2 corrected such deviated expression in the nuclears (Fig. 6f ). The seahorse assay showed that YTHDF2 had a positive effect on glycolysis in breast cancer cells (Fig. 7a and Supplementary  Fig. 6a). The expression of YTHDF2 in breast cancer cells promoted tumorigenesis both in vitro and in vivo (Fig. 7b-c and Supplementary Fig. 6b-c). Since YTHDF2 destabilized LATS1 mRNA by reading the m6A methylation in LATS1, we predicted that deleting the expression of YTHDF2 could promote the tumor-suppressive effects of LATS1. Consistent with this hypothesis, tumor proliferation, survival, and invasion were inhibited by YTHDF2 depletion, and this tumor suppressive effect was rescued by inhibiting the expression of LATS1 (Fig. 7c-f and Supplementary Fig. 6c-f ).

Discussion
The Hippo signaling pathway plays an important role in the development and progression of breast cancer. It mainly controls organ size by regulating cell proliferation and apoptosis. Once the Hippo signaling pathway is inhibited, cells overcome contact inhibition and enter a state of uncontrolled proliferation. As the core effector molecule of the Hippo signaling pathway, LATS1 is key to tumor therapy research, and the regulatory element targeting LATS1 has also become a hot topic in tumor research. Among multiple transcription factors, RNA methylation is a new and important research direction in epigenetics. The m6A methylation modification of mRNA occurs via the participation of methyltransferase by dynamic control, and protein identification is determined by methylation, which affects protein localization, translation, degradation, and expression, ultimately contributing to the development of tumors. Currently, studies on m6A methylation in tumors have suggested that m6A modification determines the fate of RNA, but its mechanism has not been clarified. Therefore, we focused on the mechanism and impact of m6A modification on LATS1 mRNA. We demonstrated that Mettl3-mediated m6A methylation of LATS1 mRNA is recognized by YTHDF2, which reduces LATS1 expression and inactivates the Hippo pathway in breast cancer. We found that the deletion of METTL3 significantly suppressed proliferation, migration, and invasion in MCF-7 and T47D breast cancer cells, as well as tumor progression in vivo. Combined analysis of the multiple sequencing results revealed that METTL3 participates in the inhibition of LATS1 and affects tumorigenesis and metabolism in breast cancer cells. Considering the important role of LATS1 in tumor suppression, we investigated the exact mechanism of METTL3-mediated m6A modification of LATS1. We found that METTL3 mediates the methylation modification of m6A sites of LATS1 mRNA. Such m6A sites of LATS1 mRNA can be recognized by YTHDF2, which promotes the degradation of LATS1 mRNA and eventually tumorigenesis and glycolysis in breast cancer cells. Our findings suggest that METTL3 exerts oncogenic activity by inhibiting the expression of LATS1 through m6A modification. Therefore, our discovery of m6A modification of LATS1 mRNA poses a new regulatory mechanism for the expression of LATS1.

Conclusion
We discovered a negative regulatory mechanism of LATS1 mRNA, in which METTL3 mediated m6A modification of certain sites in LATS1 mRNA, and this modification was recognized by YTHDF2. Thus, the expression of LATS1 and the phosphorylation of proteins in the Hippo pathway in breast cancer were inhibited, ultimately leading to glycolysis and tumor development.  Fig. 7 Rescuing the protein level of LATS1 by altering the expression of YTHDF2 in breast cancer cells could help suppress tumorigenesis. a YTHDF2 deletion remarkably suppressed glycolysis progression in MCF-7 cells, *** p < 0.001. b Knocking out the expression of YTHDF2 significantly inhibited tumor growth in vivo, n = 7, * p < 0.05, ** p < 0.01. c The colony formation ability of MCF-7 cells was suppressed by knocking out the expression of YTHDF2 and reversed by inhibiting the expression of LATS1 at the same time. d YTHDF2 overexpression led to the rapid proliferation of MCF-7 cells and such stimulation could be rescued by LATS1 knockdown, n = 3, ** p < 0.01. e YTHDF2 deletion raised the apoptosis level of MCF-7 cells, while LATS1 knockdown rescued this stimulation, n = 3, ** p < 0.01, *** p < 0.001. f The suppressed invasion ability of MCF-7 cells affected by YTHDF2 knockout could be rescued by LATS1 knockdown, n = 3, ** p < 0.01, *** p < 0.001