The Comprehensive Analysis of N6-Methyadenosine Writer METTL3 and METTL14 in Gastric Cancer

Methyltransferase-like 3 (METTL3) and methyltransferase-like 14 (METTL14) were two core components of the N6-methyadenosine (m6A) methyltransferase complex (MTC) and played a basic role in maintaining an appropriate m6A level of target genes. In gastric cancer (GC), previous researches on the expression and role of METTL3 and METTL14 were not consistent, and their specific function and mechanism have remained elusive. In this study, the expression of METTL3 and METTL14 was evaluated based on the TCGA database, 9 paired GEO datasets, and our 33 GC patient samples, and METTL3 was highly expressed and acted as a poor prognostic factor, whereas METTL14 showed no significant difference. Moreover, GO and GSEA analyses were performed, and the results pointed out that METTL3 and METTL14 were jointly involved in multiple biological processes, while they could also take part in different oncogenic pathways independently. And BCLAF1 was predicted and identified as a novel shared target of METTL3 and METTL14 in GC. In total, we conducted a comprehensive analysis of METTL3 and METTL14 in GC including their expression, function, and role, which could provide a novel insight into the research of m6A modification in GC.


Introduction
As a novel kind of posttranscriptional regulation, N6methyadenosine (m6A) RNA modifcation was one of the most abundant and prevalent RNA modifcations in eukaryotes [1][2][3]. It was a reversible and dynamic process that was installed by m6A methyltransferases "writers" and deleted by m6A demethylases "erasers" [4,5], and m6A sites on target RNAs that recognized by m6A-binding proteins "readers" afect multiple aspects of RNA metabolism, including their transcript, splicing, processing, translation, and decay [6].
To date, a large number of related studies have focused on the relationships between m6A modifcation and diseases, especially cancers [7]. Emerging evidence suggested that m6A modifcation played a critical role in multiple cancer processes through various mechanisms [8], and alteration of m6A levels on tumor-related genes dramatically afected the cancer development, proliferation, and metastasis, [7,9,10]. It was reported that methyltransferase-like 3 (METTL3)-mediated m6A modifcation of HDGF mRNA promotes gastric cancer (GC) progression and has prognostic signifcance [11]. Wang et al. [12] demonstrated that YTHDF1 promoted ARHGEF2 translation and RhoA signaling in colorectal cancer (CRC), and Pu et al. [13] reported that IGF2BP2 promoted liver cancer growth through an m6A-FEN1-dependent mechanism. Moreover, m6A modifcation was also involved in lung cancer [14], gliomas [15], and other diferent cancers.
In the digestive system, GC was one of the most common cancers worldwide, with a high incidence rate and mortality [16]. Te occurrence and metastasis of GC were considered a multifactor, multistep process, in which aberrant regulation of m6A modifcation was widely involved [11,[17][18][19].
METTL3 and METTL14 were two core components of the m6A methyltransferase complex (MTC) which was primarily responsible for m6A methylation [18,20]. Results of crystal structure indicated that METTL3 acted as the catalytic subunit bound to the methyl donor Sadenosylmethionine (SAM) and catalyzed methyl group transfer, and METTL14 was a cofactor that was necessary for substrate RNA binding and METTL3 conformation stabilization [20][21][22]. In GC, one major issue concerned the existing body of research suggested that METTL3 was overexpressed and promoted GC progression [11,19,23,24], while downregulated METTL14 acted as a tumor suppressor [25,26]. Tese two core components of MTC paradoxically exhibited opposite expressions and roles in GC progression. Previous studies have reported dysregulation of m6A modifcation in GC tumorigenesis [18][19][20], but few types of research focused on the expression and role of METTL3 and methyltransferase-like 14 (METTL14) and their relation in the process of m6A methylation. Terefore, to elucidate the molecular mechanism of m6A modifcation in GC, we performed a comprehensive analysis of METTL3 and METTL14 in GC, including their expression, function, and role, as shown in Figure 1.

Materials and Methods
2.1. TCGA Database Analysis. GEPIA2 [27], an enhanced web (https://gepia2.cancer-pku.cn/) server for large-scale expression profling and interactive analysis, was used to explore the expression of METTL3 and METTL14 between GC samples and normal samples in the TCGA database, and 444 eligible samples (408 GC tumor samples and 36 normal samples) were obtained. Diferential expression analysis was compared by the limma package in R software.

Prognostic Prediction.
Kaplan-Meier (K-M) (https:// kmplot.com/analysis/), one of the most comprehensive and authoritative online survival analysis websites, was used to predict the prognostic value of genes. To reveal the correlation between the expression of genes and the overall survival (OS) time of patients, the signifcance was computed using the Cox-Mantel (log-rank) test. Te diference between the cohorts is numerically characterized by the hazard rate (HR), which is based on the diferential descent rate of the two cohorts, and more information could be found by the authors of [37].

Gene Set Enrichment
Analysis. Te clusterProfler package in R software was used to perform gene set enrichment analysis (GSEA), and h.all.v6.2.symbols.gmt was chosen as the annotated gene set. Te enrichplot package in R software was used to draw the plots.

GO Functional Annotation.
Gene ontology (GO) functional annotation was used to explore the biological function of METTL3 and METTL14. Te clusterProfler package in R software was used to perform enrichment analysis, and the result was divided into three parts, including biological processes (BP), cellular components (CC), and molecular function (MF). Te enrichplot package, colorspace package, stringi package, and ggplot2 package in R software were also used. Te bar plots were drawn to visualize the top results.
2.6. Tissue Samples from GC Patients. 33 pairs of gastric and paracancer tissues were collected from Jiangsu Cancer Hospital (Nanjing, China), stored in RNA later (QIAGEN, Germany), and frozen in the freezer at −80°. All tissues were from patients who had been diagnosed pathologically as new cases of primary gastric cancer without radiotherapy.

Cell Lines Culture and Transfection.
Human gastric mucosal epithelial cells (GES-1) and two cancer cells (AGS, HGC-27) were obtained from the Key Laboratory of Environmental Medicine Engineering, Southeast University (Nanjing, China) and certifed by DNA fngerprinting. Te cells were cultured in a 5% CO 2 humidifed atmosphere at 37°C. GES-1 and HGC-27 were cultured in RPMI 1640 medium (Gibco, Gaithersburg, USA) supplemented with 10% fetal bovine serum (FBS). AGS was cultured in DMEM (Gibco) supplemented with 10% FBS. Te human METTL3 knockdown lentivirus, METTL14 overexpressing lentivirus, and the corresponding negative control lentivirus were synthesized by Hanheng Biotechnology Co (Nanjing, China), and the viral vector was pHBLV-CMV-MCS-3fag-EF1-ZsGreen-T2A-Puro. After 24 h of lentivirus treatment, the medium was removed and replaced with a fresh complete medium. After 72 h of lentivirus treatment, fuorescence was observed under the microscope, and puromycin was used to screen stably transfected cells.

RNA Isolation and RT-qPCR.
Gene expression levels in tissues and cells were measured by real-time fuorescent quantitative PCR quantitative reverse transcription PCR (RT qPCR). TRIzol reagent (Invitrogen, USA) was used to extract total RNA from cells and tissues, and purity and concentration were determined using a NanoDrop 2000 spectrophotometer (Termo Fisher Scientifc, USA). RT reactions and RT-qPCR were performed using the reverse transcription system kit (GenStar, Beijing, China), and the reverse transcription procedure was accomplished in two steps. General Biotech Co., Ltd. (Shanghai, China) provided all of the RNA primers. Supplementary Table S1 listed the mRNA primer sequences for candidate genes and housekeeping genes. Te comparative Ct (cycle threshold) was used for the comparison of gene expression, and the relative mRNA expression value was calculated as 40 −ΔCt [38].

Western
Blotting. Western blotting (WB) was used to detect the protein level expression, cells from transfected and negative control groups were collected, and proteins were extracted using a lysate mixture (198 μl RIPA and 2 μl PMSF). After measuring the concentration of protein samples, they were mixed with SDS loading bufer in equal volume and adjusted to the same concentration. Acrylamide gels were prepared (TGX ™ FastCast ™ Acrylamide Kit), 15 μl of the loading bufer was added to each well, and 15 μl Maker was added to the last well for indication. 90-volt electrophoresis was performed for about 2 h, and the membrane was transferred at 70-volt for 4 h and then closed with 5% skimmed milk powder. Te primary antibodies used here were shown as follows: BCLAF1(Proteintech, Cat No. 26809-1-AP), GAPDH (ABclonal, AC001), METTL3 (ABclonal, A8370), and METTL14 (ABclonal, A8530).
2.11. Statistical Analysis. For statistical analysis, SPSS 26.0 software (IBM Corp, USA), R software, and Excel software were utilized. All data were presented as mean ± standard deviation (SD) or median, and GraphPad Prism 9 was used for statistical analysis and graphing. Meta-analysis and forest plots were completed by Revman (Cochrane). A paired t-test was used to examine the diferences in gene expression levels between tumor and paracancer tissues. In diferential expression analysis, |log 2 FC| > 0.05 and P − value < 0.01 were considered to be statistically signifcant. In prognostic prediction, log-rank P < 0.05 was considered to be statistically signifcant. In GO functional annotation and GSEA, adjust P (P adj) value <0.05 and false discovery rate (FDR) q value <0.25 were considered to be statistically signifcant.  Journal of Oncology reported [11,19,23] that METTL3 was overexpressed in GC although METTL14 was low-expressed [43]. To begin, we explored their expression in GC based on the TCGA database, which contained 408 GC samples and 36 normal samples. As shown in Figure 2(a), the expression of METTL3 was signifcantly upregulated (P < 0.01) while METTL14 showed no signifcant diference.
For the inconsistent results between datasets and our tissues, a comprehensive meta-analysis was conducted to determine the expression of METTL3 and METTL14 in GC. All GEO datasets are mentioned previously, our tissues were involved, and I 2 > 50% indicated signifcant heterogeneity, and the random-efects model was used. Te pooled standard mean diference (SMD) of METTL3 was 0.95 (95% CI: 0.66 to 1.24, P < 0.001) (Figure 3(a)). Te pooled SMD of METTL14 was −0.09 (95% CI: −0.49 to 0.31, P � 0.67) ( Figure 3(b)). Terefore, we concluded that METTL3 was signifcantly overexpressed in GC, and METTL14 was low expressed but had no statistical signifcance.

Prediction of Prognostic Value.
Te online bioinformatics tool K-M plotter was used to calculate the hazard ratio (HR) and P value for OS. 1065 patients with accompanying clinical data were included in the K-M plotter database GC cohort [44]. As shown in Figure 4, high METTL3 expression was found to be a negative prognostic factor in GC patients, with those who had higher METTL3 expression having a shorter OS time than those who had lower METTL3 expression (HR � 1.46, log-rank P < 0.01). Contrarily, GC patients with higher METTL14 expression had a longer OS time (HR � 0.86), but the diference showed no signifcance.

M3DEGs and M14DEGs Jointly Enriched in Multiple Biological Processes in GC.
To discover more about the biological function of METTL3 and METTL14, GC patient samples in GSE66229 were chosen. Te median expression of METTL3 was used to divide GC patient samples into two groups, and METTL14 handled the same. Ten, diferential gene analysis was performed. Tere were 1246 diferentially expressed genes between the high-and low-METTL3 groups (METTL3 DEGs, M3DEGs) and 2098 diferentially expressed genes between the high-and low-METTL14 groups (METTL14 DEGs, M14DEGs). 218 genes between M3DEGs and M14DEGs were overlapped and named M3M14DEGs ( Figure 5(a)).
GO function annotation was frst performed on these M3DEGs and METTL14 DEGs. Te results were divided into three parts: MF, BP, and CC, and the bar plots were drawn to visualize the top results of each part (Figures 5(b) and 5(c)). As expected, the results of M3DEGs and M14DEGs were both signifcantly enriched in the modifcation and regulation of RNA, and 32 GO terms were overlapped, including regulation of mRNA processing (GO: 0050684), ncRNA processing (GO: 0034470), RNA splicing (GO: 0008380), regulation of mRNA metabolic process (GO: 1903311), ribonucleoprotein complex biogenesis (GO: 0022613), and ncRNA metabolic process (GO: 0034660). Moreover, GO function annotation of M3M14DEGs demonstrated similar results ( Figure 5(d)). Tese results indicated that METTL3 and METTL14 jointly participated in multiple biological processes in GC, which supported the cooperative role of METTL3 and METTL14 in m6A modifcation.

METTL3 and METTL14 Participated in the Specifc
Carcinogenic Pathways in GC. GSEA on M3DEGs and M14DEGs was also performed to gain a deeper insight into how METTL3 and METTL14 play a role in the GC process. Te Hallmark gene set in the MSigDB collections was chosen as the reference gene set. As the enrichment plots shown, E2F_TARGETS (NES � 2.54, P adj � 0.0038, Figure 6(a)), G2M_CHECKPOINT (NES � 2.31, P adj � 0.0038, Figure 6(b)), and MYC _TARGETS_V1 (NES � 1.76, P adj � 0.0047, Figure 6(c)) were most commonly enriched in M3DGEs, showing the association between M3DGEs and these carcinogenic signaling pathways; M14DEGs was signifcantly enriched in INTERFERON_GAMMA_RES-PONSE (NES � −2.77, P adj � 0.0047, Figure 6(d)) and negatively correlated with MTORC1_SIGNALING (NES � 2.31, P adj � 0.0047, Figure 6(e)) and GLYCOLYSIS (NES � 2.18, P adj � 0.0047, Figure 6(f )). Tese results pointed out that METTL3 and METTL14 took part in diferent carcinogenic pathways in GC cancer progression, which hinted that METTL3 and METTL14 might target diferent genes in the GC process except in addition to coworking.

Target Genes Prediction of METTL3 and METTL14
3.3.1. BCLAF1 as a Novel m6A Target in GC. 4 m6A target prediction databases [39][40][41][42] were utilized to predict the m6A-modifed genes of METTL3 and METTL14, and 54 validated targets of METTL3 and METTL14 were discovered. We intersected M3M14DEGs with 54 validated targets to get our target gene: BCL2 associated transcription factor 1(BCLAF1) (Figure 7(a)), which might be comodifed by METTL3 and METTL14 in GC. BCLAF1 was signifcantly high expressed in GC (Figure 7(b)), and the correlation analysis between METTL3 and BCLAF1, and METTL14 and BCLAF1 in GC revealed a strong positive connection (Figure 7(c)). As a result, we hypothesized that METTL3 and METTL14 working together to produce m6A modifcation on BCLAF1 would enhance its expression.
Terefore, lentiviral infection was used for the construction of cell transfection models. Based on the highexpressed METTL3 and low-expressed METTL14 in GC cells (Figure 7(d)), we established the METTL3 knockdown (METTL3-KD) cell transfection models and METTL14 overexpressed (METTL14-OE) cell transfection models in AGS and HGC-27 cell, and the transfection efciency was verifed by RT-qPCR and WB (Figures 7(e) and 8(b)). Interestingly, we found that the mRNA expression of METTL14 was upregulated after METTL3 knockdown (Figure 7(f )).
Our RT-qPCR results showed that BCLAF1 was dramatically decreased after METTL3 knockdown and was signifcantly upregulated after overexpressing METTL14 (Figure 8(a)). Meanwhile, these results were verifed in protein level in HGC-27 and AGS cells by WB, as shown in Figure 8(b).     Journal of Oncology

m6A Modifcation on PTEN Mediated by METTL3/ METTL14 Played an Opposite Role in Its Expression.
Moreover, this study retrieved that the articles related to METTL3 or METT14 in GC, METTL3-modifed genes, and METTL14-modifed genes mentioned in GC articles were met in the middle (Figure 8(c), Supplementary Table S2), and phosphatase and tension homolog (PTEN) was overlapped [26,45]. Acting as a classical tumor suppressor in the cancer process, PTEN was a key negative regulator in the PI3K signaling pathway [40]. In GC, PTEN was low expressed, and Yan et al. [45] found that METTL3 facilitated m6A-YTHDF2-dependent PTEN mRNA degradation (Figure 9(b)). Interestingly, Yao et al. [26] discovered that METTL14-mediated m6A modifcation on PTEN enhanced its mRNA stability (Figure 9(c)). Intrigued by these inconsistent fndings, our transient transfection cell models were used to validate their fndings. Knockdown of METTL3 reduced m6A modifcation on PTEN and signifcantly increased its mRNA expression, whereas overexpressed METTL14-enhanced m6A modifcation on PTEN also signifcantly increased its mRNA expression (Figure 8(d)).

Discussion
In this study, we frst determined the expression of METTL3 and METTL14 in GC. A comprehensive meta-analysis based on 9 paired GEO datasets and 33 GC tissue samples validated that METLL3 was signifcantly high expressed while the expression of METTL14 showed no signifcant diference in GC, which was consistent with the result of the TCGA database. Survival analysis results showed that METTL3 was a poor prognostic factor for GC patients while METTL14 had less prognostic value.
GO function annotation was also performed, and the result showed that M3DEGs and M14DEGs both enriched in 32 GO terms, including regulation of mRNA processing, RNA splicing, ncRNA processing, ribonucleoprotein complex biogenesis, ncRNA metabolic process, regulation of mRNA metabolic process, RNA metabolic process. It was considered that METTL3 and METTL14 jointly participated in multiple biological processes, implying the cooperative role of METTL3 and METTL14 in m6A modifcation.
Target genes of METTL3 and METTL14 were predicted in this study, and BCLAF1 was found. It was upregulated and reported as an oncogene [46][47][48]. m6A site was detected on the coding sequence (CDS) of BCLAF1 mRNA that METTL3 and METTL14 could combine with [49], and BCLAF1 revealed a strong positive correlation with METTL3 and METTL14. In our cell transfection models, the mRNA expression and the protein level of BCLAF1 were decreased after METTL3 knockdown and were upregulated after overexpressing METTL14. Taken together, we indicated that METTL3 and METTL14 jointly mediated m6A modifcation on BCLAF1 and promoted its mRNA expression (Figure 9(a)).
On the other hand, GSEA results demonstrated that M3DEGs and M14DEGs took part in diferent oncogenic pathways in the GC process, hinting that METTL3 or METTL14 might mediate m6A methylation on diferent target RNAs independently. PTEN was one of the classical tumor suppressors in the GC process, and the m6A modifcation on PTEN was deeply discussed. It was reported that METTL3 facilitated m6A-YTHDF2-dependent PTEN mRNA degradation while METTL14-mediated m6A modifcation on PTEN enhanced its mRNA stability (Figures 9(b) and 9(c)) [26,45].
It was necessary to emphasize that the role of m6A modifcation in all cancers had two sides. According to our understanding, the biological function of m6A modifcation on target RNA and cancer progress was a multistage process and is determined by multiple factors (Figure 9(d)): frstly, m6A writers and erasers determined the m6A level of target RNA. Besides, the m6A level of target RNA and the m6Abinding proteins "readers" combined on target RNA determined its expression level. Finally, the own function and downstream pathway target RNA also should be considered.
Combining the GSEA results previously, we hypothesized that except for the synergistic efect of METTL3 and METTL14 on m6A MTC, each of them could mediate m6A nidifcation on target RNAs independently, and the m6A site could be identifed by diferent readers under certain conditions. Terefore, the situation that METTL3-mediated RNA modifcation and METTL14-mediated RNA modifcation played an opposite role in the expression of PTEN and cancer progress could be explained.
While METTL3 and METTL14 both contained a methyltransferase catalytic domain, current theories mainly considered [20,21] that METTL3 methylated m6A sites and METTL14 stabilized METTL3 conformation, producing m6A sites on target RNAs together. Our hypothesis needed further experimental validation and proof of chemical structure.

Conclusion
In conclusion, the present study conducted a comprehensive analysis of METTL3 and METTL14 in GC, including their expression, function, and role in GC. BCLAF1 was identifed as a shared target of METTL3 and METTL14. Tis study provided a deeper insight into the function of m6A modifcation in the cancer process and hoped it could be benefcial to the mechanism exploration of m6A modifcation.

Data Availability
Te data that support the fndings of this study are available from the corresponding author upon reasonable request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.  Figure 9: Te biological function of m6A modifcation in the cancer progress is a multistage process and is determined by multiple factors [26,45].

Authors' Contributions
Ge Yiling and Zhang Yihan; Liu Tong and Feng Yanlu contributed to the fgures and tables of the article. All authors revised and approved the fnal manuscript.