Gene expression analysis data
We utilized the TIMER2 approach to analyse METTL3 expression levels across various cancer types in TCGA dataset. As shown in Fig. 1a,METTL3 expression levels in the tumour tissues of Bladder Cancer (BLCA), Cholangio Carcinoma(CHOL), Esophageal Carcinoma(ESCA), Head and Neck Squamous Cell Carcinoma (HNSC), Liver Hepatocellular Carcinoma (LIHC), Lung Adenocarcinoma (LUAD), Prostate Adenocarcinoma (PRAD), Rectal Adenocarcinoma (READ), Stomach Adenocarcinoma (STAD) (P < 0.001), Lung Squamous Cell Carcinoma(LUSC) (P < 0.01), Uterine Corpus Endometrial Carcinoma (UCEC) (P < 0.05), is higher than the corresponding control tissues. The expression levels of METTL3 in the tumour tissues of Thyroid Carcinoma (THCA) (P < 0.001), Kidney chromophobe (KICH), Pheochromocytoma and Paraganglioma (PCPG) (P < 0.05) were lower than those in the corresponding control tissues.
After including the normal tissues of the GTEx dataset as controls, we further evaluated the difference in METTL3 expression between normal and HNSCC tumour tissues. Figure 1b illustrates that the median expression level of METTL3 was higher in HNSCC tissues than in normal tissues.(Fig. 1b)
The CPTAC dataset results showed that the median Z-score of METTL3 total protein expression in normal HNSCC tissues was higher than that in primary tumour tissues (P < 0.05). (Fig. 1c)
Additionally, the "Pathological Stage Plot" module in GEPIA2 was employed to assess the correlation between METTL3 expression and the pathological stages of HNSCC. (Fig. 1d)
Survival analysis data
HNSCC cases were categorized into high and low expression groups based on METTL3 levels, mainly utilizing TCGA and GEO datasets to explore the correlation between METTL3 expression and HNSCC patient prognosis. Disease-free survival (DFS) analysis (Fig. 2b) revealed that high METTL3 expression is associated with a poor DFS prognosis in HNSCC patients within the TCGA project (P = 0.011). The hazard ratio (HR) for the high expression group was 1.5, with a p-value of 0.012, suggesting that patients with high METTL3 expression were 50% more likely to have disease recurrence than those with low expression. As shown in Fig. 2a, there was no significant difference in overall survival between the low and high METTL3 expression groups (P = 0.87). The hazard ratio (HR) for the high expression group was 0.98 with a p-value of 0.87, suggesting no significant risk difference between the groups.
Genetic alteration analysis data
We examined the genetic alteration status of METTL3 in different tumour samples from TCGA cohort. Figure 3a indicates that gene altered in 2.62% of 496 cases, with “amplification” type of CNA being slightly common in HNSCC at 1.21%, and the alteration frequency of METTL3 “mutation” is 1.01%, with “deep deletion” being the least frequent at 0.4% in HNSCC.
The types, sites, and case numbers of METTL3 genetic alterations are presented in Fig. 3b. We found that missense mutations are the predominant type of genetic alteration in METTL3 in HNSCC, with the R529C change within the MT-A70 domain inducing a missense mutation in the METTL3 gene and a mutation occurring in the X107_splice with a shallow deletion affecting a portion of the CNA of the METTL3 gene. At 529 site of METTL3, arginine (R) was replaced with cysteine (C). The R529C site was observed in the 3D structure of METTL3 protein (Fig. 3c).
Additionally, we explored the potential association between genetic alterations in METTL3 and the clinical survival prognosis of HNSCC patients. The data in Fig. 3c indicate that HNSCC patients with METTL3 alterations do not exhibit a better clinical prognosis.
Immune infiltration analysis data
Tumour-infiltrating immune cells have a significant effect on the development, progression, and prognosis of cancer [11]. In this study, we used algorithms such as TIMER, EPIC, MCPcounter, CIBERSORT, CIBERSORT-ABS, QUANTISEQ, and XCELL to explore the correlation between different levels of immune cell infiltration and METTL3 expression in HNSCC. The TIMER and XCELL algorithms were used to generate scatterplot data for HNSCC. We have found that in HNSCC, the immune infiltration of CD8 + T-cells based on the TIMER algorithm is negatively correlated with METTL3 expression(Fig. 4b, Rho=-0.165, P = 2.32e-04), while based on the XCELL algorithm, the immune infiltration of naive CD8 + T-cells is positively correlated with METTL3 expression(Fig. 4c, Rho = 0.102, P = 2.35e-02). Additionally, no significant correlation was found between METTL3 expression and the estimated infiltration value of cancer-associated fibroblasts in HNSCC. (Fig. 4a)
Enrichment analysis of METTL3-related partners
To further investigate the molecular mechanisms of the METTL3 gene in HNSCC, we utilized the STRING tool to screen for targeting METTL3-binding proteins and the METTL3 expression-correlated genes. Through a series of pathway enrichment analyses, we identified 50 METTL3-binding proteins, all of which are supported by experimental evidence. Figure 5a shows the interaction network of METTL3-binding proteins. We used the GEPIA2 tool to combine the HNSCC expression data and obtained the top 100 genes that correlated with METTL3 expression. Figure 5bshows that METTL3 expression levels were positively correlated with ACIN1 (R = 0.74), APEX1 (R = 0.71), PARP2 (R = 0.76), SUPT16H (R = 0.76), and TMEM55B (R = 0.68) expression levels (all P < 0.001). The heatmap data also displayed a positive correlation between METTL3 and the five genes in HNSCC (Fig. 5c). There was no common members of the METTL3-binding and correlated genes in the Intersection analysis of the two groups (Fig. 5d).
The two datasets were combined to perform KEGG and GO enrichment analyses. The KEGG data in Fig. 5esuggests that the “spliceosome” is likely to be involved in the effect of METTL3 on HNSCC.
The GO enrichment analysis data further indicated that these genes were related to the biological processes of proteins, DNA and ATP, such as Protein folding, ATP hydrolysis activity, DNA repair, and cellular stress response (Fig. 5f).