Mutations and CNVs of m6A RNA methylation regulators in SKCM
Among the SKCM cases with CNV (Copy number variations) data from TCGA dataset, the CNVs mutation pattern of variation frequency showed that the number of copies gained (Red) is less than the number of copies lost (Green) (Fig. 1a). In detail, the frequency of CNVs of gene WTAP was the highest (9.75%,46/472), that of VIRMA was the lowest from 23 m6A regulatory genes (6.36%,30/472), which were m6A “writer” genes, implying an important role of m6A writer genes in SKCM. As shown in Table 1, a total of the CNV mutation patterns by this sequencing analysis were depended on 24 chromosomes. We also observed the circles to indicate the relationship clearly (Fig. 1b). The results also revealed that CNVs led to loss of copy number, while the outermost ring represents the human chromosomes and highlights the m6A methylation regulators on this map.
Expression of m6A methylation regulators and clinicopathological features in SKCM
To assess the distinct expression pattern of m6A regulators in the character and function of SKCM, we scrupulously investigated the RNA expression profiles of 23 m6A regulatory genes between 471 tumor samples and 273 normal skin pairs, derived from the available TCGA and GTEx datasets. The expression levels of “writers” (i.e., ZC3H13, RBM15, and RBM15B,), “readers” (i.e., YTHDF1, YTHDF2, YTHDF3, LRPPRC, and IGFBP2), and “erasers”(ALKBH5) were memorably higher in SKCM tissues than in normal adjacent tissues(p < 0.001) (Fig. 2a). Furthermore, we downloaded their corresponding clinical data and incorporated 460 samples from TCGA datasets and 79 samples from GEO datasets, then assessed by Kaplan-Meier survival analysis paired with Log-rank test. The results revealed the 13 m6A regulated genes were poor survivals(p < 0.05) (Fig. 2b). To further understand the interactions among m6A regulators, we found that network map showed the top 20 significant survival favorable factors and WTAP gene and LRPPRC had the strongest correlation(Fig. 2c).
Furthermore, the results of genetic expression in the TCGA database, 97/467 (20.77%)patients showed genetic alteration of m6A methylation regulators. The 9 m6A methylation regulators showed in Fig. s1d. Missense mutation was the highest mutation in the six variant classifications. YTHDC1,ZC3H13,LRPPPRC, YTHDC1, and YTHDF1 were the most frequently altered m6A methylation genes(an alteration rate of 3%). And the results of single nucleotide variants (SNVs) indicated T > A had the highest incidence in six base substitutions (C > A, C > G, C > T, T > A, T > C and T > G). The waterfall map summarized the high mutated genes and their mutation classification of SKCM(Fig. 3a). Additionally, the results indicated that the expression of IGFBP1 gene was significant difference with ZC3H13 wild and ZC3H13 mutation samples(p = 0.037)(Fig. 3b).
Significant correlation of consensus clustering for m 6 A RNA methylation regulators
The k = 3 was identified with optimal clustering increasing from k = 2 to 9 based on the similarity displayed by the expression levels of m6A regulators and the proportion of ambiguous clustering measure((Figs. s1a, s1b and s1c). A total of 471 patients with SKCM were clustered into three subtypes, namely, m6Acluster.A(n = 278), cluster B(n = 199), and m6Acluster.C(n = 70) (Fig. 4a). And the overall survival(OS, p = 0.003) in three subtypes is significant correlation(Fig. 4b). Principal component analysis(PCA) could better distinguish between patients in three clusters(Fig. 4c). On basis of correlation analysis of clinical characteristics, the expression of individual m6A methylation regulator was the lowest in m6Acluster.B, especially the expression levels of IGFBP2. Therefore, the clinicopathological features between the three subtypes were then compared in the heat map(Fig. 4d). Above this outcome we got 15 differential genes and performed the ssGSEA to conduct enrichment analysis(Fig. 4e).The results from gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis indicated the 15 differential genes were significantly enriched in ten related pathways such as regulation of metal ion transport, and regulation of cell morphogenesis(Figs. s2a and s2b).
Consensus clustering of three distinct m6A patterns based on the m6A RNA methylation regulators associated with TIME
All m6A regulators were enrolled in univariate and multivariate cox regression. Finally, 4 variables genes including CYP2U1, SEMA6A, GRIA2, and TSPAN13 were selected in cox regression(Table. 2). We obtained m6A-related genes between the three distinct m6A patterns, including m6A.gene.cluster.A, B, and C(Fig. 5a). It also depended on the proportion of ambiguous clustering measure (Figs. s2c, s2c and s2e). Then, we compared these genes with m6A methylation regulators. And the boxplot indicated that most m6A regulators were high expression in m6A.gene.cluster.C(Fig. 5b). And the overall survival(OS) between three subtypes was significant correlation (p < 0.001)(Fig. 5c). Subsequently, the correlation with clinical features and four variables’ genes was showed in the heat map(Fig. 5d).
Association of IFN-γ with m6A RNA methylation in SKCM
To explore the involvement of IFN-γ with m6A RNA methylation, we assessed differential expression in two subtypes and the correlation of IFN-γ with m6A regulators. The expression level of IFN-γ was upregulated in SKCM tissues compared with normal adjacent tissues(p < 0.001; Fig. 6a). Subsequently, we evaluated m6A score and divided them into high and low subgroups to investigate the effect of m6A regulators on the TME of SKCM. The expression level of IFN-γ in the m6A score cluster was distinctly higher than that in the high m6A score cluster (p = 0.0074; Fig. 6b). In the TCGA cohort, the expression of IFN-γ had a significantly positive association with WTAP, YTHDC2, RBM15 and FMR1 expression levels, whereas a significantly negative correlation was noted with METTL3, METTL16,VIRMA, RBM15B, YTHDC1, YTHDF1, YTHDF2, YTHDF3, LRPPRC, HNRNPA2B1, IGFBP1, IGFBP2, IGFBP3, RBMX, and FTO expression levels. Among them, IFN-γ and WTAP were the most positive correlation(Fig. 6c). The association of m6A types and SKCM patients’ prognosis was indicated by comparing survival rates of two m6A score subgroups. Therefore, the OS of the low m6A score were related to poor survival with expression profiles of 23 m6A regulatory genes(p < 0.001)(Fig. 6d). Subsequently, we analyzed the fraction of 23 immune cell types between the three m6A clusters, which were significant difference in these immune infiltrate level. The m6Acluster.B showed higher infiltration levels of Activated CD4.T.cellna, and Eosinophilia(Fig. 6e), whereas m6Acluster.C was more correlated with Mast cellna. Subsequently, we inferred that there is complex correlation among m6Acluster, m6A.gene.cluster, m6A scores, and survival state. The m6A.gene.cluster.A was corresponding to high m6A score, and half was alive, we drew a Sankey diagram to explore these correlations(Fig. 6f). The co-expression heat map of diversified immune cells seen in Fig. 6g. It indicated that Activated CD4.T.cellna was positively correlated with Activated CD8.T.cellna and Immature.B.cellna. Finally, we assessed the patients in m6Acluster.C and m6A.gene.cluster.C had higher m6A score.(Figs. 6h and 6i).
Gene Set Enrichment Analysis in SKCM
In Gene Set Enrichment Analysis(GSEA), the functional meanings pathway analysis is based on the KEGG database. It shows a strong significant enrichment in “Chemokine signaling pathway” when the gene set was put in order by NES(NES = 2.81, p < 0.0001; Fig. 7a). The following is “Fc gamma R-mediated phagocytosis”(NES = 2.74, p < 0.0001), “Natural killer cell mediated cytotoxicity” (NES = 2.69, p < 0.0001), and “Toll-like receptor signaling pathway” (NES = 2.68, p < 0.0001). The detail information of former 10 significantly enriched pathway are listed in Table s1. As the same based on Reactome Pathway Knowledge, we found that “Programmed Cell Death” was the most significant term(NES = 3.08, p < 0.0001; Fig. 7b), followed by “C-type lectin receptors” (NES = 3.07, p < 0.0001), “Dectin-1 signaling”(NES = 3.03, p < 0.0001), “Signaling by Interleukins” (NES = 2.95, p < 0.0001). Other details are listed in Table s2.
The biological process ontology analysis subset of gene ontology(BP subset of GO analysis) in GSEA also showed enrichment of gene sets involved in “GOBP activation of innate immune response” (NES = 3.07, p < 0.0001; Fig. 7c), “GOBP innate immune response activating signal transduction” (NES = 2.99, p < 0.0001), and “GOBP response to interleukin 12”(NES = 2.98, p < 0.0001). The other details are listed in Table s3.
In addition, GSEA revealed immunologic signature gene sets “GSE19941 IL10 KO vs. IL10 KO and NFKBP50 KO LPS STIM macrophage DN”(NES = 3.29, p < 0.0001; Fig. 7d), “GSE12845 IGD NEG blood vs. naïve tonsil bcell UP” (NES = 3.26, p < 0.0001), and“GSE25088 ROSIGLITAZONE vs. IL4 and ROSIGLITAZONE STIM STAT6 KO MACROPHAGE DAY10 DN” (NES = 3.21, p < 0.0001). Other information are summarized in Table s4.
Other clinicopathological features and TMB outcome by m6A score
The OS was obtained when TMB was dichotomized at intermediate to high versus low, and the survival differences were statistically significant(p < 0.001, Fig. 8a). Then, we depend on the m6A score cluster was significant(p < 0.001, Fig. 8b). Hence, the mutational landscapes. TTN gene is the most in high- m6A score and low-m6A score (Figs. 8c and 8d). High- m6A score in alive is 58%, and low- m6A score in alive is 40%. The survival state including alive and dead were obvious differences(p = 0.029, Figs. s4a and s4b). Other clinicopathological features such as stage and age, we complain 1-(p = 0.003), 3–4(p < 0.001)and > 60, <=60(Figs. s4c and s4d) in the next.