The expressions of m6A regulators were related to the development of ccRCC
According to the previous studies, we selected 21 m6A regulators for further investigation. As shown in Figure 1A-B, the expressions of 15 m6A regulators were significantly different between ccRCC and normal kidney tissues, among which the expressions of METTL3, WTAP, RBM15, FTO, ALKBH5, YTHDC2 and IGF2BP3 were higher in ccRCC samples than that of normal samples. Compared with ccRCC samples, the expressions of METTL14, ZC3H13, RBM15B, YTHDF2, YTHDF3, IGF2BP2, HNRNPC and HNRNPA2B1 significantly increased in normal kidney samples. The result of Kaplan-Meier analysis showed that nine m6A regulators including METTL3, WTAP, YTHDC2, IGF2BP3, METTL14, ZC3H13, YTHDF3, IGF2BP2, and HNRNPA2B1 were associated with the prognosis of ccRCC patients. ccRCC patients with higher expressions of METTL3, IGF2BP3, WTAP, IGF2BP2 and HNRNPA2B1 had a more favorable prognosis, while patients with higher expressions of METTL14, ZC3H13, YTHDF3 and YTHDC2 had a less survival time (Figure 1C). These results indicated that the expressions of m6A regulators were related to the development of ccRCC.
Identification of two subtypes in ccRCC based on the expressions of m6A regulators
Consensus clustering was then applied to classify ccRCC patients into clusters based on the expressions of these 15 m6A regulators. Clearly, K=2 was identified with optimal clustering (Figure 2A-B). A total of 479 ccRCC patients were clustered into two subgroups with 337 patients in cluster 1 and 142 patients in cluster 2. Besides, the results of PCA clearly showed two distinct distribution of the two clusters (Figure 2C). Kaplan-Meier analysis showed that the patients of cluster 1 had a better OS compared to those of cluster 2 (Figure 2D).
Biological function analysis of DEGs between the two clusters
To better characterize the difference of biological process involved in the development of ccRCC between the two clusters, we compared the expressions of genes in the two clusters and found a total of 4,429 DEGs (Figure 3A). To get further insights into the biological characteristics of the DEGs, we performed GO and KEGG pathway analysis. GO analysis showed that the DEGs were enriched in immune-related functions, such as regulation of immune effector process, production of molecular mediator of immune response, immune response-activating cell surface receptor signaling pathway, lymphocyte mediated immunity, B cell mediated immunity, immunoglobulin mediated immune response and regulation of humoral immune response (Figure 3B). Moreover, the result of KEGG pathway analysis revealed several immune-related pathways, such as cytokine-cytokine receptor interaction, calcium signaling pathway, complement and coagulation cascades and bile secretion (Figure 3C). These results demonstrated that m6A regulators were involved in the immunity of ccRCC patients.
Identification of different TME between the two clusters
It has been reported that TME is closely associated with immune response in ccRCC progression [17]. Therefore, we studied the TME characteristics of the two clusters to further explore the role of m6A regulators in ccRCC. The expression pattern of m6A regulators and TME features in cluster 1 and cluster 2 were shown by the heatmap (Figure 4A). Specifically, the immune and ESTIMATE scores were remarkably higher in cluster 2 (P = 0.0027 and P = 0.018, respectively), and the stromal score of cluster 2 was relatively higher with no statistical significance compared to cluster 1 (P = 0.8) (Figure 4B-D), suggesting that m6A regulators were correlated with immune infiltration and tumor purity.
Identification of different immune landscape between the two clusters
Given the significant difference of TME between the two clusters, we performed the CIBERSORT to estimate the immune infiltration of the two clusters. The distribution of immune cells were compared between cluster 1 and cluster 2 (Figure 5A-C). The results showed that the abundance of 11 immune cell types were significantly different between two clusters, with higher proportions of naive B cells, CD4+ naive T cells, regulatory T cells, activated CD4+ memory T cells and M0 macrophages and lower plasma cells, M1 macrophages, resting NK cells, monocytes, resting rest cells and neutrophils in cluster 2. These results indicated that clustering subgroups based on the expressions of m6A regulators were closely related to the immune microenvironment in ccRCC.
Identification of different expressions of HLA, immune checkpoint molecules and Th1/IFNγ pathway signature between the two clusters
HLA, immune checkpoints and Th1/IFNγ pathway related genes can interact with TME and play vital roles in the immunoregulation of multiple tumors. Therefore, we further investigated whether m6A regulators influence on the expressions of HLA and immune checkpoint molecules. Interestingly, we found 12 HLA gene expressions significantly differed between two clusters, among which expressions of 11 HLA genes (P < 0.05), including HLA-W, HLA-T, HLA-P, HLA-K, HLA-J, HLA-H, HLA-G, HLA-F, HLA-E, HLA-DQB1, and HLA-DPA2, were significantly up-regulated in cluster 1 (Figure 6A). The 19 immune checkpoints molecules were found to be remarkably differentially expressed between two clusters, of which cluster 2 had higher expressions of 17 immune checkpoints (P < 0.05), including BTLA, CD27, CD276, CD28, CD40LG, CD47, CD80, CD86, ICOS, LAG3, LGALS9, PDCD1, PVR, TIGIT, TNFRSF18, TNFSF14 and TNFSF4 (Figure 6B). As for the genes involved in Th1/IFNγ pathway, the expressions of IFNGR2, IFNγ and STAT1 were up-regulated in cluster 2 (P < 0.05), whereas the expressions of JAK1 and JAK2 were up-regulated in cluster 1 (P < 0.05) (Figure 6C). These results indicated that m6A regulators may affect the immune response and TME by regulating the expressions of HLA, immune checkpoint molecules and Th1/IFNγ pathway gene signatures.
It has been reported that both TME and immune checkpoints are associated with therapeutic response [22]. Therefore, in consideration of different TME and expressions of immune checkpoints in the two clusters based on the expressions of m6A regulators, we investigated the response of two clusters to PD-1 and CTLA-4 blockade, which has been used in the clinical treatment of ccRCC patients [23]. As shown in Figure 6D, cluster 2 presented a very poor response to anti-CTLA-4 therapy (Bonferroni corrected P = 0.048). These results indicated that m6A regulators have a close relationship with therapeutic efficacy.