Identification of eRNA associated with prognosis of GC
We used a Perl script to convert the eRNA transcript ID into a gene symbol for subsequent analysis. Then, the expression of eRNA was extracted from the expression matrix of GC and combine it with the survival time. As showed in Table 1, 23 eRNAs significantly related to the OS of GC were filtered by Kaplan-Meier method (p < 0.05). Different from other eRNA, HAGLR has five predicted target genes. Surprisingly, levels of these 23 eRNA were significantly correlated with their predicted levels of target gene mRNAs (r > 0.4, p < 0.001; Table 1).
LncRNA WAKMAR2 is a Key eRNA in GC
LncRNA WAKMAR2 was selected as eRNA for further study, and its expression level was positively correlated with its predicted target gene TNFAIP3 level. Compared with patients in the WAKMAR2 high expression group, patients in the WAKMAR2 low expression group had a shorter OS (Figure 1A, p < 0.05). As showed in Figure 1B, WAKMAR2 and TNFAIP3 mRNA levels are moderately correlated (r = 0.55, p < 0.001). It is worth noting that we studied the prognostic effect of WAKMAR2 in other cancer types and its correlation with TNFAIP3 mRNA levels. The impact of WAKMAR2 on OS and TNFAIP3 was specific for 8 types of cancer only, which were GC, Adrenocortical carcinoma, Breast invasive carcinoma, Brain Lower Grade Glioma, Mesothelioma, Pancreatic adenocarcinoma, Pheochromocytoma and Paraganglioma and Thymoma (Table 2).
Association between WAKMAR2 expression and clinicopathological features
In order to verify the potential clinical utility of WAKMAR2 expression, the clinicopathological features of 371 GC patients were included in this study (Table 3). Figure 2 summarizes the associations between WAKMAR2 expression and clinicopathological features. Compared with Stage I and Stage II, WAKMAR2 expression level was higher in Stage III (Stage I vs III, p < 0.05; Stage II vs III, p < 0.01). However, compared with stage II and stage III, WAKMAR2 expression was lower in stage IV (Stage II vs IV, p < 0.05; Stage III vs IV, p < 0.001). The difference in WAKMAR2 expression level between Grade2 and Grade3 was statistically significant (G2 vs G3, p < 0.001). Furthermore, higher WAKMAR2 expression levels correlated with advanced T stage (T2 vs T3, p < 0.01; T2 vs 4, p < 0.01). However, no difference was observed between age, gender, N stage, M stage, family history, neoplasm status, pylori infection, and radiation therapy.
Functional enrichment analysis
To further elucidate WAKMAR2 function, we used correlation analysis to identify significantly co-expressed genes in GC. Including TNFAIP3, a total of 2335 transcripts were significantly associated with WAKMAR2 (p<0.001). Co-expressed genes underwent GO enrichment analysis to identify the functions (Figure 3A). In BP category, "T cell activation", "T cell differentiation" and "lymphocyte differentiation" has been enriched, which means that co-expressed genes affect the function of the immune system in tumor microenvironment. Enriched CC terms included “phagocytic cup”, “mast cell granule”, “immunological synapse”, and the enriched MF terms included “cytidine deaminase activity”, “G protein−coupled chemoattractant receptor activity”, “chemokine receptor activity”. In addition, to determine the co-expressed gene enrichment pathway, we conducted KEGG enrichment analysis, including " Purine metabolism ", " Primary bile acid biosynthesis " and " Taurine and hypotaurine metabolism " (Figure 3B). We analyzed the correlation between genes which were enriched in the "immune response-activating cell surface receptor signaling pathway" and WAKMAR2 (p < 0.001). Immune genes with Spearman correlation coefficient > 0.40 were listed in Table 4.
To evaluate the clinical significance of immune checkpoint-related genes
The median value of WAKMAR2 expression was used as a cut-off value, patients were divided into high and low-expression groups. In this study, high-expression group patients exhibited higher gene expression of PDL1 (p < 0.001, Figure 4A), CTLA4 (p < 0.001, Figure 4B), PDCD1 (p < 0.001, Figure 4C) and LAG3 (p < 0.001, Figure 4D). However, the expression level of B7-H3 was higher in the low-expression group (p < 0.001, Figure 4e). The results of the study indicated that patients in the high-expression group were expected to be candidates for ICIs. Inhibitors against B7-H3 seem to have better therapeutic effects on patients in the low-expression group.
Differences in the abundance of immune cells between groups with high and low WAKMAR2 expression
Co-expressed genes have been shown to be involved in immune regulation, so we wanted to further analyze the differences in the immune fractions between the WAKMAR2 high-expression group and the WAKMAR2 low-expression group. The fractions of infiltrating immune cells were calculated by the CIBERSORT algorithm. Among the 22 leukocyte subtypes, B cells naive, T cells CD8, T cells CD4 memory activated and Macrophages M1 was positively correlated with WAKMAR2 expression (Figure 5). However, in the WAKMAR2 low-expression group, T cells CD4 memory resting, Macrophages M0, Macrophages M2, Mast cells activated, and Neutrophils accounted for a high proportion (Figure 5).
Correlation of WAKMAR2 expression level with TMB and MSI in different cancer types
As depicted in Figure 6A, the expression of WAKMAR2 was negatively correlated with TMB in Stomach adenocarcinoma (r = -0.288, p<0.001), Esophageal carcinoma (r = -0.222,p<0.01), Head and Neck squamous cell carcinoma (r = -0.154,p<0.001), Kidney renal clear cell carcinoma(r = -0.206,p<0.001), Liver hepatocellular carcinoma (r = -0.222,p<0.001), Lung adenocarcinoma (r = -0.121,p<0.01), Lung squamous cell carcinoma (r = -0.130,p<0.01), Pancreatic adenocarcinoma (r = -0.417,p<0.001), Pheochromocytoma and Paraganglioma (r=-0.295,p<0.001), Prostate adenocarcinoma (r = -0.267,p<0.001), Thyroid carcinoma (r = -0.161,p<0.001) and Thymoma (r = -0.768,p<0.001). On the contrary, the expression of WAKMAR2 was negatively correlated with TMB in Adrenocortical carcinoma (r = 0.277, p<0.05), Brian Lower Grade Glioma (r= 0.298, p<0.001), Ovarian serous cystadenocarcinoma (r = 0.133, p<0.05).
Similarly, the expression of WAKMAR2 was significantly correlated with MSI in Stomach adenocarcinoma (r = -0.128,p<0.05),Breast invasive carcinoma (r = 0.089,p<0.01), Lymphoid Neoplasm Diffuse Large B-cell Lymphoma (r = -0.460,p<0.01), Esophageal carcinoma (r = -0.220,p<0.01), Kidney renal clear cell carcinoma (r = -0.117,p<0.05), Kidney renal papillary cell carcinoma (r = -0.119,p<0.05), Rectum adenocarcinoma (r = 0.164,p<0.05), Thyroid carcinoma (r = 0.123,p<0.01) (Figure 6B).