2.1 The expression of RNF208
We used the Oncomine database to investigate the mRNA expression levels of RNF208 in different tumors and the corresponding normal tissues to determine the differential expression between the tumor and normal tissues. RNF208 expression was higher in the breast, colorectal, gastric, and prostate cancers and lower in brain and CNS, esophageal, head and neck, kidney, pancreatic cancers, leukemia, and sarcoma (Fig. 1A). To further verify the expression of RNF208, we analyzed the expression of RNF208 in common tumors to determine the differential expression among the different cancers. The expression of RNF208 in normal tissues of LGGs and glioblastoma multiforme (GBM) was prominently higher than in tumor tissues (Fig. 1D). Simultaneously, RNF208 expression in Grade II was higher than Grade III (Fig. 1B). The HPA database also confirmed that the expression of RNF208 in brain tissue was higher than that in LGGs (Fig. 1C). According to the WHO grading system, gliomas were divided into four categories depending on how malignant they are, i.e. Grade I, Grade II, Grade III, Grade IV (12). We assessed the expression of RNF208 in WHO-grade groups in LGGs by using the data that was downloaded in the TCGA database. Cox regression was performed to display that high expression of RNF208 was significantly related to tumor grade (which divided into Grade II and Grade III, p = 0.001) and gender (Table 1, p = 0.015). To confirm this conclusion, we additionally conducted cox regression using data downloaded from the CGGA database. Similarly, high RNF208 expression was notably correlated with tumor grade (Table 1, p < 0.001).
2.2 Survival outcomes
As our above results indicated that RNF208 expression was significantly correlated with grade groups, we accessed the prognostic effect of RNF208 in LGGs using data downloaded from the TCGA database. High expression of RNF208 had a better prognosis than low RNF208 expression in GBM and LGGs (Fig. 2A, P < 0.001). Whereafter, we performed the survival curve of RNF208 in LGGs in GEPIA. Similarly, low expression of RNF208 had a poor prognosis (Fig. 2B, P = 0.029). As exhibited in Table 2, we executed univariate survival analysis using COX regression. Some elements, involving WHO grade system (HR = 3.085, P-value < 0.001) which was divided into Grade II and Grade III, age (HR = 5.548, P-value < 0.001) that was divided into groups of 60 or older and 60 or younger and RNF208 expression (HR = 0.454, P-value = 0.004), are markedly related to cumulative survival (Fig. 2D). In multivariate analysis, grade group is an independent prognostic element of a positive prognosis. Receiver operating characteristic (ROC) curves were used to evaluate the accuracy of survival (Fig. 2C). The area under the curve (AUC) for population survival was 0.747. These results demonstrate the accuracy of RNF208 in predicting survival in LGGs. RNF208 is a biomarker for good prognosis in LGGs.
2.3 Immune-related analysis
Studies showed that tumor immune microenvironment was critical to gliomas(13). Subsequently, we explored the immune cell infiltration levels in LGGs in TIMER database to research the relation between RNF208 expression level and immune infiltrating cells. B cell, CD8 + T cell, CD4 + T cell, macrophage, neutrophil, and dendritic cell showed significantly negative correlations with RNF208 (Fig. 3A). Moreover, the infiltration of B cells, CD8 + T cells, CD4 + T cells, macrophages, neutrophils, and dendritic cells and the expression of RNF208 were significantly correlated with the prognosis of LGGs (Fig. 3B). This suggests that RNF208 plays an important role in the regulation of immune cell infiltration in LGGs. To understand the relationship between RNF208 and various immune cells in more detail, we evaluated the proportion of 22 immune cells in LGGs with high or low expression of RNF208. T cells CD4 memory activated, NK cells activated, Macrophages M1 and Neutrophils were significantly different. (Fig. 3C). Correlation heat map of the proportions of 22 immune cells also showed varying degrees of correlations (Fig. 3E). TIMER database was applied to assess the correlation between RNF208 expression and marker levels in specific cell subpopulations including CD8+T cells, monocytes, M1 and M2 macrophages, total T cells, NK cells, B cells, TAMs, Tfh cells, neutrophils, DCs, Th17 cells, Th1 cells, Th2 cells, Tregs and exhausted T cells. We adjusted for tumor purity and found that RNF208 expression was associated with all of these markers (Table 3). Subsequently, we evaluated the correlation between RNF208 and immune checkpoints including PD1(PDCD1), PD-L1(CD274), PD-L2(PDCD1LG2), TIM-3(HAVCR2), and CTLA4 in GEPIA. We found that RNF208 had a negative correlation with all above, especially TIM-3 and PD-L2(Fig. 3D). Furthermore, the relation between RNF208 and checkpoints was evaluated in CGGA. In general, RNF028 was negatively associated with these immune checkpoints, especially PD-L2 (Supplementary Fig. 1).
2.4 RNF208-related genes
For further researching on the molecular mechanism of RNF208 in the occurrence of LGGs, we attempted to screen out the targeted RNF208 binding protein and RNF208 expression-related genes. A total of 52 RNF208 binding proteins supported by experimental evidence are obtained using the STRING database (Fig. 4A). We then used the UALCAN database to obtain all genes related to RNF208 expression in LGGs. The top 10 RNF208-related genes were compared in different clinical phenotypes (Fig. 4B). RNF208 expression is positively related to HSD11B1L, FBXL15, PIN1, AGAP3, and PPFIA3 (Fig. 4C). The heat map indicated that these five genes were positively related to RNF208 in most detailed cancer types (Fig. 4D). The cross-analysis of the above two groups shows that there are four common members, namely, H3F3C,HFE༌LITAF༌NANOS3(Fig. 4E).
2.5 Downstream signaling pathways in GSEA
To investigated related signaling pathways of different expression levels of RNF208 in LGGs, we performed GSEA analysis, which determine whether the gene sets show statistically significant (Fig. 5A). We partitioned the data sets into two sets-high expression and low expression with the median expression of RNF208 as a limit to analyze the differences among the groups. The most important pathways for down-regulated gene sets in the significance requirement (NOM p-value < 0.05, FDR q-value < 0.05) are listed in our study (Table 4). Twenty signaling pathways have discrepantly enriched in the RNF208 low expression phenotypic traits including JAK-STAT signaling pathway, apoptosis, Toll-like signaling pathway, B cell receptor signaling pathway, TGF-β signaling pathway,cell adhesion molecules CAMs, and so on (Fig. 5B).