Figure 3
The auto-regulation between STAT1 and miR-155-5p.
(a) Differential co-expression of the BiTM regulations formed by STAT1 and miR-155-5p. For each cancer type, the z-score of the transformed ρ of the auto-regulation formed by STAT1 and miR-155-5p in normal and tumor samples is shown. The two dash lines show a z-score of -2.5 and 2.5, respectively. The significance of |z-score| > 2.5 corresponds to P < 0.01. (b) The expression profiles of STAT1 and miR-155-5p in normal and tumor samples across the seven TCGA cancer types. The cancer type labeled in black indicates that the difference in expression profiles between tumor and normal is significant (P < 0.05, by edgeR). Green box: normal; red box: tumor. (c) The top 20 significantly enriched regulatory functional modules of STAT1. The GO functions colored by green, blue and red display mitosis, apoptosis and major histocompatibility complex (MHC) class I related functional modules, respectively. The column labeled with ″Combined″ represents the combined P-value of seven cancer types by Fisher′s method. For each cancer type, the P-value of significance for each functional module is indicated by the color code below the heat map. (d) The log2 fold ratio of STAT1 target genes in MHC I related functions. The log2 fold ratio is calculated by edgeR (tumor vs. normal). For each cancer type, the log2 fold ratio of each STAT1 target gene is indicated by the color code below the heat map. (e) The proposed hypothesis of the downstream effects caused by the loss of equilibrium mediated by the STAT1-miR-155-5p feedback loop. The potential molecular mechanism describes how the disequilibrium of the regulatory feedback loop between STAT1 and miR-155-5p triggers cancer immunoediting to escape from immunosurveillance.
