DEGs pf UM based on immune scores and stromal scores
The original mRNA expression data of 80 UMs were obtained from TCGA, following by the calculation of ESTIMATE algorithm of immune scores, stromal scores and ESTIMATE scores. The median value of the scores was taken as the midpoint to divide patients into the high- and low-expression groups with the survival of patients calculated by KM-plot. Overall, immune scores were significantly associated with the survival of patients (P=0.038) (Fig1. B). To explore the potential association of immune microenvironment with immune scores and stromal scores of UM patients, the differences between both high/low immune scores and high/low stromal scores were analyzed. A total of 1538 differential genes were screened by immune score-related differential analysis, involving 1101 up-regulated genes and 437 down-regulated genes, while 1693 differential genes were screened by stromal score-related differential analysis, including 1555 up-regulated genes and 138 down-regulated genes (Fig2. A, B). The Venn plot showed the intersection of them that there were 888 up-regulated genes and 126 down-regulated genes in total (Fig2.C), which we regarded as DEGs.
Gene annotation and pathway enrichment of DEGs
We attempted to explore the enrichment of signaling pathways related to the DEGs in microenvironment, GO and KEGG were both demonstrated to enrich the DEGs. As shown in Fig2.D, DEGs were mainly enriched in adaptive immune response based on somatic recombination of immune receptors built from immunoglobulin superfamily domains, immune response−activating cell surface receptor signaling pathway, lymphocyte mediated immunity and regulation of lymphocyte activation pathways. Subsequently, we demonstrated KEGG pathway enrichment, showing that DEGs were majorly enrichen in Allograft rejection, Cytokine−cytokine receptor interaction and Hematopoietic cell lineage signaling pathways(Fig2.E).
Construction of PPI network and identification of key genes
To further research the interaction among the DEGs and then screen the critical genes, we constructed a PPI network based on the STRING online database, exhibiting in Fig3 by Cytoscape visualization. We maintained the top 10 genes, basing on KM-plot method to investigate whether these genes affected the survival of UM patients. Two genes which regarded as key genes were negatively associated with overall survival according to our result (both P<0.05) and were shown in Fig4.A.
Relationship between key genes and immune infiltration
Tumor microenvironment has a marked impact on the diagnosis, survival and clinical treatment sensitivity. Therefore, we further explored the potential molecular mechanism of risk scores affecting the progression of UM by investigating the relationship among them. As shown in Fig4.B, we observed that two key genes, B2M and HLA-B, were highly corelated with most of the immune cells, especially with Dendritic cells resting, Macrophages M1, Mast cells resting, Monocytes, NK cells activated, NK cells resting, T cells CD4 memory activated, T cells CD4 memory resting, T cells CD8, T cells follicular helper and T cells regulatory. In addition, we retrieved the critical genes related to UM based on GeneCards database, keeping the top 20 genes with the highest score for correlation analysis with B2M and HLA-B, which revealed that they were both significantly correlated with UM disease genes Fig4.C, especially closely linked with BAP1, BRCA2, CDK4, CTNNB1, KIT, MC1R, MDM2, RNAS and SF3B1. Next, the relationship of B2M and HLA-B with immune infiltration was implemented by TIMER database separately, whose results showed that both genes were closely associated with CD8+ T Cell and Neutrophils (Fig5.A). These results suggested that B2M and HLA-B were highly linked with both the immune state and the UM process, providing the potential targets of immunotherapy and the latent effective prognosis biomarkers.
Signaling pathways of UM
The specific signaling pathways involved in B2M and HLA-B were dug to investigate the underlying molecular mechanisms that may influence tumor progression. GSVA results revealed that the differences between the two groups of patients mainly enriched in TNFA_SIGNALING_VIA_NFKB, PEROXISOME and PI3K_AKT_Mtor_signaling pathways and the like (Fig5). Since there were 42 pathways involving B2M(Fig5.B) and 41 pathways concerned HLA-B(Fig5.C) in total, suggesting that the prognosis of UM patients who were grouped by B2M and HLA-B gene expression levels may affected by disturbance of these signaling pathways, provides these functions as potential targets for further research.
Drug sensitivity of B2M and HLA-B
Based on the data from the GDSC database, we predicted the chemotherapy sensitivity of each tumor sample through the R package "pRRophetic" to further discuss the risk score and the sensitivity of common chemotherapeutic drugs. Fortunately, we found that HLA-B was conspicuously related to the sensitivity of patients to Paclitaxel (Fig5.D), suggesting Paclitaxel may be an effective medicine to the treatment of HLA-B mutated patients.