Immune and inflammation-related genes play crucial roles in KIRC
We constructed an immune, inflammation or KIRC-directed neighbour network (IIKDN network), which is a sub-network of the PPI network (Fig. 1a). The IIKDN network contained 5391 nodes and 15,411 edges. The degree of all the nodes in the IIKDN network showed a scale-free distribution (R-square = 0.900) (Fig. 1b). We also divided the genes in the IIKDN network into four types. We defined the gene as “three types” if the gene was immune-related, inflammation-related gene and KIRC-related. We defined the gene as “two types” if the gene was both immune- and inflammation-related or immune- and KIRC-related or inflammation- and KIRC-related. We defined the gene as “one type” if the gene was only immune-related or inflammation-related or KIRC-related. We defined the gene as “other type” if the gene belonged to none of the above mentioned gene types. In the IIKDN network, there were 36 “three types” genes, 272 “two types” genes and 1056 “one type” gene (Fig. 1c, d). We found that the “three types” genes had the highest degree (average degree = 21.9), and the average degree of the “other type” genes was far lower than that for the immune-, inflammation- and KIRC-related genes (Fig. 1e). The result suggested immune- and inflammation-related genes play more essential roles than other genes in KIRC. The result also indicated complex links among the immune-related, inflammation-related and KIRC-related genes. We also discovered that the average degree of KIRC-related genes was the highest and showed that KIRC-related genes still play essential roles in the IIKDN network (Fig. 1f). The top five genes including TP53, SRC, GRB2, ESR1 and SMAD3 exhibit a direct connection and the highest degree in the network (Fig. 1g). Notably, SRC and GRB2 were immune-related genes. SMAD3 was a “three type” gene, and ESR1 was an inflammation-related and KIRC-related gene, which indicated that immune- and inflammation-related genes play hub roles in the IIKDN network. More than half of the interactions of KIRC-related genes were associated with immune- and inflammation-related genes (Fig. 1h). The number of other gene neighbours was smaller than that of immune- and inflammation-related gene neighbours for KIRC-related genes. All the above results indicated that immune- and inflammation-related genes play crucial roles in KIRC.
Immune- and inflammation-related genes directly interact with KIRC-related genes and identification of core clusters
We extracted a KIRC-related gene-directed network (KIRCD network) from the IIKDN network to further explore the association among immune-related, inflammation-related and KIRC-related genes (Fig. 2a). The KIRCD network only contained the KIRC-related genes and genes directly connected to them. There were 3360 nodes and 12,058 edges in the KIRCD network. We found 853 immune- and inflammation-related genes in the KIRCD network (Fig. 2b). Specially, we discovered that the nodes in the KIRCD network exhibited a higher network clustering, average degree and clustering coefficient than the nodes in the IIKDN network (Fig. 2c1–c3). This result of comparing topological features indicated that the KIRCD network is closer in structure.
Next, we further explored the associations among immune-, inflammation- and KIRC-related genes on the expression level based on the gene expression profile of KIRC patients. In the KIRCD network, there were 64.9% significant co-expressed interactions and 72.2% positive co-expressed interactions (Fig. 2d). We observed that the density curves for the PCC values showed classical bimodal distribution (Fig. 2e). The above results that we analysed in the KIRCD network indicated that the association among immune-related, inflammation-related and KIRC-related genes not only showed on the topological structure but also showed on the expression pattern. Next, we revealed the communications among immune-related, inflammation-related and KIRC-related genes by performing a module analysis. We extracted five core clusters with top scores (Fig. 2f1–f5). We discovered that the immune-, inflammation- and KIRC-related genes interacted in each cluster. These core clusters demonstrated the close relationships among the immune response, inflammation and KIRC.
Characteristics of expression pattern for core clusters to reveal the associations among immune-, inflammation- and KIRC-related genes
The core clusters showed close structure features, and we further considered if they also hold close communications on the expression level. First, we characterized the co-expression pattern of all genes in five core clusters. Most interactions in all five core clusters were significant co-expressed, and they indicated that most genes showed strong correlations (Fig. 3a). We also discovered that immune-, inflammation- and KIRC-related genes had higher PCC values when other genes were removed in the five clusters (Fig. 3b). We also identified differentially expressed genes in each core cluster based on the expression profile of KIRC and control samples after describing the co-expression pattern. We discovered that more than 80% genes in the core clusters showed significant differential expressions (Fig. 3c). In all five core clusters, there were more up-regulated genes than down-regulated genes (Fig. 3d). We further analysed the core clusters and found some strongly correlated gene pairs and functional modules. For example, in the first core cluster, there were 8 up-regulated and 1 down-regulated genes, and 81.8% genes were differentially expressed (Fig. 3e). The immune- and inflammation-related gene LAT and the immune-related gene ZAP70 showed significant differential expression in KIRC patients (P = 4.79e−55, 4.08e−65). LAT and ZAP70 showed strong positive correlation (PCC = 0.7, P < 0.001) and indicated that some immune- and inflammation-related genes play their roles by interacting in KIRC. In the second core cluster, most genes were differentially expressed (Fig. 3f). We also found a functional module including four immune-related genes, an inflammation-related and KIRC-related gene EGFR and an inflammation-related and immune-related gene LAT (Fig. 3f). Most of these six genes showed strong positive correlation, and this close interacting functional module may play an important role in KIRC. A similar phenomenon was observed in the fourth and fifth core clusters (Fig. 3g, h). In the fourth core cluster, three immune-related genes encoded complement C1q subcomponent subunit family genes including C1QA, C1QB and C1QC. The three genes all showed strong positive correlation (PCC = 0.98, 0.90 and 0.93) and formed a functional immune-related module in KIRC. All the above results revealed that the immune-, inflammation- and KIRC-related genes play their roles in KIRC by interacting within these five core clusters.
Core clusters showed complex genomic characteristics including somatic mutation, CNV and DNA methylation
Multi-dimensional genomic analysis promoted the depth of understanding for the associations among immune-, inflammation- and KIRC-related genes in core clusters. First, we discovered that most genes contained a certain number of somatic mutations in all core clusters (Fig. 4a1–a5). However, the numbers of somatic mutation in different genes were diverse. In addition, we found that the average number of somatic mutations in all clusters were almost similar (Fig. 4b). In all core clusters, the gene ERBB4 contained the highest number of somatic mutations, and other genes in the second cluster contained less somatic mutations. The mutation types on ERBB4 were complex, including missense mutation, intron mutation, silent mutation, frame shift insert and nonsense mutation (Fig. 4c). Specifically, ERBB4 interacted with the KIRC-related genes ERBB2, ERBB3 and MUC1 and the immune-related gene GRB2. Next, we discovered that certain genes contained CNV in the core clusters. For example, all genes in the first core cluster contained CNVs including amplification and deletion (Fig. 4d). PIK3R1 was a famous KIRC-related gene, and more than 120 samples exhibited CNV alterations in this gene. Other immune-related genes such as PTPN6 also contained some CNV alterations. Through somatic mutation and CNV analysis, we found that cancer genes usually contained more genomic alterations than immune- and inflammation-related genes. It could be inferred that immune- and inflammation-related genes could play a synergistic role with KIRC-related genes in KIRC.
Next, we also analysed the DNA methylation pattern for the genes in core clusters and found that most genes exhibited differential DNA methylation. For example, all genes in the first cluster had differential DNA methylation sites (Fig. 4e). Similar to somatic mutation and CNV, the KIRC-related gene ERBB2 exhibited the most differential DNA methylation sites. Notably, we discovered that the genes showed strong correlations with respect to the methylation level, similar to gene expression, and this indicated the interactions between immune, inflammation and KIRC-related genes and DNA methylation level (Fig. 4f). The network structure, expression pattern and DNA methylation pattern showed coincident correlation among immune-, inflammation- and KIRC-related genes in core clusters.
Core clusters in KIRC has prognostic potential
To evaluate the potential value of core clusters as prognostic biomarkers in KIRC, we created a risk-score formula according to the expression of all the genes in each core cluster to generate OS (overall survival) prediction (see “Methods” section). We used median risk score as the cut-off point to test the survival of the KIRC patients. We calculated the risk scores of all the genes in each cluster for each patient and then ranked the patients according to their risk score. Next, the KIRC patients would be divided into high-risk or low-risk groups. All five core clusters were significantly associated with survival, and they could serve as prognostic biomarkers (Fig. 5a1–a5). In addition, KIRC patients in the high-risk group exhibited a significantly shorter median OS than those in the low-risk group. KIRC patients could be grouped based on the risk score of these genes in the core clusters (Fig. 5c1–c5, d1–d5). The results indicated that immune-, inflammation- and KIRC-related genes could collectively influence KIRC patient survival and serve as specific prognostic biomarkers.
Core clusters associated with critical biological functions and the JAK/SAT signalling pathway
We performed GO enrichment analysis based on all the genes in five core clusters, respectively. We found that these genes were enriched in different GO terms (Fig. 6a, Additional file 1: Table S1). We found that the genes in the second, fourth and fifth core clusters were associated with certain immune-related GO terms such as regulation of the innate immune response, innate immune response activating cell-surface receptor signalling and negative regulation of immune system system. In addition, some protein modification related GO terms including protein phosphorylation and activation of protein kinase activity were discovered. Studies have reported the important role played by aberrant phosphorylation in oncogenesis and immune disorders [19].
Notably, we discovered that the genes in these five core clusters were all related to the JAK/SAT signalling pathway (Fig. 5b). The JAK/SAT signalling pathway is now recognized as an evolutionarily conserved signalling pathway employed by diverse cytokines, interferons, growth factors, and related molecules [20]. The changes in the pathway are functionally relevant in various human diseases, especially cancer and immune-related conditions [21]. Not only have genome-wide association studies demonstrated that the JAK/SAT signalling pathway is highly related to human autoimmunity but also targeting JAKs is now a reality in immune-mediated disease. Cytokine–cytokine receptor interaction is an important part of this pathway, and these cytokines are essential for immune and inflammatory responses [22]. In our analysis, we found that several genes in the five core clusters play essential roles in this pathway, which indicated that these key genes identified by us were highly associated with the immune response and inflammation.