Over-expression of different KMT2s family members in patients with GC
To examine the potential correlation between the different KMT2s members and GC, we compared the TCGA data by using GEPIA database (http://gepia.cancer-pku.cn/) to explore the expression of KMT2s in cancer tissues and the normal tissues. We found that transcriptional levels of KMT2A/B/C/D had a trend to be higher in cancer tissues than the normal tissues, while transcriptional levels of KMT2E/F were significantly higher in cancer tissues (Figure 1A-B).
Having examined the mRNA expression patterns of KMT2s in GC, we tried to explore the protein expression patterns of KMT2s in GC by the Human Protein Atlas. As showed in Figure 2A-F, the expressions of KMT2A/B/C/D/E/F proteins were also higher in cancer tissues than the normal tissues. In summary, our results showed that transcriptional and proteomic expressions of KMT2s were overexpressed in patients with GC.
Correlation between mRNA expressions of different KMT2s family members with the clinicopathological parameters of patients with GC
Relationship between mRNA expressions of different KMT2s with clinicopathological features of GC patients were analyzed by using UALCAN (http://ualcan.path.uab.edu) including patients’ individual cancer stages and tumor grades. Our results showed that KMT2A/B/C/D/E/F mRNA expressions were significantly related with patients’ individual cancer stages, and patients with advanced stages or grades tended to express higher mRNA of KMT2s. As showed in Figure 3, the highest mRNA expression of different KMT2s were found in stage 3. The reason why mRNA expressions of KMT2A/B/C/D/E/F in stage 3 seemed to be higher than that in stage 4 may be due to the small sample size. Consistent with the analysis of patients’ individual cancer stages, our work found that mRNA expression of different KMT2s family members were also correlated with tumor grades. As tumor grades increased, the expression levels of KMT2s mRNA tended to be higher. As showed in Figure 4 A-F, the highest expression of KMT2A/B/D/F were all found in grade 3, while the highest expression of KMT2C/E were in grade 2. Taken together, the results above indicated that mRNA expressions of 6 KMT2s family members were significantly related with clinicopathological characteristics in GC patients.
Correlation between methylation of KMT2s with its expression and clinical data in GC
DNA methylation is a common heritable epigenetic modification in the genome of eukaryotic cells, plays an important role in regulating cell proliferation, differentiation, and individual development, and abnormal DNA methylation levels are closely related to tumor development. We used MEXPRESS to analyze the correlation between methylation of KMT2s with its expression and clinical data in GC. As showed in Figure 5A, KMT2A expression was positively related with KMT2A promoter methylation (r=0.448***), tumor stage (p=0.01) and histological type (p=4.311e-5). In Figure 5B, KMT2B promoter methylation was positively related with KMT2B expression (r=0.479***), tumor stage (p=0.002) and histological type (p=0.002). In Figure 5C, KMT2C promoter methylation was positively related with KMT2C expression (r=0.429***), tumor stage (p=0.018). In Figure 5D, KMT2D promoter methylation was positively related with KMT2D expression (r=0.303***). In Figure 5E, KMT2CE promoter methylation was positively related with KMT2E expression (r=0.403***), tumor stage (p=0.014). In Figure 5F, KMT2F (SETD1A) promoter methylation was positively related with KMT2E expression (r=0.357***), tumor stage (p=0.021) and histological type (p=0.047). These findings indicated that high promoter methylation levels of KMT2A/B/C/D/E/F contributed normal expression of oncogenes (KMT2A/B/C/D/E/F) to induce the development and progression of GC.
Prognostic values of expression levels of KMT2s mRNA in patients with GC
We further explore the critical efficiency of KMT2s family members in the survival of patients with GC. Kaplan-Meier plotter (http://kmplot.com/analysis/) was performed to analyze the association between KMT2s mRNA expressions with the survival including overall survive (OS), progression-free survival (FP) and post-progression survival (PPS). The Kaplan-Meier curve and log rank test analyses revealed that patients of GC with higher of KMT2A/B/C/D/E/F mRNA expression levels have a shorter OS, FP and PPS in 150 months’ follow up (Figure 6), P < 0.05. Thus, our results demonstrated that increased KMT2A/B/C/D/E/F expression were predicted to be associated with poor prognosis in GC.
Genetic alteration of KMT2s family members and the network in GC
We used the cBioPortal online tools (https://www.cbioportal.org/) to analyze the KMT2s alterations, correlations and networks based on sequencing data from GC patients in the TCGA database. KMT2s were altered in 117 of 293 (40%) GC patients. These alterations of KMT2A/B/C/D/E/F were in the following (Table 1): KMT2A: amplification (2%), mutation (10.6%); KMT2B: amplification (2.4%), homozygous (1.7%) and mutation (7.5%); KMT2C: amplification (0.7%), homozygous (1.7%) and mutation (15.4%); KMT2D: amplification (0.3%), homozygous (1%) and mutation (21.8%); KMT2E: amplification (2.7%), homozygous (0.7%) and mutation (5.8%); KMT2F: amplification (0.3%) and mutation (5.5%).We also calculated the co-expression of KMT2s with each other for stomach adenocarcinoma, and Pearson’s correction was included. We found significant and positive correlations in the followings KMT2s:KMT2A with KMT2C, KMT2D and KMT2E; KMT2B with KMT2D and KMT2F; KMT2C with KMT2A and KMT2E; KMT2D with KMT2A, KMT2B and KMT2F; KMT2E with KMT2A and KMT2C; KMT2F with KMT2B and KMT2D (Figure 7C). Next, we constructed the network for KMT2s and the neighboring genes with the alteration frequency >5% (Figure 7D).
Predicted functions and pathways of the changes in KMT2s and their frequently altered neighboring genes in patients with GC
The functions of KMT2s and the genes significant correlated with KMT2s alterations were predicted by analyzing GO and KEGG pathways in the database for STRING (https://string-db.org/). To address this, we used the KMT2s-neighboring genes that were altered at frequencies > 5% in the GC TCGA database (Figure 7D and Table 1). As showed in Figure 8, analysis of significantly enriched GO terms suggested that those proteins were mainly located in (GO:0005654) nucleoplasm, (GO:0005694) chromosome, (GO:0035097) histone methyltransferase complex, (GO:0043231) intracellular membrane-bounded organelle, (GO:0044451) nucleoplasm part and (GO:0043232) intracellular non-membrane-bounded organelle. These proteins are primarily involved in (GO:0034968) histone lysine methylation, (GO:0006325) chromatin organization, (GO:0018193) peptidyl-amino acid modification, (GO:0006355) regulation of transcription, DNA-templated and (GO:0051252) regulation of RNA metabolic process. while they also serve as (GO:0018024) histone-lysine N-methyltransferase activity, (GO:0042800) histone methyltransferase activity (H3-K4 specific), (GO:0140100) transferase activity, (GO:0140110) transcription regulator activity and (GO:0003677) DNA binding. Moreover, KEGG pathway analysis showed enrichment in (hsa00310) lysine degradation, (hsa05226) gastric cancer such as MAPK and PI3K/AKT pathways, (hsa05210) colorectal cancer, (hsa04550) signaling pathways regulating pluripotency of stem cells and (hsa04934) Cushing’s syndrome. Among these significant pathways, pathways of lysine degradation and gastric cancer were regulated by KMT2s in GC (Figure 9A-B).
KMT2s networks of kinase, miRNA and transcription factor targets in GC
To further predict the targets of KMT2s in GC, we analyzed the kinase, miRNA and transcription factor targets generated by Gene Set Enrichment Analysis (GSEA). The top 5 most significant target networks of KMT2A/B/C/D/E/F were included in Table 2, respectively. Importantly, we found that all of the 6 KMT2s had common kinas targets: AKT and MAPK1, which suggested KMT2s might be involved in pathways such as MAPK and PI3K/AKT signaling pathway.