Expression of PRDXs in hepatocellular carcinoma in different databases
mRNA levels of PRDXs in various common cancer cell lines were obtained from CCLE database. mRNA levels of PRDXs were found to be maintained at 6-9, relatively higher compared to most other cancer cell lines (Figure 1 and Supplementary figure 1-2). The expression levels of PRDXs in normal liver and HCC tissues were furtherassessed using UALCAN database (Figure 2). PRDX1 (Figure 2A), PRDX2 (Figure 2B), PRDX5 (Figure 2E) and PRDX6 (Figure 2F) were significantly upregulated in liver cancer, while PRDX4 (Figure 2D) was markedly downregulated in HCC. Besides, there was no difference in PRDX3 expression between HCC tissues and normal liver tissues (Figure 2C).
What’s more, the mRNA expression of PRDXs in HCC, adjacent normal tissue, cirrhotic and healthy samples was obtained from HCCDB database. PRDX1 was verified to be upregulated in HCC tissues compared with para-carcinoma tissues in all HCC datasets but HCCDB11 (Figure 3A), and PRDX5 was also upregulated in HCC tissues in all datasets except for HCCDB11 and HCCDB16 (Figure 3B). PRDX2 was attested to have higher expression levels in HCC tissues in HCCDB7 and HCCDB18 datasets (Supplementary figure 3A). Moreover, the downregulation of both PRDX3 and PRDX4 in HCC tissues was illustrated by HCCDB1, HCCDB3, HCCDB4, HCCDB13, HCCDB15 and HCCDB16 dataset. Besides, PRDX3 was also downregulated in HCCDB6 (Supplementary figure 3B), and PRDX4 was downregulated in HCCDB12 (Supplementary figure 4A). Similarly, results of HCCDB1, HCCDB4, HCCDB6, HCCDB15 and HCCDB17 datasets indicated the downregulated expression levels of PRDX6 in HCC, and data of HCCDB7 dataset exhibited conversed results (Supplementary figure 4B).
The protein levels of PRDXs gene family in hepatocellular carcinoma and the correlation between PRDXs expression and clinical characteristics
The protein expression of PRDXs was collected from the immumohistochemical staining results of Human Protein Atlas database. It was obvious that the protein levels of PRDX1 (Figure 4A), PRDX2 (Figure 4B) and PRDX5 (Figure 4E) were significantly upregulated in HCC tissues, while that of PRDX3 (Figure 4C) and PRDX4 (Figure 4D) was downregulated. Moreover, there was no differential expression of PRDX6 between HCC tissues and normal liver tissues (Figure 4F). The correlation analysis between expression of PRDXs and age, cancer stage, gender, and tumor grade was further performed based on the data of TCGA database by UALCAN. Except for the upregulation of PRDXs in HCC patients compared with normal people, the significant association between expression of some members of PRDX gene family and age (Supplementary table 1), cancer stage (Supplementary table 2), gender (Supplementary table 3) and tumor grade (Supplementary table 4) can be observed. In addition, correlation analysis between the methylation levels of PRDXs in HCC and cancer stage as well as tumor grade was also carried out. The results showed that the methylation levels of PRDX1 and PRDX3 in HCC patients were obviously higher than that in normal people, whereas that of PRDX2/4/5 were lower (Supplementary table 5). Also, several significant correlation between methylation levels of PRDXs and cancer stage (Supplementary table 6) or tumor grade (Supplementary table 7) can be observed.
mRNA expression of PRDXs and its association with the overall survival (OS) of patients with HCC
According to the analysis results of Kaplan-Meier Plotter database, overexpression of PRDX1 was associated with a poor prognosis for patients with HCC (HR = 1.63) (Figure 5A). However, the lowly expressed of PRDX2 (Figure 5B) and PRDX3 (Figure 5C) led to the poor prognosis value of patients with HCC. The expression of other PRDX family members (PRDX4/5/6) was not significantly correlated with the survival of HCC patients (Figure 5D-F).
The genomic alterations of PRDXs in HCC
The frequency and types of PRDXs alterations in HCC based on sequencing data from patients with HCC in the TCGA database were monitored by cBioPortal (Figure 6A). The results revealed that PRDX1 and PRDX4 were both altered in 26 of 372 (7%) patients with HCC. Besides, both of PRDX2 and PRDX5 were altered in 33 of 372 (9%) patients with HCC. PRDX3 was altered in 40 of 372 (11%) HCC patients, and 18% (63/372) of HCC patients exhibited PRDX6 alteration. Moreover, the specific types and frequency of PRDXs alterations in HCC showed in Table 1. These alterations were mRNA upregulation in 24 cases (6.4%) of PRDX1, 30 cases (8.1%) of PRDX2, 19 cases (5%) of PRDX3, 31 cases (8.3%) of PRDX5, and 33 cases (8.9%) of PRDX6. Furthermore, the tab Network in cBioPortal was used to reflect interactions between PRDXs and the neighborhood genes in HCC (Figure 6B). The neighbor genes of PRDXs with the most frequent alterations were MYC (18.6%), MAPK1 (13.1%) and PBK (11.4%) (Table 1).
KEGG pathway analyses of co-expression genes correlated with PRDXs in HCC
The 372 HCC patients in the TCGA were used to analyze mRNA sequencing data by the LinkFinder module of LinkedOmics. As shown by the volcano plot in Figure 7A, 11082 genes showed negative correlation with PRDX1 and a total of 8840 genes were positive correlation with PRDX1. The top 50 of positive correlated significant genes presented in Figure 7B by heat map, whereas the top 50 of negative correlated significant genes showed in Figure 7C by heat map. There was a strong negative correlation between the expression of PHF2 and PRDX1 (Pearson correlation = 0.62, P = 2.547e-40), whereas PSMB2 was strongly positively correlated with PRDX1 (Pearson correlation = 0.72, P = 1.961e-59). In addition, the specific analysis results of genes that negatively or positively correlated with PRDX family members were shown in panels A-C of Supplementary figure 5-9. Combining all of the above results, it indicated that PRDXs had extensive influence on the transcriptome. Furthermore, the significant KEGG pathway analysis of genes differentially expression in correlation with PRDX1 by gene set enrichment analysis (GSEA) revealed enrichment in Metabolic pathways, TGF-beta signaling pathway, and so on (Figure 7D). It suggested that the enriched pathways of genes differentially expression in correlation with PRDX3, PRDX4 and PRDX5 were also involved in Metabolic pathways (panels D of Supplementary figure 6-8). To further investigate the miRNA targets of PRDX1 in HCC, the LinkInterpreter of LinkedOmics were utilized, and the results revealed that the network was associated with (GTCTTCC) MIR-7, (CTTTGTA) MIR-524, (TTGCACT) MIR-130A, MIR-301, MIR-130B (Figure 7E). Furthermore, miRNA targets networks of the other members of PRDX family were also analyzed and showed in panels E of Supplementary figure 5-9.