ADH family members were down-regulated in human HCC
Previous study assessed the gene expression of ADH family members in TCGA data, and demonstrated that ADH1A, ADH1B, ADH1C, ADH4, and ADH6 expression were significantly decreased in HCC tissues compared with normal controls [10]. In the present study, we further analyzed the transcriptome expression level of all ADH genes in 20 liver cancer and 12 tumor adjacent normal tissues by employing GEO dataset. Consistently, we observed a remarkable reduction in ADH1A, ADH1B, ADH4, and ADH6 expression in HCC as compared to normal tissues (Figure 1). Of note, there were no statistically significant differences in the expression of ADH1C and ADH5 between the two groups (Figure 1A).
Correlation between ADHs gene expression and methylation
In attempt to elucidate the mechanism underlying the modulated expression of different ADH genes in HCC, we first evaluated DNA methylation level of all ADH genes from 41 normal and 374 HCC in TCGA. As shown in Figure 2, the methylation level of ADH1A, ADH1B, ADH1C, ADH4 and ADH6 were negatively correlated with their gene expression. Moreover, no significant association between ADH5 expression and methylation was detected (Figure 1B). It has been illustrated that multiple methylated CpG sites located in promoter region could induce transcriptional silencing of genes [11]. We found that cg24368912 within the sequence of ADH1B promoter, and cg09112717, cg13256891, cg15443145, cg21548116 and cg25073725 in ADH5 promoter had significantly higher methylated levels in HCC than normal controls (Figure 2). Conversely, DNA methylation levels at cg03806087, cg23949936 in ADH1A promoter, cg00689360 in ADH1C promoter, cg12011299 in ADH4 promoter, and cg00268009 in ADH5 promoter were evidently reduced in HCC as compared with healthy samples (Figure 2).
Correlation between ADHs gene expression and gene copy number
Being one of the predominant forms of genetic alterations, copy number variation has been documented to play a vital role in regulating gene expression during tumor progression [12]. Accordingly, we further performed an analysis of the gene level focal segment copy number derived from 190 HCC cases in TCGA data. However, we did not observe significant alterations of gene copy number for the gene expression of ADH1A, ADH1B, ADH1C, ADH4, and ADH6 in HCC (Figure 3). In contrast, ADH5 expression was markedly downregulated with a reduced gene copy number (Figure 3). Thus, our data implicated that copy number variation might not apparently affect the gene expression level of ADH genes in HCC subjects.
Identification of the potential TFs that regulate ADH gene expression
To better clarify other regulatory factors for ADH gene expression, we conducted the prediction of potential TFs, which are proteins that participate in transcriptional regulation of genes through binding specific DNA sequences [13]. From RegNetwork database, SRF, NFYA and HNF4A were identified as TFs that target ADH1B, ADH1C and ADH4 respectively (Figure 4A). Interestingly, five TFs (TFAP2A, CEBPB, E2F1, SP1 and NFIC) were all recognized as regulators of ADH6 gene expression (Figure 5A). Next, we ascertained the gene expression levels of these predictive TFs in HCC tissues in contrast with normal tissues. Remarkably, all the predictive TFs were up-regulated in HCC except CEBPB (Figure 4B). Furthermore, we examined the correlation between above TFs and overall survival. The results of Kaplan-Meier analysis revealed that HCC patients with high expression levels of NFYA, E2F1 and TFAP2A were strongly associated with shortened overall survival (Figure 5).
Identification of the potential miRNAs that regulate ADH gene expression
As small non-coding RNA molecules, miRNAs post-transcriptionally modulate gene expression [14]. Thus, we also predicted the potential miRNAs that were involved in ADH gene regulation by employing RegNetwork database. miR-377 and miR-944 were discovered to respectively impact ADH1B and ADH5 expression (Figure 6A). Notably, we found that ADH4 was mediated by 7 miRNAs, including miR-181c, miR-185, miR-376c, miR-410, miR-561, miR-657 and miR384 (Figure 6A). In addition, ADH6 was observed to be regulated by miR-105, miR-299-3p, miR-498 and miR-522 (Figure 6A). To understand the effects of the predictive miRNAs exerted on HCC progression, we initially measured their expression levels. We noticed that the expression of miR-377, miR-376c, miR-410, miR-299-3p were evidently suppressed in HCC subjects in comparison with normal controls (Figure 6B). On the contrary, miR-185 and miR-105 expression were significantly enhanced in HCC (Figure 6B). Further analysis of the association between miRNAs expression and survival time indicated that high expression of miR-561 and miR-498 were both predictors of poor overall survival in HCC patients (Figure 7).