Clinical significance of long non-coding RNA DUXAP8 and its protein coding genes in hepatocellular carcinoma

Abstract

Backgrounds: Hepatocellular carcinoma (HCC) is a lethal malignancy worldwide that is difficult to diagnose during the early stages and its tumors are recurrent. Long non-coding RNAs (lncRNAs) have increasingly been associated with tumor biomarkers for diagnosis and prognosis. This study attempts to explore the potential clinical significance of lncRNA DUXAP8 and its co-expression related protein coding genes (PCGs) for HCC. Method: Data from a total of 370 HCC patients from The Cancer Genome Atlas were utilized for the analysis. DUXAP8 and its top 10 PCGs were explored for their diagnostic and prognostic implications for HCC. A risk score model and nomogram were constructed for prognosis prediction using prognosis-related genes and DUXAP8. Molecular mechanisms of DUXAP8 and its PCGs involved in HCC initiation and progression were investigated. Then, potential target drugs were identified using genome-wide DUXAP8-related differentially expressed genes in a Connectivity Map database. Results: The top 10 PCGs were identified as: RNF2, MAGEA1, GABRA3, MKRN3, FAM133A, MAGEA3, CNTNAP4, MAGEA6, MALRD1, and DGKI. Diagnostic analysis indicated that DUXAP8, MEGEA1, MKRN3, and DGKI show diagnostic implications (all area under curves ≥0.7, p≤0.05). Prognostic analysis indicated that DUXAP8 and RNF2 had prognostic implications for HCC (adjusted p=0.014 and 0.008, respectively). The risk score model and nomogram showed an advantage for prognosis prediction. A total of 3 target drugs were determined: cinchonine, bumetanide and amiprilose and they may serve as potential therapeutic targets for HCC. Conclusion: Functioning as an oncogene, DUXAP8 is overexpressed in tumor tissue and may serve as both a diagnostic and prognosis biomarker for HCC. MEGEA1, MKRN3, and DGKI maybe potential diagnostic biomarkers and DGKI may also be potentially prognostic biomarkers for HCC.

Keywords: DUXAP8; hepatocellular carcinoma; long non-coding RNA; molecular mechanism; protein-coding gene.

Figure 1

Body map of the expressions of DUXAP8 and its co-expression-related protein-coding genes. A-K: Body map of the expressions of DUXAP8, RNF2, MAGEA1, GABRA3, MKRN3, FAM133A, MAGEA3, CNTNAP4, MAGEA6, MALRD1, and DGKI, respectively.

Figure 2

Scatter plots of DUXAP8 and its co-expression-related protein-coding genes in tumor and non-tumor tissues. A-K: Scatter plots of DUXAP8, RNF2, MAGEA1, GABRA3, MKRN3, FAM133A, MAGEA3, CNTNAP4, MAGEA6, MALRD1, and DGKI, respectively.

Figure 3

Diagnostic receiver operator curves of DUXAP8 and its co-expression-related protein-coding genes. A-K: Diagnostic receiver operator curves of DUXAP8, RNF2, MAGEA1, GABRA3, MKRN3, FAM133A, MAGEA3, CNTNAP4, MAGEA6, MALRD1, and DGKI, respectively.

Figure 4

Joint-effect analysis of diagnostic receiver operator curves of DUXAP8 and diagnosis-related genes. A-F: Diagnostic receiver operator curves of DUXAP8 and MAGEA1; DUXAP8 and MKRN3; DUXAP8 and DGKI; MAGEA1 and MKRN3; MAGEA1 and DGKI; and MKRN3 and DGKI, respectively.

Figure 5

Kaplan-Meier plots of DUXAP8 and its co-expression-related protein-coding genes and joint-effect analysis of DUXAP8 and RNF2. A-L: Kaplan-Meier plots of DUXAP8, RNF2, MAGEA1, GABRA3, MKRN3, FAM133A, MAGEA3, CNTNAP4, MAGEA6, MALRD1, DGKI, and joint-effect analysis plot of DUXAP8 and RNF2, respectively.

Figure 6

Risk score model, Kaplan-Meier plot, and time-dependent ROC curves. A: risk score model constructed using risk scores, patient survival status, DUXAP8 and RNF2 expression heat maps; B: Kaplan-Meier plot of low and high risk groups; C: Time-dependent ROC curves of 1, 3, and 5year OS.

Figure 7

Nomogram and co-expression network of DUXAP8 and co-expression-related protein-coding genes. A: Nomogram constructed using DUXAP8, RNF2, tumor stage, radical resection, and hepatitis B virus infection status; B: Co-expression network between DUXAP8 and the co-expression-related protein-coding genes.

Figure 8

Gene set enrichment analysis of DUXAP8 of gene ontologies and KEGG pathways. A-I: Gene ontology results of DUXAP8; J-L: KEGG pathway results of DUXAP8.

Figure 9

Gene set enrichment analysis of RNF2 of gene ontologies and KEGG pathways. A-I: Gene ontology results of RNF2; J-L: KEGG pathway results of RNF2.

Figure 10

Co-expression matrix and gene-gene interaction and protein-protein interaction networks of LINC00668 and co-expression-related protein-coding genes. A: Co-expression matrix among DUXAP8, RNF2, MAGEA1, GABRA3, MKRN3, FAM133A, MAGEA3, CNTNAP4, MAGEA6, MALRD1, and DGKI; B: Co-expression network of gene-gene interactions among protein-coding genes; C: Protein-protein interaction network among RNF2, MAGEA1, GABRA3, MKRN3, FAM133A, MAGEA3, CNTNAP4, MAGEA6, MALRD1, and DGKI.

Figure 11

Heatmap and volcano plot of differentially expressed genes of DUXAP8. A: Heatmap of differentially expressed genes of DUXAP8; B: Volcano plot of differentially expressed genes of DUXAP8.

Figure 12

2D and 3D structure of the chemical compound of the 3 target drugs. A-C: 2D structure of cinchonine, bumetanide and amiprilose, respectively; D-F: 3D structure of cinchonine, bumetanide and amiprilose, respectively.

Figure 13

Enriched gene ontology terms network using differentially expressed genes.