Downregulation of Methionine Cycle Genes MAT1A and GNMT Enriches Protein-Associated Translation Process and Worsens Hepatocellular Carcinoma Prognosis

Abstract

The major biological methyl donor, S-adenosylmethionine (adoMet) synthesis occurs mainly in the liver. Methionine adenosyltransferase 1A (MAT1A) and glycine N-methyltransferase (GNMT) are two key enzymes involved in the functional implications of that variation. We collected 42 RNA-seq data from paired hepatocellular carcinoma (HCC) and its adjacent normal liver tissue from the Cancer Genome Atlas (TCGA). There was no mutation found in MAT1A or GNMT RNA in the 42 HCC patients. The 11,799 genes were annotated in the RNA-Seq data, and their expression levels were used to investigate the phenotypes of low MAT1A and low GNMT by Gene Set Enrichment Analysis (GSEA). The REACTOME_TRANSLATION gene set was enriched and visualized in a heatmap along with corresponding differences in gene expression between low MAT1A versus high MAT1A and low GNMT versus high GNMT. We identified 43 genes of the REACTOME_TRANSLATION gene set that are powerful prognosis factors in HCC. The significantly predicted genes were referred into eukaryotic translation initiation (EIF3B, EIF3K), eukaryotic translation elongation (EEF1D), and ribosomal proteins (RPs). Cell models expressing various MAT1A and GNMT proved that simultaneous restoring the expression of MAT1A and GNMT decreased cell proliferation, invasion, as well as the REACTOME_TRANSLATION gene EEF1D, consistent with a better prognosis in human HCC. We demonstrated new findings that downregulation or defect in MAT1A and GNMT genes can enrich the protein-associated translation process that may account for poor HCC prognosis. This is the first study demonstrated that MAT1A and GNMT, the 2 key enzymes involved in methionine cycle, could attenuate the function of ribosome translation. We propose a potential novel mechanism by which the diminished GNMT and MAT1A expression may confer poor prognosis for HCC.

Keywords: GNMT; MAT1A; human hepatocellular carcinoma; methionine cycle.

Figure 1

The expressions of MAT1A, GNMT, and MAT2A in HCC. Gene expression profiling interactive analysis (GEPIA) GEPIA2 database was employed for bioinformatics analysis. (A–C) MAT1A, GNMT, and MAT2A expression in paired tumor and adjacent normal tissue. *, p < 0.05. (D) Pearson’s correlation was used to elucidate MAT1A expression in relation to GNMT expression. (E,F) Kaplan–Meier plots showed that high expression of MAT1A and GNMT (F) are both associated with better overall survival rate of patients with HCC. HR, hazard ratio (G) Kaplan–Meier plot showed that elevated MAT2A expression is associated poor overall survival rate of patients with HCC. HR, hazard ratio.

Figure 2

Searching for the potential biological processes associated with low expression of MAT1A and GNMT. (A) MAT1A and GNMT expressions in paired tumor and adjacent normal tissues in the HCC cases (n = 42). (B) Pearson’s correlation for MAT1A and GNMT expression in the 42 HCC cases. (C) Flow chart for searching the candidate biological processes that could be driven by low expression of MAT1A and GNMT; GSEA was used to process the 42 HCC RNA-seq data. REACTOME_TRANSLATION gene set was found to be significantly enriched in HCC with low MAT1A. The median of mRNA expression level was used as the cutoff point for the dichotomization of MAT1A and GNMT. A score greater than median was defined as “high” expression, whereas a score of less than or equal to median was defined as “low” expression, and the categories further were mapped into GSEA databases. (D) and in HCC with low GNMT (E). (F) Heatmaps of the REACTOME_ TRANSLATION enrichment genes in HCC with high and low MAT1A and GNMT categories (G) A total of 71 overlapped genes of the REACTOME_ TRANSLATION gene set that are enriched in both low MAT1A and low GNMT HCC samples.

Figure 3

Kaplan–Meier curves for the 43 of REACTOME_TRANSLATION genes.

Figure 3

Kaplan–Meier curves for the 43 of REACTOME_TRANSLATION genes.

Figure 4

MAT1A and GNMT expression decreased cell proliferation and invasion in HepG2 cells. MAT1A and GNMT expression also decreased reactome protein EEF1D, and EIF3B mRNA levels. (A) Doubling time of 4 HepG2 derived stable cell lines with various MAT1A and GNMT-expression levels were determined by cell counting. Points, mean; bars, SE. A: Control cells. B: Clone #1, MAT1A and GNMT-overexpressing cells. C: Clone #2, MAT1A and GNMT-overexpressing cells. (B) Overexpression of MAT1A and GNMT decreases invasive ability of HCC cells. MAT1A, GNMT, and β-actin (loading control) protein expression levels were evaluated by immunoblotting in HepG2 and HepG2/MAT1A/GNMT. *, p < 0.05. (C) List of quantitative PCR primers. (D) Pearson’s correlation matrix of GNMT, MAT1A, MAT2A, EEF1D, and EIF3B mRNA expression. (E) Linear plot of correlation estimated from doubling time and EEF1D mRNA expression.

Figure 5

The correlations between MAT1A and GNMT with reactome genes in liver hepatocellular carcinoma patients using the Cancer Genome Atlas (TCGA) web server program. Pearson’s correlations showed that MAT1A mRNA levels inversely correlated with (A) EIF3B and (B) EEF1D; GNMT mRNA levels inversely correlated with EIF3B (C) and EEF1D (D); MAT1A (E) and GNMT expressions are positively correlated.

Figure 6

Promoter and gene body methylation status of EEF1D.