Integrated analysis of transcriptomic and metabolomic data demonstrates the significant role of pyruvate carboxylase in the progression of ovarian cancer

Abstract

The aim of this study was to explore prognosis-related biomarkers and underlying mechanisms during ovarian carcinoma progression and development. mRNA expression profiles and GSE49997 dataset were downloaded. Survival analyses were performed for genes with high expression levels. Expression level of candidate genes was explored in four ovarian cancer cells lines. Pyruvate carboxylase (PC) was found to be one of significantly differentially expressed gene (DEG). The role of PC knockdown was analyzed in SKOV cells using cell proliferation, flow cytometric, and Transwell migration and invasion assays. DEGs and metabolites in PC-shRNA (shPC)-treated samples vs. control groups were identified. PC was a prognosis-related gene and related to metabolic pathway. Knockdown of PC regulated cell proliferation, cell cycle progression, and migration and invasion of SKOV-3 cells. Transcriptome sequencing analyses showed STAT1 and TP53 gained higher degrees in PPI network. A total of 44 metabolites were identified. These DEGs and metabolites in PC samples were related with neuroactive ligands receptor interaction, glycine, serine and threonine metabolism, and ABC transporter pathways. PC may affect the tumor biology of ovarian cancer through the dysregulation of glycine, serine, and threonine metabolism, and ABC transporter pathways, as well as STAT1 and TP53 expression.

Keywords: differentially expressed genes; epithelial ovarian cancer; metabolomics sequencing; prognosis; transcriptome sequencing.

Figure 1

Prognosis-related genes screened from TCGA and GSE49997 datasets. The gene expression profiles from TCGA and GEO databases were subjected to survival analyses. A total of 1153 and 1022 genes related with prognosis were obtained from TCGA and GSE49997 datasets, respectively. After Venn diagram analysis, 66 overlapping genes were identified as candidate genes related to EOC progression.

Figure 2

Survival curve for four candidate genes closely related with the prognosis of epithelial ovarian cancer. Results show that GGPS1, NTPCR (also known as C1orf57 or MGC13186), PC, and PPOX are significantly associated with the prognosis of epithelial ovarian cancer. The survival curves based on TCGA and GSE49997 datasets are listed, separately.

Figure 3

Knockdown of PC can significantly inhibit cell proliferation, cell cycle progression, and cell migration and invasion. (A) The relative expression of GGPS1, NTPCR, PPOX, PC, PRICKLE2, TCF7L1, and PPP3CA in four cancer cell lines (SKOV3, CAOV-3, OV-1063, and OVCAR-3). Gene expression levels of candidate genes (GGPS1, NTPCR, PPOX, PC, PRICKLE2, TCF7L1, and PPP3CA) were examined in different ovarian cancer cells lines using real-time qPCR analysis. (B) The relative expression of PC in SKOV-3 cell after PC knockdown using shRNAs by real-time qPCR analysis. (C) Decreased expression of PC can significantly inhibit SKOV-3 cell proliferation. (D) Cell cycle analysis of SKOV-3 cells following PC knockdown. The effect of PC on cell cycle progression was examined using flow cytometry. PC knockdown significantly inhibited cell cycle transition from G1 to S phase. (E) Cell migration and invasion of ovarian cancer cells after PC knockdown. The effects of PC on migration and invasion of SKOV3 cells were evaluated using the Transwell system. PC knockdown significantly inhibited the invasive and metastatic abilities of ovarian tumor cells.

Figure 4

Differentially expressed genes and pathways enrichment analysis after PC knockdown based on transcriptome sequencing. (A) Sample correlation based on differential gene expression. The correlation between samples was analyzed using Pearson’s correlation coefficient based on gene expression values. There were significant positive correlations between samples. (B) Principal component analysis results. The different colored dots represent the sample group under the condition. (C) Heatmap of differentially expressed genes between shPC and shNC samples. Two-dimension clustering analysis results were visualized using heatmaps for differentially expressed genes from PC-knockdown samples compared to the normal control group. The gene expression profiles were significantly different between groups. Red represents high expression levels while blue represents low expression levels. (D) KEGG pathway enrichment analysis for differentially expressed genes. The top ten pathway terms ranked by p-value were visualized using dot plot. The vertical axis represents KEGG pathways and the horizontal axis shows differentially expressed genes. A category with a smaller p-value represents a more significant difference. (E) Gene set enrichment analysis results. The red line refers to the highest enrichment score.

Figure 5

PPI analysis and miRNA regulatory networks. (A) PPI networks were constructed to visualize the relationships of differentially expressed genes screened from PC-knockdown samples compared to control samples. Red dots represent up-regulated genes and green dots refer to down-regulated genes. The points with a blue border refer to ovarian cancer-related genes. (B) miRNA regulatory network. A complex regulatory network was constructed to visualize connections of miRNAs, transcription factors, and genes related to ovarian cancer. The dots in red and green represent up- and down-regulated genes, respectively. Triangular nodes refer to miRNAs and v-shaped nodes represent transcription factors.

Figure 6

Data normalization and partial least squares discriminant analysis for metabolomics data. Two-dimensional principal component analysis (A) and three-dimensional partial least squares discriminant analysis (B) were conducted to ensure the detection stability of metabolomics data analysis. (C) Orthogonal partial least squares discriminant analysis by the X-Score model.

Figure 7

Metabolomics data analysis for the identification of potential metabolic pathways regulated by PC. (A) Heatmap of differential metabolites in shPC vs. shNC groups in positive mode and negative mode. A bidirectional clustering analysis heat map was used to visualize metabolite levels in shPC vs. shNC samples under positive (left) and negative modes (right). The horizontal and vertical axes represent the samples and metabolites, respectively. Green represents down-regulated levels while red refers to up-regulated levels. (B) Volcano map of differential metabolites under positive (left) and negative (right) modes in shPC vs. shNC samples. The points in pink, blue, and grey refer to metabolites with up, down, and normal regulated levels, respectively. The dot size represents the VIP value. (C) Functional enrichment analysis for differential metabolites. The vertical and horizontal axes represent pathway categories and count number, respectively. The dot size represents the metabolite ratio of pathway enrichment. The color changes from blue to red refer to decreasing p-values. A dot with a smaller p-value represents a more significant difference for the pathway category. (D) Integrated analysis of transcriptomic and metabolomics data to identify crucial pathways regulated by PC.