Different Approaches for the Profiling of Cancer Pathway-Related Genes in Glioblastoma Cells

Abstract

Deregulation of signalling pathways that regulate cell growth, survival, metabolism, and migration can frequently lead to the progression of cancer. Brain tumours are a large group of malignancies characterised by inter- and intratumoral heterogeneity, with glioblastoma (GBM) being the most aggressive and fatal. The present study aimed to characterise the expression of cancer pathway-related genes (n = 84) in glial tumour cell lines (A172, SW1088, and T98G). The transcriptomic data obtained by the qRT-PCR method were compared to different control groups, and the most appropriate control for subsequent interpretation of the obtained results was chosen. We analysed three widely used control groups (non-glioma cells) in glioblastoma research: Human Dermal Fibroblasts (HDFa), Normal Human Astrocytes (NHA), and commercially available mRNAs extracted from healthy human brain tissues (hRNA). The gene expression profiles of individual glioblastoma cell lines may vary due to the selection of a different control group to correlate with. Moreover, we present the original multicriterial decision making (MCDM) for the possible characterization of gene expression profiles. We observed deregulation of 75 genes out of 78 tested in the A172 cell line, while T98G and SW1088 cells exhibited changes in 72 genes. By comparing the delta cycle threshold value of the tumour groups to the mean value of the three controls, only changes in the expression of 26 genes belonging to the following pathways were identified: angiogenesis FGF2; apoptosis APAF1, CFLAR, XIAP; cellular senescence BM1, ETS2, IGFBP5, IGFBP7, SOD1, TBX2; DNA damage and repair ERCC5, PPP1R15A; epithelial to mesenchymal transition SNAI3, SOX10; hypoxia ADM, ARNT, LDHA; metabolism ATP5A1, COX5A, CPT2, PFKL, UQCRFS1; telomeres and telomerase PINX1, TINF2, TNKS, and TNKS2. We identified a human astrocyte cell line and normal human brain tissue as the appropriate control group for an in vitro model, despite the small sample size. A different method of assessing gene expression levels produced the same disparities, highlighting the need for caution when interpreting the accuracy of tumorigenesis markers.

Keywords: cancer pathway; glioblastoma; mRNA; multicriterial analysis.

Figure 1

The mean of Ct ratio of mRNA levels. (a) The mRNA expression levels in control group. (b) The mRNA expression levels in group of tumour cell lines. Ct ratio, gene of interest/housekeeping genes. Each mRNA level included three replicates.

Figure 2

Scatter plot of principal component analysis. The grey elliptical line represents data of tumour cell lines, dots outside the line are control samples. All values of each sample were measured in triplicates.

Figure 3

Pearson correlation of individual genes expressions of control (a) and tumour (b) cell lines.

Figure 4

Statistical analysis of gene expression correlations. (a) Descending order of correlations. (b) Kernel density of correlations. The green line represents data from control, red line from tumour and grey line is the mix of both groups.

Figure 5

Pearson correlations between gene expression profiles of analysed cell lines. (a) Correlation of gene expression profiles within control group, (b) within tumour group and (c) between samples of control and tumour group.

Figure 6

Heat map of fold change regulation between test and control group. Test group (TG): A172, SW1088 and T98G; Control group (CG): HDFa, hRNA and NHA.

Figure 7

Multi-criterial correlation plots of gene expression in tumour cell lines relative to the controls. R⁺ represents a negative proportion of the numerator’s dissimilarity tendencies, R⁻ positive. A 45-degree line separates the regions of overexpression and underexpression; at this line, overexpression and underexpression are therefore balanced. The scenarios when noise predominates are covered by the (0.1) × (0.1) square. There are three main regions in the graph to categorize according to the level of gene expression with three ways of marking the corresponding points: statistically significant and overexpressed with R⁺ > 1 (red circles), statistically significant underexpressed with R⁻ > 1 (blue triangles), finally statistically less significant (black squares) bordered by 0 ≤ R⁺ ≤ 1, 0 ≤ R⁻ ≤ 1.