Investigating the different mechanisms of genotoxic and non-genotoxic carcinogens by a gene set analysis

Abstract

Based on the process of carcinogenesis, carcinogens are classified as either genotoxic or non-genotoxic. In contrast to non-genotoxic carcinogens, many genotoxic carcinogens have been reported to cause tumor in carcinogenic bioassays in animals. Thus evaluating the genotoxicity potential of chemicals is important to discriminate genotoxic from non-genotoxic carcinogens for health care and pharmaceutical industry safety. Additionally, investigating the difference between the mechanisms of genotoxic and non-genotoxic carcinogens could provide the foundation for a mechanism-based classification for unknown compounds. In this study, we investigated the gene expression of HepG2 cells treated with genotoxic or non-genotoxic carcinogens and compared their mechanisms of action. To enhance our understanding of the differences in the mechanisms of genotoxic and non-genotoxic carcinogens, we implemented a gene set analysis using 12 compounds for the training set (12, 24, 48 h) and validated significant gene sets using 22 compounds for the test set (24, 48 h). For a direct biological translation, we conducted a gene set analysis using Globaltest and selected significant gene sets. To validate the results, training and test compounds were predicted by the significant gene sets using a prediction analysis for microarrays (PAM). Finally, we obtained 6 gene sets, including sets enriched for genes involved in the adherens junction, bladder cancer, p53 signaling pathway, pathways in cancer, peroxisome and RNA degradation. Among the 6 gene sets, the bladder cancer and p53 signaling pathway sets were significant at 12, 24 and 48 h. We also found that the DDB2, RRM2B and GADD45A, genes related to the repair and damage prevention of DNA, were consistently up-regulated for genotoxic carcinogens. Our results suggest that a gene set analysis could provide a robust tool in the investigation of the different mechanisms of genotoxic and non-genotoxic carcinogens and construct a more detailed understanding of the perturbation of significant pathways.

Figure 1. A Venn diagram displaying the 57 gene sets that met p < 0.05 for at least one of the 12, 24 or 48 h time points.

Figure 2. Gene plot (top) from Globaltest and KEGG pathway (bottom) showing the fold change of individual genes in the p53 signaling pathway.

Red and green bars indicate up-regulated and down-regulated genes, respectively, after GTX exposure at A. 24 h or B. 48 h in comparison to NGTX exposure.

Figure 3. Principal component analysis revealed the distribution of 12 compounds in the training data.

A. PCA results for gene expression in the p53 signaling pathway gene set at 24 h [red, 5 GTX; blue, 7 NGTX]. B. PCA results for gene expression in the bladder cancer gene set at 48 h [red, 5 GTX; blue, 7 NGTX].