HazChem Human Array V3: Classification of Environmental Toxicants through Gene Expression Pattern for Risk Assessment

Abstract

Environmental chemicals such as fungicides, dioxin, or cadmium can cause changes in gene expression. Environmental toxicogenomic approaches using gene expression profiles are useful tools that might be exploited in risk assessments of environmental toxicants from natural sources or as the result of human-made pollution. The principal objective of this study was to compare the gene expression profiles of 17 environmental chemicals-16 type-identified chemicals and 1 obscured chemical-and to identify classifications that better characterized toxicity types by exposure. We then utilized 2 human cell lines, and determined the IC20 values of each. In order to classify the gene expression profiles of the 17 chemicals, we used a custom-made HazChem human array V3, based on previous studies. This array included a total of 1136 genes, all of which were specifically differentially expressed by exposure to VOCs, PAHs, POPs, and LTCs (liver-toxicity chemicals). As a result, we detected 286 of these genes that were differentially expressed by drug type, using a statistical method involving type-parametric Welch's t-test and the Benjamini-Hochberg false discovery rate (FDR-adjusted p-value < 0.01). However, one type-obscured chemical was shown to have endocrine disruption ability, and evidenced liver-toxicity somewhat close to that of POPs. Additionally, we used an SVM (support vector machines) class prediction method, and then selected 150 genes (prediction strength > 4.148) that could be used to classify the chemical types via Fisher's exact test. We identified 43 common genes via two methods as powerful class-predictive genes, and confirmed their classifications using PCA. These 43 genes may help in advanced screening chemical for similar toxicogenomic effects with 5 chemical-types.