Transcriptome-based functional classifiers for direct immunotoxicity

• Published in: Archives of toxicology 2014-03, Vol.88 (3), p.673-689
• Main Authors: Shao, Jia, Berger, Laura F., Hendriksen, Peter J. M., Peijnenburg, Ad A. C. M., van Loveren, Henk, Volger, Oscar L.
• Format: Article
• Language: English
• Subjects: activation; bioactivation; Biomedical and Life Sciences; Biomedicine; blood mononuclear-cells; cyclin g2; Environmental Health; exposure; Food safety; Gene expression; Gene Expression Regulation - drug effects; Gene Ontology; Humans; Immunohistochemistry; Immunosuppressive Agents - toxicity; Immunotoxicology; in-vitro; jurkat; Jurkat Cells - drug effects; Occupational Medicine/Industrial Medicine; Oligonucleotide Array Sequence Analysis; Pharmacology/Toxicology; Real-Time Polymerase Chain Reaction; Studies; t-cells; T-Lymphocytes - drug effects; T-Lymphocytes - immunology; Toxicity; Toxicogenetics - methods; toxicogenomics; Transcriptome
• ISSN: 0340-5761; 1432-0738
• DOI: 10.1007/s00204-013-1179-1

Current screening methods for direct immunotoxic chemicals are mainly based on general toxicity studies with rodents. The present study aimed to identify transcriptome-based functional classifiers that can eventually be exploited for the development of in vitro screening assays for direct immunotoxicity. To this end, a toxicogenomics approach was applied in which gene expression changes in human Jurkat lymphoblastic T cells were investigated in response to a wide range of compounds, including direct immunotoxicants, immunosuppressive drugs, and non-immunotoxic control chemicals. On the basis of DNA microarray data previously obtained by the exposure of Jurkat cells to 31 test compounds (Shao et al. in Toxicol Sci 135(2):328–346, 2013 ), we identified a set of 93 genes, of which 80 were significantly regulated (|numerical ratio| ≥1.62) by at least three compounds and the other 13 genes were significantly regulated by either one single compound or compound class. A total of 28 most differentially regulated genes were selected for qRT-PCR verification using a training set of 44 compounds consisting of the above-mentioned 31 compounds (23 immunotoxic and 8 non-immunotoxic) and 13 additional immunotoxicants. Good correlation between the results of microarray and qRT-PCR (Pearson’s correlation, R  ≥ 0.69) was found for 27 out of the 28 genes. Redundancy analysis of these 27 potential classifiers led to a final set of 25 genes. To assess the performance of these genes, Jurkat cells were exposed to 20 additional compounds (external verification set) followed by qRT-PCR. The classifier set of 25 genes gave a good performance in the external verification: accuracy 85 %, true positive rate (sensitivity) 88 %, and true negative rate (specificity) 67 %. Furthermore, on the basis of the gene ontology annotation of the 25 classifier genes, the immunotoxicants examined in this study could be categorized into distinct functional subclasses. In conclusion, we have identified and validated classifier genes that can be used for the development of an in vitro assay for the identification and initial characterization of hazards for direct immunotoxicity of chemicals and drugs. This assay promises to complement animal-free toxicity testing approaches within the field of direct immunotoxicity.