Gene expression profiling in primary mouse hepatocytes discriminates true from false-positive genotoxic compounds

• Published in: Mutagenesis 2010-11, Vol.25 (6), p.561-568
• Main Authors: Mathijs, K., Brauers, K. J. J., Jennen, D. G. J., Lizarraga, D., Kleinjans, J. C. S., van Delft, J. H. M.
• Format: Article
• Language: English
• Subjects: Animals; Apoptosis; Biological and medical sciences; Carcinogenicity; Carcinogens - classification; Carcinogens - isolation & purification; Carcinogens - toxicity; Cell Culture Techniques; Cells, Cultured; Chromosome aberrations; Classification; Cluster Analysis; DNA Damage; DNA microarrays; False Positive Reactions; Fundamental and applied biological sciences. Psychology; Gene expression; Gene Expression Profiling; Genomes; Genotoxicity; Hepatocytes; Hepatocytes - drug effects; Hepatocytes - metabolism; Histones - genetics; Immune response; Lymphoma; Male; Mice; Mice, Inbred C57BL; Molecular and cellular biology; Molecular genetics; Mutagenesis; Mutagenesis. Repair; Mutagenicity Tests - methods; Mutagenicity Tests - standards; Salmonella
• ISSN: 0267-8357; 1464-3804
• DOI: 10.1093/mutage/geq040

Well-established in vitro methods for testing the genotoxic potency of chemicals—such as the Ames/Salmonella test, the mouse lymphoma assay, the micronucleus test and the chromosomal aberration test—show a high false-positive rate for predicting in vivo genotoxicity and carcinogenicity. Thus, there is a need for more reliable in vitro assays. We investigated whether gene expression profiling in metabolically competent primary mouse hepatocytes is capable of discriminating true genotoxic (GTX) compounds from false-positive genotoxic (FP-GTX) compounds. Sandwich-cultured primary hepatocytes from male C57Bl6 mice were treated for 24 and 48 h with five true GTX and five FP-GTX compounds. Whole genome gene expression modifications were analysed by means of Affymetrix mouse genome 430 2.0 microarrays. Filtered genes were used for hierarchical clustering and class prediction methods. Classifiers were generated by prediction analysis of microarray using a leave-one-compound-out method and selecting the genes that were common to the 10 training sets. For the training compounds, all but one were correctly classified. Validation of the classification model with five new compounds resulted in a 100% correct classification at 24 h and 80% at 48 h. The generated classifiers were mostly involved in metabolic and biosynthetic processes, immune responses and apoptosis. Applying genes whose expression change correlates with γH2AX foci, a measure for DNA damage, did not improve the classification. The present study shows that gene expression profiling in primary mouse hepatocytes is capable of discriminating between true GTX and FP-GTX compounds.