Research Roundup
The Forum (p. A670) describes several new tools aimed at extracting the full benefits from genomics data. The Encyclopedia of DNA Elements is envisioned as a complete catalog of all the functional elements of the genome, while the Gene Expression Nervous System Atlas maps mousehuman homologs expressed in the central nervous system. The PathBLAST software program, meanwhile, compares protein interaction networks between species. In other news, scientists find an additional use for the weight-lossdrug orlistat--as a potent anticancer agent.
The Genomics of Cancer
Under the auspices of the Toxicogenomics Research Consortium, researchers the University of North Carolina at Chapel Hill focus on identifying environmental factors that cause DNA damage or oxidative stress, and ultimately cancer. The NCT Update (p. A676) looks at what these researchers are learning about the genomics of cancer, and how microarray technology in particular is revolutionizing the study of this disease.
Regulatory Acceptance of Toxicogenomics Data
Although regulatory agencies and many chemical manufacturers welcome the new wealth of data that can be produced by microarray studies, others are less pleased with the financial and legal ramifications of being required to submit such data for screening purposes. There also are very real logistical challenges involved in requiring the submission of toxicogenomics data; thus, agencies are in no hurry to mandate such requirements. The Focus (p. A678) examines the arguments for and against the required submission of toxicogenomics data by pharmaceutical and industrial chemical manufacturers.
Gene Interaction Networks
Gene expression arrays have enabled researchers to roughly quantify the level of mRNA expression for a large number of genes in a single sample. Most analyses are aimed at a qualitative identification of what is different between two samples and/or the relationship between two genes. Toyoshiba et al. (p. 1217) propose a quantitative, statistically sound Bayesian methodology for direct quantification of gene expression networks. A gene expression network linking the arylhydrocarbon receptor and the retinoic acid receptor (RAR) was hypothesized and analyzed using gene expression changes in lung airway epithelial cells after exposure to dioxin. The method demonstrates support for the assumed network and the hypothesis linking the usual dioxin expression changes to the RAR system. Simulation studies demonstrated the method works well, even for small samples. (Also see Science Selections, p. A687)
Valproic Acid Teratogenicity
The antiepileptic drug valproic acid (VPA) is a potent inducer of neural tube defects in human and mouse embryos, but the mechanism of VPA teratogenicity is unknown. Using microarray analysis, Kultima et al. (p. 1225) compare the global gene expression responses to VPA in mouse embryos during the critical stages of teratogen action in vivo with those in cultured P19 embryocarcinoma cells in vitro. A subset of genes with similar transcriptional response to VPA in both embryos and the cell model were evaluated as potential biomarkers for VPA-induced teratogenicity. Several of the identified genes may be activated or repressed through a pathway of histone deacetylase (HDAC) inhibition and specific protein 1 (Sp1) activation. The data support a role for HDAC as an important molecular target of VPA action in vivo.
Predictive Toxicology Using Toxicogenomics
Steiner et al. (p. 1236) investigated whether biological samples from rats treated with various hepatotoxicants can be classified based on gene expression profiles. A complete serum chemistry profile and liver and kidney histopathology were performed in addition to gene expression analyses. Hepatic gene expression profiles using a supervised learning method (support vector machines) to generate classification rules were combined with recursive feature elimination to improve classification and to identify a subset of probes for use as biomarkers. The predictive models discriminated between hepatotoxic and nonhepatotoxic compounds and predicted the correct hepatotoxicant class in most cases. (Also see Science Selections, p. A686)
Assessment of Applicability Domain of SAR Models for Estrogens
Despite major advances in algorithms and software, quantitative structureactivity relationship (QSAR) models have inherent limitations associated with size and chemical-structure diversity of the training set, experimental error, and many characteristics of structure representation and correlation algorithms. Tong et al. (p. 1249) report on two QSAR models based on different data sets for classification of chemicals according to their abilities to bind to the estrogen receptor. The models were developed by using a novel QSAR method, Decision Forest (DF), which combines the results of multiple heterogeneous but comparable Decision Tree models to produce a consensus prediction. Despite being based on large and diverse training sets, both QSAR models had poor accuracy for chemicals within the domain of low confidence, whereas good accuracy was obtained for those within the domain of high confidence.
Arsenicals and Gene Expression
Previous research has demonstrated that 12-O-tetradecanoylphorbol-13-acetate (TPA) treatment increases the number of skin papillomas in v-Ha-ras transgenic (Tg.AC) mice that had received sodium arsenite in drinking water. Because the liver is a known arsenic target, Xie et al. (p. 1255) examined the pathophysiologic and molecular effects of inorganic and organic arsenical exposure on Tg.AC mouse liver. Tg.AC mice were provided drinking water containing sodium arsenite, sodium arsenate, monomethylarsonic acid, and dimethylarsinic acid for 17 weeks in combination with TPA after 4 weeks. Subchronic exposure of Tg.AC mice to inorganic or organic arsenicals resulted in toxic manifestations, hepatic arsenic accumulation, global DNA hypomethylation, and numerous gene expression changes.
