Liver library: creating a microarray for hepatotoxicants. (Science Selections).

Mechanical improvements in high-throughput applications continue to increase the utility of the microarray approach for investigating toxic effects on genes. But improvements in the content of arrays may be the key to maximizing the value of these technologies, according to a paper in this month's issue [EHP 111:863-870]. In the report, researchers at Abbott Laboratories and Rosetta Inpharmatics, led by senior research scientist Jeffrey Waring, lay out the development of a microarray specifically constructed for studying the effects of hepatotoxicants.

Work in toxicogenomics has so far focused primarily on hepatotoxicity because of the importance of the liver as a site of toxic response. Whereas earlier toxicology-focused arrays were put together using DNA libraries from normal or diseased tissues, building a library from toxicant-challenged animals is a new approach. Because these animals were specifically expressing genes regulated in response to toxic exposures, it was possible for the Abbott-Rosetta team to enrich for genes regulated by toxic compounds, making their array a highly specific tool for understanding the function of rat liver undergoing toxic exposure. Understanding how gene expression changes when animals face different toxicants is especially important in light of growing evidence suggesting that even dissimilar toxicants can elicit similar response mechanisms calling similar groups of genes into play.

The array was made from cDNA derived from RNA from male Sprague-Dawley rats exposed to 52 different compounds at two levels during three-day toxicity studies. Applying the compounds orally, interperitoneally, or intravenously (depending on the compound), the scientists exposed three rats to both levels of each toxicant. They formed the pool of RNAs used to make the array from a total of 312 exposed rats. The exposure compounds induce a variety of toxic mechanisms including DNA damage, cirrhosis, oxidative stress, steatosis (accumulation of fat in the liver), and necrosis.

The scientists enriched their library for genes induced by exposure to the study toxicants by using a subtractive hybridization approach that allowed them to eliminate transcripts that were also present in nonexposed animals. Using animals exposed for three days allowed induction of gene-level responses in the liver, but avoided capturing genes involved in the later processes of secondary inflammation or fibrosis. Sequencing clones from the library allowed identification of more than 2,700 expressed putative genes. About 20% of these genes, the scientists indicate, do not appear to have been previously described.

Genes from this library make up about 25% of the array, which contains 25,000 probes. The other 75% includes rat genes with known human orthologs (which help compare gene expression patterns between species), genes allowing comparisons between specific and nonspecific hybridization, hybridization targets to allow comparisons of hybridization intensity, and other controls.

The researchers say these gene expression profiles can be used to build a predictive database encapsulating biological responses to toxic insult. If the concept of "guilt by association" is to prove accurate, they write--if compounds are considered to have toxic liabilities when they closely associate with a known toxicant--it is extremely important to equip the array with the correct genes to distinguish the mechanism of toxicity.