ABSTRACT: Time is often not characterized as a variable in ecotoxicogenomic studies. In this study, the temporal kinetics in gene expression were determined during exposure to crude oil and a subsequent recovery period. Juvenile rainbow trout, Oncorhynchus mykiss, were exposed for 96 hours to the water accomodated fractions of 0.4, 2 or 10 mg l-1 crude oil loadings. Following 96 h of exposure, fish were transferred to recovery tanks for 96 h. Gill and liver samples were collected after 24 and 96 h of exposure, and after 96 h of recovery for RNA extraction and microarray analysis. Fluorescently labeled cDNA was hybridized against matched controls, using salmonid cDNA arrays. Each exposure scenario generated unique patterns of altered gene expression. More genes responded to crude oil in the gill than in liver. In the gill, 1137 genes had altered expression at 24 hours, 2003 genes had altered expression levels at 96 h of exposure, yet by 96 h of recovery, no genes were significant ly altered in expression. The Gene Ontology terms associated with gill-responsive genes implicated membrane narcosis, a toxic mechanism for crude oil. By contrast, in the liver at 10 mg l-1, only five genes were changed at 24 h, yet 192 genes had altered expression after 96 h recovery. At 2 mg l-1 in the liver, many genes had altered regulation at all three timepoints. The 0.4 mg l-1 loading also showed 289 genes upregulated at 24 h after exposure. The Gene Ontology terms associated with altered expression in the liver suggested that the processes of protein synthesis, xenobiotic metabolism, and oxidoreductase activity were altered. The concentration-responsive expression profile of cytochrome P450 1A, a biomarker for oil exposure, did not predict the majority of gene expression profiles in any tissue or dose, since direct relationships with dose were not observed for most genes. While the genes and their associated functions agree with known modes of toxic action for crude oil, the gene lists obtained do not agree with our previously published work, presumably due to array analysis procedures. These results demonstrate that changes in gene expression with time and dose should be characterized in controlled laboratory settings before responses from field collected organisms are interpreted, and that processes for analyzing microarray data need to be developed such that standardized gene lists are developed, or that analysis is gene list independent before arrays are as a monitoring tool. Two channel experiment; Control versus exposed (samples were time matched). 3 biological replicates, two technical replicates for both exposed and control fish. Exposed and cotrol samples were paired so that fish had been sacrificed at the same time, but otherwise, samples were paired at random. One replicate per array
