Multigene biomarkers of pyrethroid exposure: preliminary experiments: 15k microarrays

We describe initial development of RNA profiling-based assays for detecting pyrethroid pesticides in water. We conducted 48-hour flow-through exposures of Pimephales promelas adults and larvae to lab water and eight nominal concentrations of each of four pyrethroid insecticides: bifenthrin, cypermethrin, esfenvalerate, and (trans) permethrin. Concentration-response curves suggest steep dose responses, and LC50s below most published values. Expression microarray analysis of dissected brains from adults and whole larvae exposed to cypermethrin and bifenthrin suggests both assay formats can be used to detect these pyrethroids at concentrations well below P. promelas LC50s, and below LC50s of about 70% of aquatic metazoan species. Geneset analysis and intersections between lists of differentially expressed genes failed to identify substantial functional overlaps between these toxicants. Geneset analysis was sensitive to parameter choice, which may reflect the large number of genes analyzed relative to the number of samples, and substantial biological variability relative to effect sizes. Several identified GO terms can be rationalized based on known pyrethroid action, but implications of other highlighted GO terms are unclear. Adult (6 months old) and larval (48 hrs post-hatch) fathead minnow were exposed for two days to the pyrethroids bifenthrin and cypermethrin at several concentrations, in order to estimate LC50s. For each pesticide, samples exposed to two concentrations well below LC50s were selected for transcriptomic analysis and classifier development. Cross-validation was used to estimate classifier performance and compare classifiers developed using dissected adult brains with classifiers developed using whole larvae.