Identification of differential gene expression profiles in rat cortical cells exposed to the neuroactive agents trimethylolpropane phosphate and bicuculline

Recent technological advancements in microfabrication combined with the rapid acquisition of full genome sequence data have led to the development of DNA arrays that have the capacity to monitor the expression levels of thousands of genes simultaneously. The development of this technology enables the use of functional genomics approaches to identify molecular markers associated with Cellular responsiveness to cytotoxic exposures. Databases containing unique cell-response profiles associated with specific toxicants or classes of toxicants can then be used in conjunction with cell-based biosensor platforms for environmental surveillance and toxicological assessment. An important issue that must be addressed, however, is whether DNA arrays can be used to identify transient gene modulation events in a reproducible manner. To address this issue, we utilized a primary embryonic rat (day 18) cortical cell model system and examined the RNA of both chemically treated and untreated cells using radioisotope-labeled cDNA probes and commercially available nylon membrane arrays. Using this approach, we examined experimental variability, basal gene expression variability, the occurrence of false positives, and the reproducibility of gene expression profiles obtained after chemical exposure. Minimal differences in gene modulation were observed between RNA samples from independently cultured cortical cells when array experiments were conducted in parallel (Pearson correlation coefficient for gene intensities = 0.98). In contrast, significant differences in gene expression were observed between array experiments conducted at different times with an identical RNA source (Pearson correlation coefficient for gene intensities = 0.91). Our results suggest the effect of basal gene activity differences in independently isolated cell cultures is negligible and that experimental variability possibly associated with the handling of RNA samples, differences in reverse transcription efficiency, hybridization, and/or signal acquisition are the primary contributors to variability in measurements. Using cDNA array analysis of unexposed cells from three independent cell culture preparations, we calculated false positive gene modulation events as a function of the threshold \ log(2) R \ > 1.0. The number of false positives using this criteria was 1 - 10 gene/ESTs/5109 actively transcribed gene/ESTs represented on the array. Using three independent replicate experiments of untreated cortical cell cultures, we determined that a threshold criterion of \ log(2) R \ > 0.63 for triplicate experiments would reduce the expected number of false positives in our experiments to less than one. Using this criterion, reproducible gene expression profiles were identified in cortical cells exposed to the neuroactive agents trimethylolpropane phosphate and bicuculline. (C) 2001 Elsevier Science B.V. All rights reserved.