文摘
Transcriptomics is increasingly used to assess biological responses to environmental stimuli and stressors such as aquatic pollutants. However, fundamental studies characterizing individual variability in mRNA levels are lacking, which currently limits the use of transcriptomics in environmental monitoring assessments. To address individual variability in transcript abundance, we performed a meta-analysis on 231 microarrays that were conducted in the fathead minnow (FHM), a widely used toxicological model. The mean variability for gene probes was ranked from most to least variable based upon the coefficient of variation. Transcripts that were the most variable in individual tissues included NADH dehydrogenase flavoprotein 1, GTPase IMAP family member 7-like and v-set domain-containing T-cell activation inhibitor 1-like while genes encoding ribosomal proteins (rpl24 and rpl36), basic transcription factor 3, and nascent polypeptide-associated complex alpha subunit were the least variable in individuals across a range of microarray experiments. Gene networks that showed high variability (based upon the variation in expression of individual members within the network) included cell proliferation, metabolism (steroid, lipids, and glucose), cell adhesion, vascularization, and regeneration while those that showed low variability (more stability) included mRNA and rRNA processing, regulation of translational fidelity, RNA splicing, and ribosome biogenesis. Real-time PCR was conducted on a subset of genes for comparison of variability collected from the microarrays. There was a significant positive relationship between the two methods when measuring individual variability, suggesting that variability detected in microarray data can be used to guide decisions on sample sizes for measuring transcripts in real-time PCR experiments. A power analysis revealed that measuring estrogen receptor ba (esrba) requires fewer biological replicates than that of estrogen receptor bb (esrbb) in the gonad and samples sizes required to detect a 50% change for reproductive-related transcripts is between 12 and 20. Characterizing individual variability at the molecular level will prove necessary as efforts are made toward integrating molecular tools into environmental risk assessments.