Normalizing for biology: accounting for technical and biological variation in levels of reference gene and...

Feeding, fasting and re-feeding is a common experimental paradigm for studying growth endocrinology. Herein we demonstrate dynamic changes in the livers of coho salmon under these conditions and how changes in liver composition can influence quantification and interpretation of liver gene expression data. A three-week fast resulted in decreases in hepatosomatic index (liver size), liver glycogen content, and liver DNA concentration. In addition, significant differences were found in liver transcript levels from fed and fasted fish for the reference genes, arp and ef1a, when these were normalized to total RNA. We took the additional step of normalizing reference gene transcript levels to the liver homogenate RNA/DNA ratio to account for differences in RNA yield/cell and the number of cells sampled, normalizing to transcript number per cell rather than transcript number per unit RNA. After this additional step no significant differences in liver transcript levels of reference genes were found. The significance of these results was illustrated by normalizing liver transcript levels of insulin-like growth factor 1 (igf1) to ef1a transcript levels or ef1a transcript levels by RNA/DNA. The different normalization strategies resulted in differing patterns of change in igf1 transcript levels between fed and fasted fish. The novelty of this work rests in a two-step normalization process, attempting to account for both 1) technical errors in reverse transcription and qPCR reactions, and 2) biological variance in liver samples.