Abstract

More than 30% of mRNAs are repressed by microRNAs (miRNAs) but most repressions are too weak to have a phenotypic consequence. The diffuse actions have been a central conundrum in understanding the functions of miRNAs. By applying the May-Wigner theory used in foodweb studies, we show that i) weak repressions cumulatively enhance the stability of gene regulatory network (GRN), and ii) broad and weak repressions confer greater stability than a few strong ones. Transcriptome data show that yeast cells, which do not have miRNAs, use strong and non-specific mRNA degradation to stabilize their GRN; in contrast, human cells use miRNAs to increase degradation more modestly and selectively. Simulations indicate that miRNA repressions should be distributed broadly to >25% of mRNAs, in agreement with observations. As predicted, extremely highly expressed genes are avoided and transcription factors are preferred by miRNAs. In conclusion, the diffuse repression by miRNAs is likely a system-level strategy for enhancing GRN stability. This stability control may be the mechanistic basis of \"canalization\" (i.e., developmental homeostasis within each species), sometimes hypothesized to be a main function of miRNAs.