Abstract
Post-transcriptional gene regulation is driven by RNA-binding proteins (RBPs). Recent global approaches suggest widespread autoregulation of RBPs through binding to their own mRNA; however, little is known about the regulatory impact and quantitative models remain elusive. By integration of several independent kinetic parameters and abundance data, we modelled autoregulatory feedback loops for six canonical and non-canonical RBPs from the yeast Saccharomyces cerevisiae, namely Hrb1p, Hek2/Khd1p, Ski2p, Npl3p, Pfk2p, and Map1p. By numerically solving ordinary differential equations, we compared non-feedback models with models that considered the RPBs as post-transcriptional activators/repressors of their own expression. While our results highlight a substantial gap between predicted protein output and experimentally determined protein abundances applying a no-feedback model, addition of positive feedback loops are surprisingly versatile and can improve predictions towards experimentally determined protein levels, whereas negative feedbacks are particularly sensitive to cooperativity. Our data suggests that introduction of feedback loops supported by real data can improve models of post-transcriptional gene expression.