Identifying the Goldilocks Model

See allHide authors and affiliations

Science Signaling  05 Feb 2013:
Vol. 6, Issue 261, pp. ec36
DOI: 10.1126/scisignal.2004025

The quantitative transcriptional responses to activation of signaling pathways depend on many factors, including chromatin structure, the presence or absence of transcription factors, the magnitude and duration of the signal, and combinatorial inputs from multiple pathways. Neuert et al. used the high-osmolarity glycerol (HOG) mitogen-activated protein kinase (MAPK) pathway of the budding yeast Saccharomyces cerevisiae to build models to predict the dynamics of the transcriptional response to osmotic stress. When cells experience osmotic shock, the kinase Hog1 translocates into the nucleus, where it activates expression of osmosensitive genes. Nuclear translocation of Hog1 was similar in all cells, but the magnitude of the transcriptional response differed between cells. Using single-molecule fluorescence in situ hybridization (smFISH) to quantify STL1 mRNA abundance, the authors built several models of varying complexity to predict the quantitative transcriptional responses of osmosensitive genes to osmotic stress. The authors used computational methods to eliminate models that were too simple or too complex to make accurate predictions. The model selected to have the most predictive power was tested against the others using smFISH to quantify STL1, CTT1, and HSP12 mRNA abundance over a range of environmental conditions and in cells harboring various mutations in genes encoding chromatin modifiers or overexpressing the transcription factor Hot1. This model revealed that the steps in the transcriptional response to osmotic stress were most affected by genetic background and environmental conditions. The activation of each target gene was determined by the amount of Hog1 that entered the nucleus, with each target having a distinct threshold for activation. Once this activation switch was thrown, gene activity was stably maintained. The magnitude and dynamics of the transcriptional response were also influenced by the rates at which each transcript was produced and degraded, which varied between target genes. The computational methods used in this study to identify the best model from a field of candidates could be extended to other complex signaling and gene regulatory systems.

G. Neuert, B. Munsky, R. Z. Tan, L. Teytelman, M. Khammash, A. van Oudenaarden, Systematic identification of signal-activated stochastic gene regulation. Science 339, 584–587 (2013). [Abstract] [Full Text]