Verification of model predictions using double-mutant promoters.

<p>(<b>A</b>) Schematic representation of double-mutant promoters which lack two transcription factor binding sites (cf. <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1003421#pcbi-1003421-g001" target="_blank">Fig. 1B</a>). (<b>B</b>) Systematic analysis of transcription factor binding site deletion effects confirms cooperativity of BRE1 and STATBS. The impact of binding site deletions was calculated by taking the luciferase activity ratios of different promoters (indicated in the legend) and expressed as a log10-fold change (y axis). As expected for a system where both sites cooperatively enhance transcription, the fold-change upon a combined deletion of BRE1 and STATBS (red) is less than the product of the single deletion fold-changes (green and blue; see text). Data points are mean and standard deviation, and model predictions represent the range of measurement-compliant parameter sets, as derived from a parameter identifiability analysis (see Methods, Supplemental <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1003421#pcbi.1003421.s010" target="_blank">Protocol S2</a>). Only BRE1 and STATBS (but not BRE2) contribute to expression upon IL6 stimulation. (<b>C</b>) and (<b>D</b>) Co-stimulation heatmaps of double mutant promoters reveal that BRE1 and BRE2 are functionally similar in the absence of STATBS. (C) Heatmaps of luciferase activity under co-stimulation conditions. (D) Two-dimensional projection of the BRE1mSTATdel and BRE2mSTATdel data in C (averaged over all IL6 concentrations). Data points are mean (panel C, bottom row) or mean +/− std (panel D) (n = 6). Model predictions were formulated as ranges based on a parameter identifiability analysis (see Methods, Supplemental <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1003421#pcbi.1003421.s010" target="_blank">Protocol S2</a>), and show measurement-compliant parameter sets with highest and lowest predicted luciferase expression (top and middle rows in panel C; edges of shaded corridors in panel D). Data and model in D were normalized to basal luciferase expression in the BRE2mSTATdel construct. (<b>E</b>) Systematic analysis of transcription factor binding site deletion effects confirms buffering of BRE1 and BRE2 single deletions. Concepts similar to panel B. The fold-change upon a combined deletion of BRE1 and BRE2 (red bars) is higher than the product of the single deletion fold-changes (green and blue bars; see text). BMP stimulation conditions were considered to ensure that BRE1 and BRE2 both contribute to expression.