Abstract
The structure of the SigB-dependent general stress regulon of Bacillus subtilis has previously been characterized by proteomics approaches as well as DNA array-based expression studies. However, comparing the SigB targets published in three previous major transcriptional profiling studies it is obvious that although each of them identified well above 100 target genes, only 67 were identified in all three studies. These substantial differences can likely be attributed to the different strains, growth conditions, microarray platforms and experimental setups used in the studies. In order to gain a better understanding of the structure of this important regulon, a targeted DNA microarray analysis covering most of the known SigB-inducing conditions was performed, and the changes in expression kinetics of 252 potential members of the SigB regulon and appropriate control genes were recorded. Transcriptional data for the B. subtilis wild-type strain 168 and its isogenic sigB mutant BSM29 were analysed using random forest, a machine learning algorithm, by incorporating the knowledge from previous studies. This analysis revealed a strictly SigB-dependent expression pattern for 166 genes following ethanol, butanol, osmotic and oxidative stress, low-temperature growth and heat shock, as well as limitation of oxygen or glucose. Kinetic analysis of the data for the wild-type strain identified 30 additional members of the SigB regulon, which were also subject to control by additional transcriptional regulators, thus displaying atypical SigB-independent induction patterns in the mutant strain under some of the conditions tested. For 19 of these 30 SigB regulon members, published reports support control by secondary regulators along with SigB. Thus, this microarray-based study assigns a total of 196 genes to the SigB-dependent general stress regulon of B. subtilis.
