Inferring a transcription regulatory network by directed perturbation

Transcription plays a key role in cellular processes and its regulation is of paramount importance. The aim of the work described in this thesis is to study the transcription regulatory network of Saccharomyces cerevisiae, employing genome-wide approaches. All the three presented research studies have in common that individual genes are deleted and resulting gene expression changes are monitored by DNA microarrays. It is first described how gene expression changes can be used as detailed molecular phenotypes to study the transcription regulatory network underlying a signalling pathway. Genome-wide expression changes of 91 viable deletion mutants of different glucose signalling and metabolic pathways are analysed. A gene signature is used to group pathway members with similar effects on transcription. A new network approach is developed that is designed to explain gene expression changes upon deletion of one pathway member through the transcriptional regulation of another pathway member. This new approach reveals hierarchy and feedback in the transcription regulatory network. In particular, it predicts that the different glucose pathways converge on the transcriptional regulation of the biosynthesis of storage carbohydrates. For understanding the transcription regulatory network, it is important to discern target genes of gene-specific transcription factors (GSTFs). Genome-wide gene expression changes of 183 viable GSTF deletion mutants are compared with available DNA binding data, re-evaluating the overlap between both data types as well as discerning direct target genes of GSTFs. Besides determining roles of previously uncharacterized GSTFs, for example Stp3, this comparison has led to the first systematic classification of GSTFs into activators and/or repressors. Of all surveyed gene-specific transcription factors that could be classified, activators account for less than 54%. The remaining 46% of gene-specific transcription factors are repressors (37%) or have a dual function (9%). The unanticipated high number of gene-specific repressors indicates that in the yeast S. cerevisiae, chromatin is not as restrictive to transcription as has previously been thought and indicates that a considerable part of the gene-specific machinery is aimed at restricting unwanted transcription. Recent studies have systematically exposed large numbers of non-additive genetic interactions, the majority of which are functionally uncharacterized. To investigate such genetic interactions between GSTFs, we systematically analysed 72 viable double deletion mutants and compared them to the respective single deletion mutants. By generating a high-resolution gene expression atlas, epistatic effects of GSTF pairs on the expression of individual genes are investigated. Known genetic interactions are confirmed, and new ones are revealed. The analysis also provides evidence for two previously uncharacterized mechanisms, one for a negative (“buffering by induced dependency” between Hac1 and Rpn4) and one for a positive genetic interaction (“alleviation by derepression” between Gln3 and Gzf3). The study provides general insight into the complex nature of epistasis and proposes new models for genetic interactions, the majority of which do not fall into easily recognizable within- or between-pathway relationships.