Computational and experimental analyses of promoter architecture in yeasts

Changes in gene expression represent a dynamic response of cells to environmental cues. A fundamental challenge is to understand how regulatory information that specifies gene expression changes is encoded in genome sequences. The multifactorial regulation of eukaryotic transcription initiation is influenced by promoter architecture, which governs the assembly of multiprotein regulatory complexes that contribute to synergistic gene activation. The motivating thrust of this work is to distill key sequence features of promoter architecture and to understand the mechanisms by which these features regulate transcription initiation in yeast.; I describe two computational approaches to associate short DNA sequences with gene expression changes in yeast. Genome-mean expression profiles indicated the regulatory potential of individual sequences by averaging out the effects of multifactorial regulation. In addition, I integrated comparative sequence data into the analysis of gene expression data, based on the expectation that promoter architecture has been phylogenetically conserved. I predicted interactions between pairs of transcription factors using a series of statistical tests to identify pairs of DNA hexamers that were jointly conserved and closely spaced.; To transform these computational observations into mechanistic insights, I developed a synthetic promoter assay to investigate how reporter gene transcription was affected by varying the spacing and sequence between transcription factor binding sites. I applied this assay to characterize promoter architecture constraints on the collaborative recruitment of the coactivator Met4p by the transcription factors Cbf1p and Met31/32p in response to methionine starvation. I found that the order of binding sites was crucial, and that distance constraints on coactivator recruitment were more flexible than those for cooperatively binding transcription factors. Intriguingly, I discovered that certain sequence contexts between the binding sites abolished gene activation.; In conclusion, the incorporation of positional information for multiple transcription factor binding sites vastly improves the accuracy of regulatory predictions. The requirements of promoter architecture may vary, depending on the particular mechanism of transcription factor interactions. In general, close spacing between transcription factor binding sites appears to be necessary, but not sufficient, for multifactorial regulation. Further studies on the key determinants of sequence context would aid the synthetic design of regulatory sequences.