Enhancers are non-coding regions of DNA that coordinate spatio-temporal regulation of gene expression. These regulatory sequences contain binding sites for sequence-specific transcription factors. Enhancers are instrumental in evolution of novel developmental and morphological features, as well as quantitative expression differences in a given population. In order to understand enhancer function and develop generalizable, predictive models for enhancers, it is important to study how enhancers function at a quantitative level. I first developed a suite of reporter gene vectors, which made quantitative measurements of gene expression from enhancer feasible. This "pHonda" suite of vectors is designed for site-specific integration into fly genome to eradicate position effects, as well as it uses a specific 5'-UTR, which allows for a more diffuse distribution of mRNA, making it more amenable to quantitative studies.Dorsal-ventral patterning is regulated by a master transcription factor, Dorsal, which is the fly homolog of mammalian NF-kappaB protein. Dorsal regulates about 100 genes in early fly embryo and coordinates dorsal-ventral patterning. A number of enhancers for these genes have been identified, which have been found to contain binding sites for the above proteins. The availability of several tested enhancer sequences, and quantitative data for concentrations of thesefactors, make it a suitable system for carrying out quantitative studies. Using systematic mutagenesis and confocal microscopy I first generated a systematic perturbation dataset for enhancer of rhomboid gene. Next, I applied thermodynamic modeling to this dataset, which uses assumptions of statistical thermodynamics to derive gene expression as a function of probabilities of binding of different factors to enhancer sequences, and tested several models for protein cooperativity and repression on this dataset. Subsequently, I used these models to predict gene expression from enhancer sequences, which were not used for modeling, and found that the top-ranked models can predict gene expression from these sequences in a tissue-specific manner. My study highlights the importance of mathematical modeling to understand the general rules of enhancer function.
Read
