UC San Diego

There are over 200 cell types in the human body, each with a unique gene expression program precisely controlled by regulatory elements encoded in the genome such as promoters, enhancers, and insulators. Methods to identify functional genomic elements have widely focused on sequence. While these methods have been successful in finding promoters and insulators, identifying other regulatory elements, namely enhancers, is still an open problem. Our understanding of human transcription is incomplete because we do not have a complete catalog of enhancers. Recently, it has become increasingly clear that an epigenetic layer of information, especially in the form of post-translational histone modifications, marks different functional regions of the genome. In Chapter 1, I use high-resolution maps of histone modifications in 1% of the human genome to show that active enhancers are marked by a chromatin signature distinct from promoters, and that this signature can be used to predict other active enhancers. In Chapter 2, I extend this method to predict active enhancers genome-wide in HeLa cells, showing that enhancers are epigenetically more dynamic than promoters or insulators. Marked enhancers are highly enriched near cell-type specifically expressed genes. This key positioning of active enhancers suggests they likely drive cell-type specific gene expression. In Chapter 3, to study a biological system more relevant to human development, I then apply this technique to embryonic stem cells before and after differentiation. Most enhancers display marked changes in chromatin states in a manner that correlates with differential expression of their predicted target genes. In addition, a set of poised enhancers are marked by a distinct chromatin signature near genes important for cell fate determination, underscoring the importance of these regulatory elements in regulating differentiation. Finally, in Chapters 4 and 5, I address the problem of what other chromatin signatures exist besides those at promoters and enhancers. I develop an unbiased de novo pattern-finding method called ChromaSig to find commonly occurring chromatin signatures. Applying ChromaSig to genome-wide maps of histone modifications, I find a novel chromatin signature marking exons and other marking distinct classes of repeat elements associated with distinct modes of gene repression.