PhD Exit Seminar – Gurdeep Singh (Mitchell Lab)

Mammalian genomes are >98% non-coding and the function of most of this DNA remains unknown. Enhancers are one of the major components of the complex non-coding genome, which regulate gene expression in a tissue specific manner by binding transcription factors (TFs). Moreover, non-coding regulatory regions are critical for normal development, inter & intra-species evolution and often inappropriately regulated in disease contexts, yet we have only a limited understanding of these regions at a sequence level. The analysis of known enhancers has failed to provide a sequence code of enhancers, how many and which specific TFs are required to build a functional enhancer. To better understand the transcriptional regulatory code, I identified embryonic stem (ES) cell enhancers using mouse-human comparative epigenomics and machine learning. I found genomic regions with conserved binding of multiple transcription factors (TFs) in mouse and human embryonic stem cells (ESCs) are enriched in a diverse repertoire of transcription factor binding sites (TFBS) including known and novel ESC regulators. The accuracy of these predictions was confirmed using enhancer reporter assays and site directed mutagenesis of conserved TFBS in natural enhancers as well as the construction of synthetic enhancers. As I found that natural enhancers have on average 12 unique conserved TFBS I hypothesized that >10 unique TFBS are important for regulatory activity. Multiple synthetic enhancers constructed from >10 unique TFBS had robust enhancer activity, whereas sequences containing <10 unique TFBS even those with optimal TFBS combination or 14 repetitions of one TFBS had minimal activity compared to a natural enhancer. These findings reveal a required feature of mammalian enhancers, TFBS diversity above a threshold. Moreover, based on the conserved TFBS threshold in ESCs, I developed a sequence-based enhancer prediction model using one mouse TF ChIP-seq to identify human enhancers and novel TF regulators with high accuracy in heart, liver and brain tissues. These identified human enhancers have links to genetic diseases and cancer. Overall, my PhD research has provided a deeper understanding of the sequence code conferring enhancer function in multiple tissues.