Nuclear Organization and Function

点击翻译按钮获取中文摘要

PROJECT SUMMARY. Proper regulation of gene expression is essential for cell differentiation and homeostasis. Most of our understanding of the mechanisms that control the transcription process comes from studies of the one-dimensional genome i.e. the 10 nm chromatin fiber. However, the genome is folded in the three-dimensional (3D) nuclear space, and the relationship between this organization and gene expression is poorly understood. Using Drosophila as a model system, where it is feasible to obtain 250 bp resolution Hi-C data, we have found that the genome is folded into only one type of domain, which we call compartmental domains. These domains precisely correlate with the transcriptional state of their sequences. Compartmental domains are also found in other lower eukaryotes. Based on this, we propose that compartmental domains represent an evolutionarily conserved principle of genome 3D organization. Drosophila and lower eukaryotes either lack CTCF or this protein is unable to stop cohesin extrusion. However, CTCF can interfere with the progression of cohesin extrusion in vertebrates, which in turn affects other types of interactions in the genome. Here we suggest extending concepts learned from the analysis of 3D organization in Drosophila to mammals by proposing an ambitious and substantive multi-disciplinary approach combining genetics, epigenomics, computational biology, and differentiation of human embryonic stem cells (hESC) into disease-relevant tissues. The hypothesis underlying the proposed experiments is based on the idea that, rather than the prevalent view of large compartments containing smaller TADs, the mammalian genome is organized by conserved principles into relatively small compartmental domains. Cohesin extrusion operates on top of the compartmental domain scaffold and affects its organization. To test this novel hypothesis, we will deplete specific proteins present in complexes required for various aspects of the transcription process. We will also deplete protein complexes responsible for H3K27me3- and H3K9me3-dependent silencing. We will then use Micro-C XL to obtain very high-resolution interaction data and examine effects of protein depletion on the formation of self-interacting domains and in the interactions between these domains. These effects will be examined in the presence and absence of cohesin in order to understand the contribution of loop extrusion to enhancer-promoter interaction frequency. We will examine the predictability of 3D genome organization from one-dimensional epigenetic information using machine learning computational tools. We will study the logic of CTCF loop formation by analyzing the local chromatin environment around CTCF sites able or unable to form loops of different strengths using a new computational tool we have developed. Principles learned from these experiments will be tested by analyzing changes in 3D organization and their relationship to gene expression during the differentiation of hESCs into pancreatic cells. Results from this work will fill critical gaps in our understanding of the relationship between 3D chromatin organization and transcription, and its possible role in human disease.