1. Developing super-resolved microscopy platform for imaging accessible chromatin sites across the genome

Conventional approaches for studying 3D genome structure rely on chromatin conformation capture (3C) techniques based on proximity ligation and high-throughput sequencing. We have developed 3D ATAC-PALM, a super-resolved fluorescence microscopy technique to capture the 3D distribution of accessible chromatin sites across the genome at the nanometer resolution. Through direct labelling of oligonucleotides with photoactivable fluorophor, we successfully transformed Tn5-assisted assay for transposase-accessible chromatin with sequencing (ATAC-Seq) to an imaging approach.  Based on lattice light sheet microscope developed by Eric Betzig's lab, we took advantage of astigmatism to generate point spread function (PSF) with different ellipticity and achieved much higher localization precision along the axial direction. This platform was subsequently applied to the imaging of accessible chromatin sites throughout the genome and within 3D nucleus at nanometer resolution. It successifully detected above 90% of all accessible chromatin sites within a single nucelus with localization precision of ~20nm in x/y dimension and ~50nm in z dimension. We also applied this platform to analyzing chromatin accessibility of mouse embryonic stem cells and discovered a clustered pattern of the accessible chromatin distribution. This technique therefore provides a direct imaging approach for the study of genome structure. 

Figure 1. Developing super-resolved microscopy platform for imaging accessible chromatin sites across the genome

2. Applying super-resolved imaging to the study of regulatory mechanisms underlying 3D genome structure

In this study, we took advanatge of 3D ATAC-PALM and revealed that Cohesin loss leads to spatial mixing of accessible chromatin domains across the genome. We further screened candidates that participate in the regulation of genome structure and discovered that BRD2 cooperates with Cohesin complex and modulates chromatin folding. Utilizing an established cell line with protein degron system in which Cohesin and Brd4 can be degraded selectively and independently, we applied 3D ATAC-PALM to the study of accessible chromatin structure before and after the depletion of each factor and combined as well. In combination with analyses by using techniques including micro-C, DNA in situ hybridization and computational simulations of polymer folding, we suggested that BRD2 and Cohesin complex play a "tug of war" in maintaining dynamical structure and function of the genome. 

Figure 2. Applying super-resolved imaging to the study of regulatory mechanisms underlying 3D genome structure

3. Discovering the unknown gene co-activation regulated by genome structure based on super-resolved imaging and single-cell omics 

In the past decade, though the implementation of 4D nucleosome project and others have revealed principles for 3D genome folding and have also uncovered that local conformational alteration can affect gene transcription and expression, we still lack a clear understanding of the exact biological function of genome structure. Specifically, does genome structure has a general role in controlling global gene expression since it has a general effect in orchastrating the genome folding? We tackled this question by firstly performing single-cell RNA-seq and ATAC-seq experiments. Instead of evaluating average gene expression/chromatin accessibility counts, we devised a statistical method to calculate expression/accessibility correlations per gene/domain pair across all the cells analyzed. Our results indicated that Cohesin depletion not only dramatically affects genome organization, but also significantly increases gene co-expression and chromatin co-accessbility among gene pairs. These findings lead to the hypothesis that "genome structure participates in the regulation of gene co-activation". We furthur validated this hypothesis by performing intron SeqFISH experiments to measure transcriptional co-bursting for all 208 actively transcribed genes in Chr 2. We then examined the mechanisms underlying this gene co-regulation with super-resolution imaging of accessible chromatin distribution and transcriptional condensates, in parallel with single-particle tracking of transcriptional factors/co-factors involved. Different from classical promoter-enhancer interaction mechanism, this is an unknown layer of transcriptonal regulation caused by spatial proximity due to genome folding and subsequently the sharing of common regulatory machinary. This research could shed light on understanding the etiology of various cohesinopathy disorders, such as Cornelia de Lange syndrome, and might also provide insights into cell fate decision and developmental processes. 

Figure 3. Discovering the unknown gene co-activation regulated by genome structure based on super-resolved imaging and single-cell omics