Genome-wide coherence between enhancer states and 3D chromosome conformation in mouse T cell differentiation

Complex genomes are intricately organised in 3D nuclear space, yet the extent to which genome form and function are interconnected remains subject to intense debate. To elucidate this relationship, we generated developmental stage-specific maps of enhancer state, gene expression, and chromosome conformation during CD4 T cell differentiation. While progression from the DP to the CD4 SP stage of thymocyte differentiation was accompanied by activation and repression of hundreds of genes and enhancers, Hi-C maps of DP and CD4 SP were visually highly similar, and discrete Hi-C measures such as TADs, contact domains, loops, or compartments failed to unambiguously link genome organisation to gene regulatory activity. At face value, these findings appear to support proposals of a disjoint between genome form and function. By contrast, Twins, a deep learning approach trained to distinguish technical from biological variation in Hi-C data, revealed compelling coherence between gene regulatory changes and genome organisation. Analysis of local Hi-C contacts demonstrated that the strength of Hi-C contacts quantitatively reflected developmental stage-specific enhancer states. These findings demonstrate correspondence between genome form and function in mouse T cell differentiation, and suggest a rationale for the control of gene expression by distal regulatory elements. Chromatin contact profiles in wild type or CTCF/Rad21 double knockout cells were generated by in situ Hi-C. Chromatin contact profiles in control or enhancer insertion cells were generated by capture Hi-C. Examination of enrichment of total H3 and H3K27ac in primary cells by ChIP-seq.

复杂基因组在三维细胞核空间中呈现高度精密的组织结构，然而基因组形态与功能之间的关联程度仍存在激烈的学术争论。为阐明这一关联，本研究在CD4阳性T细胞分化过程中，构建了发育阶段特异性的增强子状态、基因表达及染色体构象图谱。
尽管胸腺细胞分化中从双阳性（Double Positive, DP）到CD4单阳性（Single Positive, SP）阶段的进程伴随数百个基因与增强子的激活与沉默，但DP与CD4 SP细胞的高通量染色体构象捕获（Hi-C）图谱视觉上高度相似，且诸如拓扑关联结构域（Topologically Associating Domain, TAD）、接触结构域、染色质环及染色质区室等离散Hi-C分析指标均未能明确将基因组组织结构与基因调控活性关联起来。
从表面上看，这些结果似乎支持“基因组形态与功能互不关联”的观点。相比之下，用于区分Hi-C数据中技术变异与生物学变异的深度学习方法Twins，揭示了基因调控变化与基因组组织结构之间存在令人信服的一致性。对局部Hi-C接触信号的分析显示，Hi-C接触强度可定量反映发育阶段特异性的增强子状态。
上述结果证实了小鼠T细胞分化过程中基因组形态与功能之间存在对应关系，并为远端调控元件调控基因表达的机制提供了理论依据。野生型及CTCF/Rad21双敲除细胞的染色质接触图谱均通过原位Hi-C（in situ Hi-C）技术构建。对照组及增强子插入细胞的染色质接触图谱则通过捕获Hi-C（capture Hi-C）技术构建。通过染色质免疫共沉淀测序（Chromatin Immunoprecipitation Sequencing, ChIP-seq）技术分析原代细胞中总H3及H3K27ac的富集情况。