Differential cofactor dependencies define functionally distinct types of human enhancers (PRO-seq)

Development and homeostasis of all multicellular organisms rely on differential cell-type-specific gene expression that is regulated by non-coding genomic enhancer elements. Enhancers function through the transcription-factor-mediated recruitment of cofactors, a structurally and functionally diverse set of proteins that activate RNA polymerase II transcription at promoters. Despite the important roles of enhancers and cofactors in transcriptional regulation, it is still not clear if all cofactors are required for all enhancers or if different enhancers have distinct cofactor dependencies and whether differential cofactor dependencies could be used to functionally categorize enhancers. Here, we quantified enhancer activities along the entire human genome using the massively parallel enhancer-activity assay STARR-seq in HCT116 cells, following the rapid auxin-inducible degradation of eight different cofactors. We identified groups of enhancers with distinct cofactor requirements, including enhancers whose activity is insensitive to the depletion of the core Mediator subunit MED14 or the bromodomain protein BRD4, respectively. In particular, Mediator seemed dispensable for P53-responsive enhancers and MED14-depleted cells in which transcription globally failed were still able to induce endogenous P53 target genes such as p21. Similarly, BRD4 was not required for genes with a CCAAT- and TATA-box proximal enhancer, including histone genes and LTR12 family retrotransposons, and for the induction of TATA-box containing heat-shock genes. Taken together, we show that different types of enhancers with distinct cofactor dependencies exist, including enhancer types that function in the absence of well-known cofactors and are employed to regulate specific gene classes and transcriptional programs. This represents the first functional categorization of enhancers by their cofactor dependencies that improves our understanding of alternative ways to activate transcription, which can aid in developing more precise interventions to modulate gene expression. Nascent RNA transcription measured by PRO-seq in WT, MED14- or BRD4-AID-tagged HCT116 cells, upon P53 induction by Nutlin-3a and/or respective COF depletion. Each COF-AID-tagged cell line was treated with auxin (IAA) to deplete the respective COF or water (mock) as a control. Each cell line was in addition treated with Nutlin-3a to induce P53 signaling on top of either auxin or mock treatment. For MED14-AID cell line, two different treatment regimens were performed: 1) 3h auxin/mock treatment followed by 3h Nutlin-3a treatment, or 2) 12h auxin/mock treatment followed by 6h Nutlin-3a treatment. All experiments were performed in 2 biological replicates.所有多细胞生物的发育与稳态，均依赖于由非编码基因组增强子元件（non-coding genomic enhancer elements）调控的、具有细胞类型特异性的差异基因表达。增强子（enhancer）通过转录因子介导的辅因子（cofactor）招募发挥功能——辅因子是一类结构与功能均具多样性的蛋白质，可激活启动子处的RNA聚合酶II（RNA polymerase II）转录。尽管增强子与辅因子在转录调控中发挥关键作用，但目前仍未明确：是否所有辅因子均为所有增强子所必需，不同增强子是否具有独特的辅因子依赖性，以及是否可通过差异辅因子依赖性对增强子进行功能分类。本研究在HCT116细胞中，于快速生长素诱导降解（auxin-inducible degradation）8种不同辅因子后，利用大规模并行增强子活性检测技术STARR-seq，定量检测了全基因组范围内的增强子活性。我们鉴定出了具有不同辅因子需求的增强子类群，包括分别对核心中介体亚基MED14（Mediator subunit MED14）或溴结构域蛋白BRD4（bromodomain protein BRD4）的缺失不敏感的增强子。尤为特别的是，中介体（Mediator）似乎对P53应答增强子（P53-responsive enhancers）并非必需；即便在全局转录普遍受阻的MED14缺失细胞中，仍可诱导p21等内源性P53靶基因的表达。类似地，对于带有CCAAT盒（CCAAT-box）与TATA盒（TATA-box）近端增强子的基因（包括组蛋白基因（histone genes）与LTR12家族逆转录转座子（LTR12 family retrotransposons）），以及含TATA盒的热休克基因（heat-shock genes）的诱导过程，BRD4均非必需。综上，本研究证实具有不同辅因子依赖性的各类增强子广泛存在，包括可在缺失经典辅因子的情况下发挥功能、并用于调控特定基因类群与转录程序的增强子类群。这是首次基于辅因子依赖性对增强子进行功能分类的研究，有助于加深我们对转录激活替代途径的理解，可为开发更精准的基因表达调控干预手段提供支撑。本研究通过PRO-seq检测了野生型（Wild Type, WT）、MED14或BRD4经AID标记（auxin-inducible degron tag, AID tag）的HCT116细胞在经Nutlin-3a诱导P53表达、和/或进行相应辅因子（COF）降解处理后的原初RNA转录情况。每种经AID标记的辅因子细胞系均接受生长素（indole-3-acetic acid, IAA）处理以降解对应辅因子，或以纯水处理作为模拟处理对照（mock）。此外，每种细胞系还额外接受Nutlin-3a处理，以在生长素或模拟处理对照的基础上诱导P53信号通路激活。针对MED14-AID细胞系，我们设置了两种不同的处理方案：1）先进行3小时生长素/模拟处理对照处理，再进行3小时Nutlin-3a处理；或2）先进行12小时生长素/模拟处理对照处理，再进行6小时Nutlin-3a处理。所有实验均设置2次生物学重复。