Defining the multi-dimensional code of zinc finger specificity-Resubmission-1

项目摘要 Cys2His2锌指DNA结合域是人类最常见的结构域，也是DNA结合的 这些蛋白质中的绝大多数的特异性仍未确定。其中许多区域的突变， 无论有没有已知的DNA结合数据，都与阿尔茨海默病的一系列疾病有关 (REST)到癌症(如鼻涕虫、WT1、CTCF)。因此，对这些蛋白质的表征具有很大的价值。 不幸的是，用于确定转录因子的DNA结合特异性的常用方法 未能解决锌指问题，至少部分原因是无法完全定义大目标 哺乳动物平均锌指蛋白所需的特异性。即使ChIP-Seq数据存在，它也是 有限，因为基因组的大小不允许我们捕捉到一个因子的全部结合潜力 可提供≥21bp的靶序列。因此，在没有全面了解蛋白质结合的情况下 潜在的，基因组中的SNPs将继续代表我们无法 预测。总而言之，几十年的研究启发了我们对这个领域的理解，但我们仍然处于 当谈到它作为转录因子的功能时，它是黑暗的。最近，我们采取了另一种方法来 定义这一领域，展示了一个由单个锌手指组成的人工屏幕允许我们 预测多指蛋白质的特异度比所有先前预测的准确度更接近或更高 算法。然而，这种方法没有考虑到相邻手指对 彼此之间。我们已经制作了一个相当于锌手指能力的全面快照 只在众多潜在的上下文环境中的一个中。在这里，我们建议扩展此方法并筛选 在一套包容的上下文环境下的锌手指。我们将考虑最常见的直接和 对相邻手指结合的间接影响以及影响锌的几何形状的因素 手指与DNA接合。我们将使用这些结果来提供有关相邻锌指的完整图景 确定它们的特异性，并通过搭建这些两指模型，预测和设计 大型多指蛋白质。通过这种方式，我们将定义锌指特异性的多维代码 这使我们能够预测所有锌指DNA结合的特异性，以及任何相邻手指的上下文 会改变这种特异性，以及导致邻近手指不相容和丧失 DNA结合功能。我们将应用这个模型来预测所有人类锌指蛋白的特异性， 通过体内鉴定一组知情的转录因子来验证这些预测，并进行测试 用设计者、人为因素预测多指捆绑的机理。

Project Summary The Cys2His2 zinc finger DNA-binding domain is the most common domain in human yet the DNA-binding specificities for the great majority of these proteins remain undefined. Mutations in many of these domains, both with and without known DNA-binding data, have been linked to a host of diseases from Alzheimers (REST) to Cancer (e.g. Slug, WT1, CTCF). Therefore, the characterization of these proteins holds great value. Unfortunately common methodologies used to determine the DNA-binding specificity of transcription factors have failed to address the zinc finger, at least in part because of an inability to fully define the large target specificities required of the average mammalian zinc finger protein. Even when ChIP-Seq data exists it is limited because the size of the genome does not allow us to capture the full binding potential of a factor that could offer a ≥21bp target sequence. As a result, without a comprehensive understanding of a protein’s binding potential, SNPs across the genome will continue to represent potential binding sites that we are unable to predict. In sum, decades of research have enlightened our understanding of this domain but we are still in the dark when it comes to its function as a transcription factors. Recently we have taken an alternative approach to define this domain, demonstrating that a synthetic, one-by-one screen of individual zinc fingers allows us to predict the specificity of multi-fingered proteins with similar or greater accuracy than all prior prediction algorithms. However, this approach fails to take into consideration the influences that adjacent fingers have on one another. We have produced the equivalent of a comprehensive snapshot of what a zinc finger is capable of in just one of many potential contextual environments. Here we propose to scale this approach and screen the zinc finger under an inclusive set of contextual environments. We will consider the most common direct and indirect influences on adjacent finger binding as well as factors that impact the geometry with which the zinc fingers engage the DNA. We will use these results to provide a complete picture of how adjacent zinc fingers determine their specificity and by scaffolding these two-fingered models, predict and design the specificity of large, multi-fingered proteins. In this way, we will define a multi-dimensional code of zinc finger specificity that allows us to predict all zinc finger DNA-binding specificities, how any neighbor finger context would modify this specificity, and the factors that result in adjacent finger incompatibility and loss of DNA-binding function. We will apply this model to predict the specificity of all human zinc finger proteins, validate these predictions through in vivo characterization of an informed set of transcription factors, and test predicted mechanisms of multi-fingered binding with designer, artificial factors.