Determining the sequence and structure specificities of RNA-binding proteins

描述（由申请人提供）：人类基因组编码数百种含有rna结合结构域的蛋白质，其中大多数特征不明确，基因组分析表明转录后基因调控的广泛使用：在5‘和3’非翻译区（utr）有高度的序列保守性，选择性剪接很普遍，并且有许多亚细胞转录本定位，翻译的差异调节和转录本衰变调节的个体例子，通常在疾病相关的背景下。了解人类基因调控的一个关键方面将是绘制转录后调控网络，绘制这些网络的一个重要步骤是获得所有rna结合蛋白的rna结合活性的准确描述。我们已经开发了一种称为RNAcompete的方法，该方法使用单一结合反应来测量rna结合蛋白对数十万小rna （27- 35nt长）的相对偏好，这些小rna专门设计用于包含广泛的一级序列和二级结构。除了快速和系统之外，rnaccompete产生的结合活性描述通常优于传统的基序模型。在这里，我们建议使用rnaccompete来获得所有已知和预测的人类rna结合蛋白的rna结合活性的完整指数。我们的具体目标是：(1)将目前基于阵列的rnaccompete方法应用于所有294种人类RNA结合蛋白及其所有470个单个RNA结合域。(2)进一步发展rnaccompete方法，以创建更复杂的库，并使用下一代测序作为输出，以促进对具有多个rna结合结构域的蛋白质进行更详细的分析，并最终分析rna结合蛋白的复合物。(3)建立一个rna结合谱数据库，该数据库由我们在目标1和目标2中分析的文献编译而成。这一目标的一个组成部分将是探索rna结合活性的模型，以便提供细胞rna中潜在结合位点的最准确预测。(4)大RRM和KH域类序列和结构识别的决定因素分析。破译（或反驳）这些流行的rna结合结构域的氨基酸序列特征与它们结合的rna类型之间存在的映射，对于理解它们的功能，以及确定结合偏好的知识如何在物种和不同蛋白质之间转移，将是重要的。

DESCRIPTION (provided by applicant): The human genome encodes hundreds of proteins that contain RNA-binding domains, most of which are poorly-characterized, and genomic analyses indicate widespread use of post- transcriptional gene regulation: there is high sequence conservation in 5' and 3' untranslated regions (UTRs), alternative splicing is prevalent, and there are many individual examples of subcellular transcript localization, differential regulation of translation, and regulation of transcript decay, often in a disease-relevant context. A key aspect of understanding human gene regulation will be to map post-transcriptional regulatory networks, and an essential step in mapping these networks is to obtain an accurate description of the RNA-binding activity of all of the RNA-binding proteins. We have developed a method called RNAcompete which measures, using a single binding reaction, the relative preference of an RNA-binding protein to hundreds of thousands of small RNAs (27-35 nt long) specially designed to encompass a broad range of primary sequences and secondary structures. In addition to being rapid and systematic, RNAcompete produces descriptions of binding activity that are generally superior to conventional motif models. Here, we propose to use RNAcompete to obtain a complete index of RNA-binding activities for all known and predicted human RNA-binding proteins. Our Specific Aims are: (1) Application of the current array-based RNAcompete method to all 294 human RNA- binding proteins and all of their 470 individual RNA-binding domains. (2) Further development of the RNAcompete methodology to create more complex pools and use next-generation sequencing as an output, to facilitate more detailed analysis of proteins that have multiple RNA-binding domains, and, eventually, complexes of RNA-binding proteins. (3) Creation of a database of RNA-binding profiles, both compiled from the literature and produced by our analyses in Aims 1 and 2. A component of this aim will be to explore models for RNA-binding activities, in order to provide the most accurate predictions of potential binding sites in cellular RNAs. (4) Analysis of the determinants of sequence and structure recognition for the large RRM and KH domain classes. Deciphering (or refuting) the existence of a mapping between amino- acid sequence features of these prevalent RNA-binding domains and the types of RNAs they bind will be important for understanding their function, and also in determining how knowledge of the binding preferences can be transferred across species and among different proteins. PUBLIC HEALTH RELEVANCE: Virtually all human genes produce RNA, and many genes are controlled by proteins that bind to the RNA. We propose to use a new method we have developed in order to obtain a complete index of RNA-protein interactions. This work should ultimately identify regulatory mechanisms that control both normal health and disease.