Unbiased identification of spliceosome vulnerabilities across cancer

项目总结： 尽管选择性剪接是细胞多样性和发育过程中增长的主要驱动因素之一， 剪接机制可以在癌症中被劫持，促进转移、免疫逃逸、侵袭和抗肿瘤 细胞凋亡的作用。虽然剪接因子突变发生在1%-15%的患者中，但具体取决于癌症， 新出现的数据表明，常见的不受调控的癌基因，如myc，间接调节mrna。 在不同的恶性肿瘤中促进不同剪接异构体的导致癌症的加工途径。使用 最近开发的一系列非监督剪接检测和候选剪接调控预测 技术，我们发现剪接在成人和儿童癌症中广泛中断，独立于 明显的剪接因子突变。这些数据表明了一种潜在的范式转变模型，在这种模型中 广泛的协调剪接功能障碍发生在癌症中，可能是通过剪接因子的失衡 表达、信号或遗传变化。如果是真的，剪接体定向和上游治疗可能是 在癌症中广泛改变用途，专注于特定的剪接信号和牵连的监管 而不是仅仅针对特定的突变。测试这些假设并开发可重复使用的分析 对于癌症社区的资源，我们提出了以下目标。 目标1：将关键剪接途径漏洞与观察到的致癌事件联系起来 不同的癌症。我们将在全球层面上利用我们现有的集成式多- 针对数十种癌症和数千个样本的组学计算工作流程。识别的剪接事件 使用新的非监督或监督分析也将在不同的癌症内部和之间进行比较 作为不同来源的正常细胞，以确定可复制的肿瘤固有与分化相关的程序。 目标2：定义和验证儿童急性髓系白血病和多种疾病的核心剪接调控网络 人类癌症。我们将构建并验证一个新的学习模型来定义剪接调控网络 儿童急性髓系白血病，并最终跨越各种成人和儿童癌症。我们将采用当前的最佳实践 用于多证据转录调控网络推理，以拼接并严格测试我们的模型 验证数据。我们将使用大量实验性剪接因子结合数据集来改进我们的 预测。这些分析将确定新的剪接调节因子和RNA识别元件。 目标三：搭建精准剪接生物标记物检测和选择性剪接发现平台 目标抑制。我们将开发一个交互式计算界面来识别特定的RNA异构体 在目标1中获得的不同癌症中的剪接亚型与预后不良相关。通过整合剪接， 基因表达、蛋白质组学和甲基化数据，我们将能够发现 剪接事件与基因调控的不同模式有关，可能在蛋白质水平上表现出来。 将建立相关的异构体相互作用和加权的共表达网络，以确定特定剪接的优先顺序 已知的癌症途径中的事件。

PROJECT SUMMARY: Although alternative splicing is one of the major drivers of cellular diversity and growth during development, the splicing machinery can be hijacked in cancer to promote metastasis, immune escape, invasion and anti- apoptotic actions. While splicing factor mutations occur in 1-15% of patients, depending on the cancer, emerging data suggest that commonly dysregulated oncogenes such as MYC indirectly regulate mRNA processing pathways leading to cancer promoting alternative splice isoforms in distinct malignancies. Using a series of recently developed unsupervised splicing detection and candidate splicing regulatory prediction techniques, we discovered that splicing is broadly disrupted in adult and pediatric cancers independent of obvious splicing factor mutations. These data suggest a potentially paradigm shifting model, in which widespread coordinated splicing dysfunction occurs across cancers, likely via imbalances in splicing factor expression, signaling or genetic alternations. If true, spliceosome directed and upstream therapies may be broadly repurposed across cancers, focused on specific splicing signatures and implicated regulatory pathways rather than on specific mutations alone. To test these hypotheses and develop reusable analytical resources for the cancer community, we propose the following aims. Aim 1: Implicate key splicing pathway vulnerabilities with observed oncogenic events across diverse cancers. We will characterize alternative splicing on a global-level with our existing integrative multi- omics computational workflow across dozens of cancers and thousands of samples. Splicing events identified using novel unsupervised or supervised analyses will be compared within and between distinct cancers as well as normal cells of different origins to define reproducible tumor intrinsic vs. differentiation associated programs. Aim 2: Define and validate the core splicing regulatory networks in pediatric AML and diverse human cancers. We will build and validate a novel learning model to define the splicing regulatory network in pediatric AML and ultimately across diverse adult and pediatric cancers. We will adapt current best practices for multi-evidence transcriptional regulatory network inference to splicing and rigorously test our models with validation data. A large library of experimental splicing factor binding datasets will be used to improve our predictions. These analyses will identify novel splicing regulators and RNA recognition elements. Aim 3: Build a discovery platform for precision splicing biomarker detection and selective splicing target inhibition. We will develop an interactive computational interface to identify specific RNA isoforms associated with poor prognosis splicing subtypes in diverse cancers obtained in Aim 1. By integrating splicing, gene expression, proteomics and methylation data on the same patients, we will enable the discovery of splicing events linked to diverse modes of gene regulation, that potentially manifest at the protein level. Associated isoform interactions and weighted coexpression networks will be built to prioritize specific splicing events in known cancer pathways.