Computational analysis of Human High Grade Gliomas

描述（申请人提供）：胶质瘤是一种具有多种遗传和表观遗传改变的致死性疾病。这些变化在癌症的发展和进展中以协调的方式协同工作。癌症系统生物学是一门新兴的学科，在这门学科中，高通量基因组数据和计算方法相结合，提供了对导致相同癌症表型的多种途径失调的连贯和系统的理解。这门新学科有望将医学实践从被动的转变为预测性的。在这里，我们将应用逆向工程计算方法来解剖和验证驱动高级别胶质瘤间充质表型的转录网络。恶性胶质瘤中间充质和血管生成相关基因的表达与非常差的临床预后相关。我们使用了ARACNe（精确细胞网络重建算法），这是由哥伦比亚国家生物医学计算中心（MAGNet）开发的工具之一，用于识别调节与预后不良相关的间充质基因表达特征的转录因子。后者是通过广泛收集恶性胶质瘤的微阵列表达谱的分层聚类来确定的。我们的分析已经确定了一个高度互连的模块，六个转录因子相互调节以及绝大多数间质基因。我们还将计算分析扩展到能够预测主转录调节剂的翻译后调节剂的新算法（MINDy, Modulator Inference by Network Dynamics）。在本提案中，我们将设计和使用新的计算工具来整合人类脑肿瘤的遗传、表观遗传和功能数据的许多来源。我们的目标是：重建和实验操作负责高级别胶质瘤间充质特征表达的转录和翻译后程序（目的1）；阐明高级别胶质瘤沉默间充质信号转录抑制因子ZNF238的机制，并测试ZNF238基因失活在小鼠胶质瘤形成中的作用（目的2）；计算鉴定和实验验证调节恶性胶质瘤间充质特征的“可药物”基因，并将其作为候选治疗靶点进行测试（目标3）；组装和传播全基因组的人类胶质瘤相互作用组（HGi），它将整合遗传、表观遗传和功能改变的各种来源，这些改变表征了高级别胶质瘤的间充质表型（Aim 4）。HGi将通过MAGNet中心的传播基础设施向科学界开放。最终，我们的目标是利用计算推断和实验验证的胶质瘤侵袭性调节因子作为治疗干预的宝贵新靶点。公共卫生相关性：高级别胶质瘤（也称为多形性胶质母细胞瘤）是人类中枢神经系统肿瘤中生长最快的类型。它们总是与预后不良有关。最近，多形性胶质母细胞瘤的两种内在能力被认为是恶性肿瘤的关键特征，即侵袭邻近正常大脑和产生新血管的能力。人们普遍认为，目前治疗高级别胶质瘤的失败主要是由于缺乏有效的工具来对抗侵袭性脑肿瘤的这些特征。这项拨款提案使用哥伦比亚大学医学中心最近开发的尖端计算技术，从大量原发性多型胶质母细胞瘤中利用微阵列表达谱，并确定负责恶性胶质瘤最具侵略性属性的关键遗传主调控因子（“间充质特征”）。我们将在细胞培养系统和实验动物中进行一系列最先进的实验，以验证新算法所做的预测。我们预计，确定指导恶性胶质瘤最具侵袭性特征的主要遗传程序将为治疗干预提供最好的目标。

DESCRIPTION (provided by applicant): Glioma is a lethal disease with multiple genetic and epigenetic alterations. These changes work in concert in a coordinated fashion in cancer development and progression. Cancer Systems Biology is an emerging discipline in which high-throughput genomic data and computational approaches are integrated to provide a coherent and systematic understanding of the diverse pathway dysregulations responsible for the presentation of the same cancer phenotype. This new discipline promises to transform the practice of medicine from a reactive one to a predictive one. Here we will apply a reverse engineering computational approach to dissect and validate the transcriptional network that drives the mesenchymal phenotype of high-grade glioma. The expression of mesenchymal and angiogenesis-associated genes in malignant human glioma is associated with very poor clinical outcome. We have used ARACNe (Algorithm for the Reconstruction of Accurate Cellular Networks), one of the tools developed by the Columbia National Center for Biomedical Computing (MAGNet), to identify transcription factors regulating a mesenchymal gene expression signature associated with poor prognosis. The latter was identified by hierarchical clustering of a wide collection of microarray expression profiles of malignant glioma. Our analysis has identified a highly interconnected module of six transcription factors that regulate each other as well as the vast majority of the mesenchymal genes. We have also extended the computational analyses to new algorithms able to predict post-translational modulators of the master transcriptional regulators (MINDy, Modulator Inference by Network Dynamics). In this proposal we will design and use new computational tools to integrate the many sources of genetic, epigenetic and functional date available on human brain tumors. Our goals are: to reconstruct and experimentally manipulate the transcriptional and post-translational programs responsible for the expression of the mesenchymal signature of high-grade glioma (Aim 1); to elucidate the mechanism by which high-grade glioma silence ZNF238, a transcriptional repressor of the mesenchymal signature, and test the role of ZNF238 gene inactivation in gliomagenesis in the mouse (Aim 2); to computationally identify and experimentally validate "druggable" genes that regulate the mesenchymal signature in malignant glioma and to test them as candidate therapeutic targets (Aim 3); to assemble and disseminate a genome-wide, Human Glioma interactome (HGi) that will integrate the diverse sources of genetic, epigenetic, and functional alterations that characterize the mesenchymal phenotype of high-grade glioma (Aim 4). The HGi will be accessible to the scientific community via the MAGNet Center dissemination infrastructure. Ultimately, we aim to exploit the computationally inferred and experimentally validated regulators of glioma aggressiveness as invaluable new targets for therapeutic intervention. Public Health Relevance: High-grade gliomas (also known as glioblastoma mutiforme) are the fastest growing type of central nervous system tumors in humans. They are invariably associated with a bad prognosis. Recently, two intrinsic capabilities of glioblastoma multiforme have been recognized as key traits of malignancy, the ability to invade adjacent normal brain and produce new blood vessels. There is general agreement that the current failures in the treatment of high-grade gliomas are mostly caused by lack of effective tools against these features of aggressive brain tumors. This grant proposal uses cutting-edge computational technologies recently developed at Columbia University Medical Center to exploit microarray expression profiles from a large collection of primary glioblastoma multiforme and identify the key genetic master regulators responsible for the most aggressive attributes of malignant gliomas ("the mesenchymal signature"). We will perform a series of state-of-the-art experiments in cell culture systems and experimental animals to validate the predictions made by the new algorithms. We anticipate that the identification of the master genetic programs directing the most aggressive features of malignant gliomas will offer the best possible targets for therapeutic intervention.