Modeling MicroRNA-Target Functional Networks for Acute Myeloid Leukemia Prognosis

描述（由申请人提供）：急性髓性白血病（AML）是最常见的成人造血恶性肿瘤类型。它有很高的疾病复发率，这是化疗耐药性的结果。最近的生物学研究表明，复发表型的一个主要组成部分存在于罕见的白血病干细胞（LSC）群体中，其特征是广泛的增殖和自我更新潜力，对标准化疗药物的反应较差。我们假设i) lsc富集人群固有的共同分子特征反映了具有不利预后特征的异质AML的生物学特性；ii)这些关联可以从大量AML样本的基因和microRNA表达谱以及其他生物学特征的系统分析中得出。这个项目的总体目标是定义lsc特征的转录因子（TF）和microRNA在异质aml中的相互作用。我们建议整合关于LSC特征和大量AML临床结果的合格数据驱动和策划知识。具体而言，我们将在途径水平上进行荟萃分析，以建立lsc特异性生物学，重点关注AML预后TF/microRNA失调。这些方法得到了我们两种开创性数学方法的支持：个体微阵列表达功能分析（FAIME）和机制锚固定的表型-基因型网络（PGNet）。FAIME提供了一种将广泛可用的基因表达谱转化为单个途径谱的新过程，从而产生更多可重复的途径特征（三个队列中有46%的重叠，经验p<0.001）。PGNet方法揭示了由疾病关键调控因子调控的基因，并能准确预测患者预后——将范式从单基因/microRNA分析转向“机制锚定分析”。利用PGNet，我们成功地预测了表观遗传调节因子HDAC9与急性淋巴细胞白血病的存活相关。创新的是，该项目将询问调节lsc特异性生物学途径和AML预后的microrna /基因，整合调节因子-调节因子相互作用和调节因子-基因相互作用。在目标1中，我们将构建lsc特异性基因通路并确定调控因子-基因相互作用。在目标2中，我们将关联lsc特异性调节因子及其与AML结果相对应的靶基因。在目标3中，我们将开发新的方法来计算模拟lsc特异性调节因子和预后基因靶点之间的关键相互作用。大量可用的AML患者资料和我们提出的方法已被证明的能力表明，我们将实现建立lsc驱动的预后模型的目标。

DESCRIPTION (provided by applicant): Acute myeloid leukemia (AML) is the most common type of adult hematopoietic malignancy. It has a high rate of disease relapse, a consequence of chemoresistance. Recent biological studies suggest that a major component of the relapse phenotype resides in a rare population of leukemic stem cells (LSC), characterized by extensive proliferative and self-renewal potential, and poor response to standard chemotherapeutic agents. We hypothesize that i) common molecular characteristics inherent to LSC-enriched populations reflect the biology of heterogeneous AML with unfavorable prognostic features; ii) these associations can be derived from systematic analysis of gene and microRNA expression profiles, and other biologic characteristics of bulk AML samples. The overarching goal of this project is to define LSC-characterized transcription factor (TF) and microRNA interactions in heterogeneous AMLs. We proposed to integrate both qualified data-driven and curated knowledge about LSC characteristics, and clinical outcomes from bulk AML. Specifically, we will perform meta-analyses on a pathway level to build LSC-specific biology, focusing on AML prognostic TF/microRNA deregulation. These approaches are supported by our two pioneering mathematical methodologies: the Functional Analysis of Individual Microarray Expression (FAIME) and the mechanism-anchored Phenotypes-Genotype Network (PGNet). FAIME provides a novel process for transforming extensive available gene expression profiles into individual pathway profiles, resulting in more reproducible pathway signatures (46% overlap among three cohorts, empirical p<0.001). The PGNet method reveals genes regulated by disease-critical regulators and can accurately predict patient outcomes - shifting the paradigm from single gene/microRNA analysis towards "mechanism anchored profiling". Using PGNet, we have successfully predicted that the epigenetic regulator HDAC9 is associated with survival in acute lymphoblastic leukemia. Innovatively, this project will interrogate microRNAs/genes that regulate LSC-specific biological pathways and AML prognostication, integrating regulator-regulator interactions and regulator-gene interactions. In Aim 1, we will build LSC-specific gene pathways and identify regulator-gene interactions. In Aim 2, we will correlate LSC-specific regulators and their target genes corresponding to AML outcomes. In Aim 3, we will develop novel approaches to computationally model the crucial interactions among the LSC-specific regulators and prognostic gene targets. Both the abundance of available AML patient profiles and the proven ability of our proposed methods suggest that we will achieve our aim to build an LSC-driven prognostic model.