Differential Regulatory Networks in Disease: Application to Macular Degeneration

描述(由申请人提供)：定义在疾病中改变的调控网络不仅可以提供对疾病潜在机制的见解，还可以提供可能的治疗靶点。一些因素，如遗传变异和甲基化位点，可以破坏转录因子和顺式调控区域(如启动子和增强子)之间的相互作用，从而改变调控网络。然而，识别疾病中改变的网络仍然是具有挑战性的。首先，为了确定在基因调控中发挥作用的遗传变异和甲基化位点，我们需要绘制病理组织特有的调节区上的遗传变异和甲基化位点。而当 DNase I超敏位点(DHSS)和组蛋白标记图谱是确定调控区域的强大工具，但并不是每个实验室都有设备来测量感兴趣组织上的DHS和组蛋白标记。因此，我们需要一种足够准确的计算算法来区分疾病样本和正常样本之间的调控区域。其次，虽然已经确定了不同疾病的大量差异甲基化位点，但它们的功能作用在很大程度上仍不清楚。DNA甲基化通常被认为是一种有效的表观遗传修饰，它阻止了TF的招募，导致转录抑制。最近的研究和我们自己的初步结果表明，一些转录因子优先与甲基化的DNA结合，这种相互作用在某些情况下会激活基因转录。因此，我们需要确定这样的转录因子，并将这些甲基化依赖的TF-DNA相互作用整合到计算平台中。第三，我们需要一个统一的计算框架，以纳入可能改变监管网络的各种不同类型的因素。为了应对这些挑战，我们将开发一个计算框架，以纳入遗传和表观遗传变异的影响，并确定由这些影响改变的调控网络。在这个框架中，我们将开发一种计算方法，通过整合各种表观遗传学数据集来预测感兴趣组织中的调控区域(目标1)。我们的方法类似于蛋白质结构预测的同源建模，充分利用了ENCODE项目现有的表观遗传学数据集。然后我们将开发一个模型，在考虑遗传变异、DNA甲基化和转铁蛋白浓度的情况下，提供转铁蛋白和DNA之间相互作用强度的定量测量(目标2)。这个模型将纳入我们的新发现，即一些TF优先与甲基化的DNA基序结合。我们的计算框架将应用于老年性黄斑变性(AMD)，这是60岁及以上美国人视力丧失的主要原因。然后将对AMD中改变的监管网络进行实验评估(目标3)。最后，我们将通过一个交互的、用户友好的数据库(AIM 4)提供我们的软件和AMD的监管网络。

DESCRIPTION (provided by applicant): Defining the regulatory networks altered in the disease can provide not only the insights on the mechanisms underlying disease, but also the possible therapeutic targets. Several factors such as genetic variation and methylation sites can disrupt the interaction between transcription factors (TFs) and cis-regulatory regions (e.g. promoters and enhancers) and thus alter the regulatory networks. However, to identify the altered networks in disease is still challenging. First, to identify the genetic variation and methylation sites that play a role in gene regulation, we will need to map the genetic variation and methylation sites on the regulatory regions that is specific to the pathological tissues. While DNase I hypersensitivity sites (DHSs) and histone mark profiles are powerful to determine the regulatory regions, it is not feasible for every laboratory to be equipped to measure DHS and histone mark on the tissues of interest. Therefore, we need a computational algorithm that is accurate enough to differentiate the regulatory regions between diseased and normal samples. Second, although a large number of differentially methylated sites have been determined for different disease, their functional role remains largely unclear. DNA methylation has been generally considered as a potent epigenetic modification that prohibits TF recruitment, resulting in transcription suppression. Recent studies and our own preliminary results showed that some TFs preferentially bind to methylated DNA, an interaction that in some cases activates gene transcription. Therefore, we need to identify such TFs and incorporate these methylation- dependent TF-DNA interactions in the computational platform. Third, we need a unified computational framework to incorporate various and these diverse types of factors that could alter the regulatory networks. To address these challenges, we will develop a computational framework to incorporate the effects of genetic and epigenetic variations and identify the regulatory networks altered by these effects. In this framework, we will develop a computational approach to predict regulatory regions in tissue of interest by integrating various epigenetic datasets (Aim 1). Our approach is analogous to homology modeling for protein structure prediction, fully utilizing the existing epigenetic datasets from ENCODE project. We will then develop a model to provide quantitative measurement of interaction strength between TFs and DNA with consideration of genetic variation, DNA methylation and TF concentration (Aim 2). This model will incorporate our new discovery that some TFs preferentially bind to methylated DNA motifs. Our computational framework will then be applied to age-related macular degeneration (AMD), which is the leading cause of vision loss in Americans aged 60 and older. The altered regulatory networks in AMD will then be experimentally evaluated (Aim 3). Finally, we will make our software and the regulatory networks in AMD available through an interactive, user-friendly database (Aim 4).