Quantitative, multiplexed and high-throughput: macroarrays of lysate microarrays

描述（由申请人提供）：哺乳动物信号网络包括具有共享成分、共同输入和重叠输出的生化途径。理解信息是如何通过这些途径流动的，需要了解整个信号网络的信息，而不是一个或两个组成部分的信息。为了在系统水平上研究信号，我们需要一种平行、定量和可靠的方式来测量许多蛋白质的丰度和翻译后修饰。此外，由于对信号的理解需要在多种条件下对许多蛋白质进行频繁的时间采样，因此这些方法必须是高通量的。在这里，我们描述的技术，模仿免疫印迹，但在一个多路复用和极其小型化的格式。细胞在96孔板中培养，并受到各种扰动（在选定的shRNA或cDNA存在下用表皮生长因子刺激）。然后将细胞裂解，裂解物以高空间密度排列在玻璃支撑的硝化纤维素衬垫上，也以96孔格式排列。通过用不同的抗体探测每个垫，可以评估信号网络的“状态”。目前，高通量的多维读数可以通过自动荧光显微镜或多路流式细胞术获得。尽管这两种技术都提供了同时跟踪一种以上蛋白质的能力，但它们依赖于使用不同颜色的荧光团，因此只能跟踪大约12种蛋白质。相比之下，这里描述的技术使单个样品可以在单独的微阵列上复制数千次，因此很容易缩放。本应用程序详细介绍了使裂解物微阵列技术严格定量的努力，并概述了使其高通量和可重复性的自动化策略。此外，由于在系统层面分析癌症的最大挑战之一是超越数据的简单描述，因此还提出了一种使用贝叶斯方法构建细胞信号传导预测模型的策略。作为概念验证，我们将重点研究A431细胞中的表皮生长因子信号。虽然这个系统已经被很好地理解了，但我们的方法应该捕捉到传统研究中不明显的蛋白质之间的高阶相互依赖性。更重要的是，我们的策略应该提供一种通用的方法，利用我们的高通量微阵列获得的数据，在研究较少的网络中发现因果关系。

DESCRIPTION (provided by applicant): Mammalian signaling networks comprise biochemical pathways with shared components, common inputs, and overlapping outputs. Understanding how information flows through these pathways requires information on signaling networks as a whole, rather than on one or two components. To study signaling at a systems level, we need ways to measure the abundance and post-translational modification of many proteins in a parallel, quantitative, and reliable manner. In addition, since an understanding of signaling requires the frequent temporal sampling of many proteins under multiple conditions, these methods must be high-throughput. Here, we describe technology that mimics an immunoblot, but in a multiplexed and extremely miniaturized format. Cells are cultured in 96-well plates and subjected to a variety of perturbations (stimulation with epidermal growth factor in the presence of selected shRNA's or cDNA's). The cells are then lysed and the lysates arrayed at high spatial density onto glass-supported nitrocellulose pads, also arranged in a 96-well format. By probing each pad with a different antibody, the `state' of the signaling network is assessed. Currently, high- throughput multidimensional readouts can be obtained either by automated fluorescence microscopy or by multiplexed flow cytometry. Although both techniques provide the ability to track more than one protein simultaneously, they rely on the use of different colored fluorophores and hence can only follow about a dozen proteins. In contrast, the technology described here enables a single sample to be replicated thousands of times on separate microarrays and is thus easily scaled. This application details efforts to make lysate microarray technology rigorously quantitative and outlines automation strategies that render it high-throughput and reproducible. In addition, since one of the biggest challenges in analyzing cancer at a systems level is to go beyond a mere description of the data, a strategy is also presented to build predictive models of cell signaling using Bayesian methods. As proof-of-concept, we will focus on epidermal growth factor signaling in A431 cells. Although this system is relatively well-understood, our approach should capture higher-order interdependencies between proteins that are not evident from traditional studies. More importantly, our strategy should provide a general way to uncover causal relationships in less well-studied networks using data derived from our high-throughput microarrays.