Decoding the Spatial Grammar of Developmental Signaling

项目摘要 胚胎使用一种复杂的化学信号语言向细胞传达指令。一个有定义的 这种语言的特点是它在空间中的使用；指令以空间分辨的信号模式进行编码 分子。学习翻译这种模式语言--阅读和编写干细胞的指令 理解--是现代发育生物学的一个明确目标。这一能力将具有变革性 通过使指导替代组织的发展成为可能，对再生医学的影响 实验室。然而，要实现这一目标，必须克服两个关键挑战。首先，我们需要新的方法 以创建和测试发育信号模式。其次，我们需要能够预测 由任意的信号模式编码的发育结果。在这里，我们提出了一种新的方法， 使用高通量的光遗传操作来解决这两个挑战 发信号。我们的策略结合了用于人类发育信号的光遗传控制的新试剂 具有空间分辨光刺激平台的干细胞。我们的图案平台将允许我们创建 在一次实验中，为500多万个细胞提供单细胞分辨率的信号模式。我们将利用 这种方法提供了独特的规模，可以构建发育信号的深度采样配对文库 图案和由此产生的组织结构。此数据集将为建模创造令人兴奋的新机会 发育信号和合理引导组织的体外发育。在这项提案中，我们列出了第一个 这种方法在协调内胚层和中胚层的节点信号通路中的应用 脊椎动物早期胚胎的形成。首先，我们将从深度学习中引入计算方法来构建 预测任意结节信号模式导致的组织结构的模型。第二，我们将使用 这些模型可以指导合理设计可用于生长内胚层的三维组织 有机化合物在体外。最后，我们将使用高通量光遗传操作来严格构建和 约束节点阵列的机械模型。据我们所知，这项工作将构成第一个高- 空间分辨率信令模式的吞吐量筛选。通过对模式空间进行经验抽样， 前所未有的深度，我们的目标是建立一条通往复杂组织合理工程的新途径。

Project Summary Embryos communicate instructions to their cells using an elaborate language of chemical signals. A defining feature of this language is its use in space; instructions are encoded in spatially-resolved ‘patterns’ of signaling molecules. Learning to translate this language of patterns—to read and write instructions that stem cells understand— is a defining aim of modern developmental biology. This ability would have transformative implications for regenerative medicine by making it possible to guide the development of replacement tissues in the laboratory. However, two key challenges must be overcome to realize this goal. First, we need new methods to create and test developmental signaling patterns. Second, we need modeling frameworks that can predict the developmental outcomes encoded by arbitrary patterns of signaling. Here, we propose a new approach that addresses both of these challenges using high-throughput optogenetic manipulation of developmental signaling. Our strategy combines new reagents for optogenetic control of developmental signaling in human stem cells with a platform for spatially-resolved light stimulation. Our patterning platform will allow us to create patterns signaling with single-cell resolution for over 5 million cells in a single experiment. We will leverage the unique scale afforded by this approach to build deeply-sampled paired libraries of developmental signaling patterns and resulting tissue structures. This dataset will create exciting new opportunities for modeling developmental signaling and rationally guiding tissue development in vitro. In this proposal, we lay out a first application of this approach to the Nodal signaling pathway, which orchestrates endoderm and mesoderm formation in early vertebrate embryos. First, we will import computational methods from deep learning to build models that predict the tissue structures resulting from arbitrary Nodal signaling patterns. Second, we will use these models to guide the rational design of three-dimensional tissues that can be used to grow endodermal organoids in vitro. Finally, we will use high-throughput optogenetic manipulation to rigorously construct and constrain mechanistic models of Nodal patterning. To our knowledge, this work will constitute the first high- throughput screen on spatially-resolved signaling patterns. By empirically sampling pattern space with unprecedented depth, we aim to establish a new path to the rational engineering of complex tissues.