High-throughput optimization of genetically-encoded fluorescent biosensors

项目总结/摘要 遗传编码荧光生物传感器使我们能够捕捉实时的生化行为“电影” 在单个细胞内，它们已经被证明是学习新生物学的宝贵工具。最 突出的例子是细胞内钙传感器，它揭示了完整的神经元的实时活动。 大脑，但也有许多其他新的工具来测量葡萄糖浓度，蛋白激酶活性，半胱天冬酶 作用，活性氧和核心代谢物，如ATP和NADH。通过病毒载体表达，或 转基因，它们可以靶向单个细胞或细胞类型，因此它们可以揭示时间依赖性 在活的混合细胞型组织的背景下，这些细胞中的信号传导或代谢的变化-并且实际上 所有哺乳动物的组织都由多种细胞类型组成，这些细胞类型在信号传导和代谢中具有不同的作用。在 相比之下，生化和质谱测量具有精致的化学灵敏度，但它们通常 包括牺牲准备（使时间课程难以学习），并喜欢同样强大的磁性 尽管使用共振光谱/成像技术，它们很少具有单细胞特异性。 但与光谱方法不同的是，生物传感器必须针对每个目标进行专门定制。这 通常涉及半理性蛋白质工程的组合-其中荧光蛋白和 配体结合蛋白以特定的方式融合在一起，然后筛选随机或靶向的 传感器变体的诱变文库。 这种筛选过程是传感器开发的主要限制，也是许多出版物 生物传感器没有充分优化--这意味着许多已发表的传感器是“原理证明”， 不容易使用，或者更糟的是，有干扰，使他们不可靠的记者，他们的名义 目标的优化具有挑战性的一个原因是传感器的许多特性必须被 同时优化：荧光响应的大小，目标的灵敏度范围， 传感器的特异性（包括来自其他配体的干扰）和对环境因素的抗性 例如pH和温度。 因此，我们的目标是开发一种高通量和高含量的筛选方法，用于遗传- 编码的荧光生物传感器，特别是那些响应配体结合或其他化学物质的生物传感器， 刺激。这种筛选方法使用了一系列成熟的微流控和成像方法，我们 已经试运行了其中的大部分。完成后，这种筛选方法应可部署在其他 实验室广泛使用。它可以在不到一天的时间内筛选104-105个传感器变体， 在一打或更多不同条件下的每个传感器变体的信息。我们也将应用这种筛选 一系列已发表和未发表的生物传感器的方法，需要特定的优化。这个项目 将使高质量生物传感器的可用性得到显著改善，以研究新的生物学。

PROJECT SUMMARY/ABSTRACT Genetically-encoded fluorescent biosensors allow us to capture real-time “movies” of biochemical behavior inside individual cells, and they have already proven to be valuable tools for learning new biology. The most prominent examples are the intracellular calcium sensors, which reveal real-time neuronal activity in the intact brain, but there are many other new tools for measuring glucose concentration, protein kinase activity, caspase action, reactive oxygen species, and core metabolites such as ATP and NADH. Expressed via viral vectors or transgenes, they can be targeted to individual cells or cell types, and thus they can reveal time-dependent changes in signaling or metabolism in these cells in the context of a living, mixed-cell-type tissue – and virtually all mammalian tissues are composed of multiple cell types with distinct roles in signaling and metabolism. In comparison, biochemical and mass-spec measurements have exquisite chemical sensitivity, but they usually involve sacrificing the preparation (making timecourses hard to learn), and like the also-powerful magnetic resonance spectroscopy/imaging technologies, they rarely have single-cell specificity. But unlike spectroscopic methods, the biosensors must be tailored specifically for each individual target. This generally involves a combination of semi-rational protein engineering – in which a fluorescent protein and a ligand-binding protein are fused together in a specific way – followed by screening of random or targeted mutagenic libraries of sensor variants. This screening process is a major limitation for sensor development, and a reason that many published biosensors are not adequately optimized – meaning that many published sensors are a “proof of principle” that cannot easily be used, or worse yet, have interferences that make them unreliable reporters of their nominal targets. One reason that optimization is challenging is that many characteristics of a sensor must be simultaneously optimized: the size of the fluorescence response, the sensitivity range for the target, the specificity of the sensor (including interference from other ligands), and resistance to environmental factors such as pH and temperature. We therefore aim to develop a high-throughput and high-content screening approach for genetically- encoded fluorescent biosensors, specifically for those that respond to ligand binding or other chemical stimuli. This screening method uses a series of well-established microfluidic and imaging methods, and we have piloted most of these already. When complete, this screening method should be deployable in other laboratories for widespread use. It will enable the screening of 104-105 sensor variants in less than a day, with information about each sensor variant in a dozen or more different conditions. We will also apply this screening approach to a series of published and unpublished biosensors in need of specific optimizations. This project will enable a dramatic improvement in the availability of high quality biosensors to study new biology.