CRCNS: Biophysical properties of parallel neural circuits serving night vision

项目描述（申请人提供）：本项目由两位神经生物学家和两位应用数学家合作完成。其主要目标是将实验和计算相结合，以详细了解单个视网膜神经元和突触的生物物理特性如何影响杆视觉期间视觉信息的并行处理。我们将对哺乳动物视网膜神经回路的组成部分进行实验表征和计算建模，这些神经回路构成哺乳动物的夜间（暗视）视觉。这个电路的解剖结构被很好地描述了：光子被杆状光感受器吸收，产生神经信号，这些信号通过一系列中间神经元分布到多种类型的视网膜神经节细胞（GCs），即视网膜的输出细胞。这些中间神经元通过化学突触和电突触相互连接。每种类型的GC对光都有独特的响应，这被认为反映了突触前回路的特性。这允许每种GC类型对视觉场景的不同特征进行编码，从而促进最终指导行为的高级大脑区域的进一步抽象。通过对每个组件进行表征和仔细建模，将组件组装成一个综合模型，并通过实验验证和完善整体模型，我们将确定平行视网膜微电路的生物物理特性如何产生不同的GC输出。智力优势：视网膜是少数几个生理刺激在体外和体内引起的输出相似的神经回路之一，使其易于实验分析。然而，视网膜编码的大量信息使得理解其功能电路变得困难。在模拟和实验的迭代过程中，我们将开发用于杆视觉的视网膜网络的详细生物物理模型。特别是：1)约束离子通道模型的数据将通过直接记录感兴趣的神经元获得；2)化学和电突触将通过对突触偶联神经元的记录直接表征；数据将用于构建杆路径的网络模型；3)利用实验确定的解剖和生理参数，构建不同神经元的室室模型；4)利用不同类型的gc的光诱发记录对整个网络模型进行验证和完善。计算模型将提供一个框架，在该框架中，可以理解视网膜网络中探测杆信号并行处理的实验。此外，我们将探索模型网络运行的参数空间，以开发关于视网膜功能的新的和实验可测试的假设。更广泛的影响：经过验证的计算模型将通过ModelDB数据库（一个神经元模型的公共存储库）提供，从而为其他研究人员提供一个工具，以开发关于视网膜处理的新的实验可测试的假设。更一般地说，通过该项目获得的数据和随附的计算模型将提供神经回路中信息处理的验证示例；这样的信息可以作为其他神经系统中可能的信息处理的例子。该计划还将为神经科学家和应用数学家提供详细的跨学科培训机会。应用数学专业的学生将获得扎实的神经科学背景，神经科学专业的学生将获得定量建模技术的基础。作为这种互动的一部分而开发的教学材料将提供给更大的神经科学界。此外，pi将寻求机会组织会议或研讨会（例如，在年度神经科学学会会议上），将致力于神经回路实验和计算结合研究的研究人员聚集在一起。

DESCRIPTION (provided by applicant): This project is a collaboration between two neurobiologists and two applied mathematicians. Its main goal is to combine experiment and computation to develop a detailed understanding of how the biophysical properties of individual retinal neurons and synapses shape the parallel processing of visual information during rod vision. We will characterize experimentally and model computationally the components composing the neural circuitry of the mammalian retina that subserves night (scotopic) vision. The anatomy of this circuitry is well-described: photons absorbed by rod photoreceptors generate neural signals that are distributed to multiple types of retinal ganglion cells (GCs), the output cells of the retina, via a series of interneurons. These interneurons are coupled to each other by both chemical and electrical synapses. Each type of GC has a unique response to light, which is presumed to reflect the properties of the circuitry presynaptic to it. This allows each GC type to encode a different feature of the visual scene, thereby facilitating further abstractions by the higher brain areas that ultimately guide behavior. By characterizing and carefully modeling each component, by assembling the components into a comprehensive model, and by validating and refining the overall model experimentally, we will determine how diverse GC outputs emerge from the biophysical properties of parallel retinal microcircuits. Intellectual merit: The retina is one of the few neural circuits for which output evoked by physiological stimuli is similar in vitro and in vivo, making it amenable to experimental analysis. The sheer volume of information encoded by the retina, however, makes understanding its functional circuitry difficult. In an iterative process of simulation and experiment, we will develop detailed biophysical models of the retinal network utilized for rod vision. In particular: 1) Data to constrain ion channel models will be acquired by direct recording from neurons of interest; 2) Chemical and electrical synapses will be characterized directly by recording from pairs of synaptically coupled neurons; data will be used to construct a network model of the rod pathway; 3) Compartmental models of different neurons will be constructed using anatomical and physiological parameters determined by experiment; 4) The overall network model will be validated and refined using light-evoked recordings from different classes of GCs. The computational model will provide a framework in which experiments probing parallel processing of rod signals within the retinal network may be understood. Moreover, we will explore the parameter space within which the model network operates to develop new and experimentally testable hypotheses about retinal function. Broader Impacts: Validated computational models will be made available via the ModelDB database, a public repository of neuronal models, thereby providing a tool to allow other researchers to develop new experimentally testable hypotheses about retinal processing. More generally, the data obtained through this project and the accompanying computational model will provide a validated example of information processing in neural circuits; such information can serve as an example of possible information processing in other neural systems. The proposed project also will provide neuroscientists and applied mathematicians the opportunity for detailed cross-disciplinary training. Applied mathematics students will acquire a solid background in neuroscience, and neuroscience students will acquire a foundation in quantitative modeling techniques. Teaching materials developed as part of this interaction will be made available to the larger neuroscience community. Additionally, the PIs will pursue opportunities to organize a meeting or symposium (e.g., at the annual Society for Neuroscience meeting) to bring together researchers working on combined experimental and computational studies of neural circuits.