All-optical readout and manipulation of neural circuits in the intact mammalian brain

大脑中的神经元通过电脉冲存储、处理和传输信息。大脑中电脉冲的时空模式如何驱动感知，或使动作得以执行？尽管神经科学最近取得了实质性进展，但这些关键问题尚未得到回答。我们建议开发新的方法来回答这些基本问题，以揭示我们的大脑是如何工作的，以执行超出人类工程师所能创造的复杂计算壮举。这种理解可能会让我们对大脑皮层的工作有更深入的了解，并有助于对抗大脑的衰弱障碍。以前关于电活动如何驱动行为的研究采取了两种互补的方法：将神经元活动与环境中发生的事情联系起来，以及在记录行为反应的同时刺激神经元活动。在第一种方法中，动物感觉系统执行的许多基本计算的特点是改变环境并记录大脑中神经元的反应。最近的技术进步现在可以同时录制显示大群神经元活动的电影。这类数据通常信息量很大，但仍然不能揭示所记录的活动是否单独对一个人的感知负责，或者可能只是在记录时巧合发生的。第二种方法是刺激神经元活动，并记录动物的反应。虽然这些实验通过提供更直接的证据来补充第一种方法，但它们只表明激活某些神经元就足够了。例如，在实验室中足以驱动动作的神经元不一定是动物正常行为时正常活动的神经元。这种细微的差异对理解神经代码至关重要，因为我们的目标是了解大脑在正常条件下是如何工作的，而不是它如何在实验室的人工条件下工作。这项提议的目标是通过关闭记录和操纵神经活动与行为之间的环路，将这两种方法结合在一起。我们已经开发出一种光学方法，它允许我们使用光来同时记录和操纵单个电脉冲水平上的许多神经元的活动。我们将进一步开发这一策略，在行为过程中对整个神经回路进行全光读出和操作。这将涉及到优化体内“全光学”讯问工具包，该工具包包含定制光学元件、新型基因工程探头和基于大规模计算平台的软件。这种方法将允许我们对功能定义的神经元组进行“实时”读出和操作，使我们能够在行为过程中“动态”地探测和询问神经代码。我们将通过在执行感觉辨别任务的小鼠的感觉皮质中进行实验来测试这种方法，在这些实验中，我们可以精确地定向操纵神经元，并检查其对行为的影响。本申请中概述的战略将提供一个新的技术平台，适用于神经科学中的许多基本问题，并有可能转化为临床应用。

Neurons in the brain store, process, and transmit information via electrical impulses. How does the spatiotemporal pattern of electrical impulses in the brain drive perception, or enable performance of an action? Such crucial questions have yet to be answered despite substantial recent progress in neuroscience. We propose to develop novel approaches to answer these fundamental questions, in order to reveal how our brains work to perform complex computational feats surpassing what human engineers can create. Such understanding may give us deep insights into the workings of the cerebral cortex, and will aid the fight against debilitating disorders of the brain.Previous research on how the electrical activity drives behaviour has taken two complementary approaches: correlating neuronal activity with what is happening in the environment, and stimulating neuronal activity while recording behavioural responses. In the first approach, many of the basic calculations performed by an animal's sensory system can be characterized by altering the environment and recording the response of neurons in the brain. Recent technical advances now enable recording of movies showing the activity of large groups of neurons at once. Such data are often highly informative, but nonetheless cannot reveal whether the recorded activity is solely responsible for one's perception, or may have just coincidentally occurred at the time of the recording. The second approach is to stimulate neuronal activity and record responses from an animal. While these experiments complement the first approach by providing more direct evidence, they show only that activating certain neurons is sufficient. For example, a neuron that is sufficient to drive an action in the lab may not necessarily be the neuron that is normally active when the animal behaves naturally. This subtle difference is crucial in understanding the neural code because the goal is to understand how the brain works under normal conditions, not how it is able to work under artificial conditions in the laboratory.The goal of this proposal is to bring these two approaches together by closing the loop between recording and manipulating neural activity and linking with behaviour. We have developed an optical approach which allows us to use light to both record and manipulate the activity of many neurons simultaneously on the level of individual electrical impulses. We will further develop this strategy for 'all-optical' readout and manipulation of entire neural circuits during behaviour. This will involve optimizing a toolkit for in vivo 'all-optical' interrogation incorporating custom optics, novel genetically engineered probes, and software built on large-scale computing platforms. This approach will allow us to perform 'real-time' readout and manipulation of functionally defined sets of neurons, allowing us to probe and interrogate the neural code 'on the fly' during behaviour. We will test this approach using experiments in the sensory cortex of a mouse performing a sensory discrimination task, in which we can manipulate neurons in precisely targeted manner and examine the influence on behaviour. The strategy outlined in this application will provide a new technology platform that is applicable to many fundamental problems in neuroscience with the potential for translation to clinical applications.