Functional connectomics of a simple brain centre for discrimination and memory

包括人类在内的动物的复杂行为之所以成为可能，是因为大脑有能力区分大量的感官线索，并有选择地形成和回忆与特定线索相关的记忆。大脑已经进化出了识别和记忆大量感官线索的方法。据估计，人类可以分辨一万亿种气味；我们能记住很多面孔或无生命的物体；2014年诺贝尔生理学和医学奖授予奥基夫、莫泽和莫泽，因为他们展示了哺乳动物大脑如何为特定位置编码感觉信息。我们想要了解大脑回路的原理，它能识别和区分如此多的线索，以及依赖于这些线索的行为背后的原理。像人类和果蝇这样多种多样的动物已经找到了应对这一挑战的共同解决方案——不同的感官信号，如面孔、声音或气味，会导致高级大脑中少数分散的神经元出现稀疏的活动；不同的线索是通过激活不同的分散神经元来区分的。由于人类的神经元数量是苍蝇的数百万倍，因此人类可以区分和记住多得多的感官物体——但它们之间共享的是稀疏编码原理。感官对象的表征是如何形成和调节的，以最大限度地提高大脑区分它们的能力，并根据动物的需要调整表征？它们必须得到最佳的调节：如果对给定的感官刺激作出反应的神经元太少，大脑可能不会注意到它；但是太多的话，对不同感觉信号做出反应的神经元就会重叠太多，不同的信号就无法被区分。此外，许多常规的感官提示根本不需要被记住——感官体验的影响取决于清醒等情况，我们的注意力被引导到哪里，或者我们有多兴奋。为了回答这个问题，我们需要知道底层电路的确切组织：所有涉及的神经元，它们所做的接触，它们对刺激的反应，以及它们彼此之间的影响。这在大脑中是最容易的，因为大脑神经元很少，但仍然具有高度的感官辨别能力，用于形成和检索记忆。因此，我们将描述果蝇幼虫（蛆）感官识别的完整电路；果蝇的主要大脑结构，蘑菇体（MB），在果蝇幼虫中只有数百个神经元，而人类有数十亿个神经元。神经元及其连接的综合图将告诉我们哪些其他大脑区域或感觉器官向蘑菇体提供信息，或利用蘑菇体提供的信息。然后，我们和其他访问它的科学家可以制定和测试有关电路如何编码和调节特定感官表征的理论。这将是任何动物的第一个这样的大脑图谱。我们将通过许多连续的幼虫大脑薄片来追踪每个神经元。我们将在非常高的放大倍数下分析切片，以揭示每个神经元与其他神经元的联系。然后，我们将尝试鉴定能够靶向其他基因表达到这些神经元的遗传蝇种群，使它们能够被荧光标记，用于监测神经元活动，并在活体准备中特异性地激活或阻断。我们将优先考虑那些解剖学上属于调节整个MB输入区的回路的神经元。最后，我们将测试MB调控如何影响其处理感觉信息能力的特定模型：我们将使用遗传手段阻断或激活特定神经元，并测试这对其他已识别神经元活动的影响，或对行为幼虫辨别气味的能力的影响。我们的工作将为更广泛的科学界提供全面的资源，以做出预测并通过实验对其进行评估。我们希望预测也适用于更复杂的大脑。

Complex behaviours in animals, including humans, are made possible by the ability of brains to distinguish among an enormous range of sensory cues, and to selectively form and recall memories associated with specific cues. Brains have evolved ways to recognise and remember enormous numbers of sensory cues. It is estimated that humans can distinguish a trillion odors; we can remember very many faces or inanimate objects; and the 2014 Nobel Prize in Physiology and Medicine recognised O'Keefe, Moser and Moser for showing how mammalian brains encode sensory information for specific places. We want to understand the principles of the brain circuitry that recognises and discriminates such a range of cues, and that underlies behaviours that depend on this. Animals as diverse as humans and fruitflies have found common solutions to this challenge - distinct sensory cues, like faces, sounds or smells, cause sparse activity in few scattered neurons in higher brain; and different cues are distinguished by activating different scattered neurons. Since humans have millions of times more neurons than flies, they can distinguish and remember enormously more sensory objects- but the principle of sparse coding is shared between them.How are representations of sensory objects formed and regulated, to maximise the ability of the brain to discriminate among them, and tune representations to the needs of the animal? They must be regulated optimally: if too few neurons respond to a given sensory stimulus, the brain may not take notice of it; but too many, and the neurons that respond to different sensory cues will overlap too much, and the different cues will not be distinguished. Also, many routine sensory cues simply do not have to be remembered - the impact of a sensory experience depends on circumstances like wakefulness, where our attention is directed, or how aroused we are. To answer this question, we need to know the exact organisation of the underlying circuitry: all the neurons involved, the contacts they make, the stimuli they respond to, and their effects on each other. This is easiest in a brain that has few neurons, but still high sensory discrimination, used to form and retrieve memories. Therefore, we will describe the complete circuitry for sensory discrimination in the fruitfly larva (maggot); the main brain structure for this, the mushroom body (MB), has only hundreds of neurons in fruitfly larvae, compared to many billions in humans. A comprehensive map of the neurons and their connections will tell us which other brain regions, or sensory organs, provide information to the mushroom body, or use information provided from it. We, and other scientists who access it, can then formulate and test theories for how the circuitry encodes and regulates specific sensory representations. This will be the first such brain map for any animal.We will do this by tracing each neuron individually through many consecutive thin sections of a larval brain. We will analyse sections at very high magnification, to reveal each neuron's contacts with other neurons. We will then try to identify genetic fly stocks that can target expression of other genes to those neurons, allowing them to be labelled fluorescently, used to monitor neuronal activity, and activated or blocked specifically in living preps. We will prioritise neurons whose anatomy suggests belonging to the circuits that regulate the whole of the MB input region. Finally we will test specific models of how regulation of the MB affects its ability to process sensory information: we will block or activate specific neurons using genetic means, and test the effects of this on activity of other identified neurons, or on the abilities of behaving larvae to discriminate odors.Our work will generate a comprehensive resource for the wider scientific community, to make predictions and assess them experimentally. We expect predictions also to be applicable to more complex brains.