Quantitative Analysis of Single Cell Learning

学习被认为需要大脑，但即使是非常简单的动物也有学习能力。 即使是单细胞也显示出原始类型的学习，但在没有神经系统的情况下，这种学习是如何发生的，目前还不清楚。 在这个项目中，一个巨大的单细胞生物Stentor将被用来探索单细胞如何学习。 Stentor细胞在它们的自然栖息地被捕食，但当被触摸时可以通过收缩成球来逃避攻击，但这种收缩会燃烧能量。 对于位于池塘植物上的Stentor细胞，它通常会被池塘植物或没有威胁的小藻类利用。 在决定是否在触摸时收缩时，Stentor细胞依赖于过去的经验。 这些细胞学会忽略轻微的、没有威胁性的接触，只有在受到更大的攻击性力量时才会收缩。 类似地，住在铁轨旁的人习惯了火车，当他们听到火车经过时，他们不会跳起来。 这种学习在所有动物中都可以看到，但它通常表现在那些有神经系统的动物身上。 单细胞可以学习吗？ 如果是，如何做到？ 在实验室中生长的单个Stentor细胞将被录像，因为它们响应于机械力而收缩，并且当不同的基因被关闭时，将测量反应。 这将揭示细胞如何在分子水平上学习。 与此同时，一个简单的行为数学模型将用于：a）预测与触摸时的感觉有关的基因; B）识别与驱动收缩有关的基因; c）识别细胞如何决定是否收缩。 该项目将首次展示单个细胞如何学习。 更广泛的影响活动将包括学生的跨学科培训沿着公共宣传活动。细胞整合多种输入，并在不同的行为反应之间进行选择，在某些情况下，似乎是从经验中学习。 细胞处理信息以产生适当行为的计算过程仍然知之甚少。 学习通常被认为是具有某种形式的神经元网络的多细胞动物的一个特征，但单细胞的学习能力似乎表明它是生命的一个更普遍的特征。 研究单个细胞学习的最易处理的系统之一是蓝斑巨细胞，这是一种对机械刺激做出反应的巨细胞。 重复的刺激导致习惯化，细胞学会忽略特定大小的刺激。 Stentor的习惯化已经有很好的记录，但机制基础尚不清楚。 在这个项目中，一位蓝腔生物学专家将与一位计算生物学专家合作，对单细胞中的学习如何进行定量理解。 该项目将联合收割机结合细胞反应的定量测量与细胞学习和基因功能分子扰动的简单双态数学模型，以询问有关学习如何发生的基本问题，确定细胞中学习和记忆的关键分子途径，并探索细胞决策的计算复杂性。 基因功能的研究将利用蛋白质组学和磷酸蛋白质组学信息来识别感觉和效应分子沿着将刺激与反应联系起来的信号连接。 一旦这些元素是已知的，它将有可能确定哪些方面的系统（感官，效应器，或信号）在学习过程中调制。这个奖项反映了NSF的法定使命，并已被认为是值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估的支持。

Learning is assumed to require a brain, but even very simple animals are capable of learning. Even single cells have been shown to display primitive types of learning, but how such learning takes place, without a nervous system, is currently not understood. In this project, a giant single cell organism, Stentor, will be used to explore how a single cell can learn. Stentor cells are preyed upon in their natural habitat but can escape from attack by contracting into a ball when touched, but this contraction burns up energy. For a Stentor cell sitting on a pond plant, it will often get tapped by pond plants or small algae that are not threatening. In deciding whether or not to contract when touched, the Stentor cell relies on past experience. The cells learn to ignore light, non-threatening touches, and only contract when hit with a larger aggressive force. In an analogous way, humans living by a railroad track get used to the train and they don’t jump when they hear it go by. This kind of learning is seen in all animals, but it is usually displayed in those with a nervous system. Can single cells learn? If so, how? Single Stentor cells grown in the lab will be videotaped as they contract in response to a mechanical force, and the response will be measured when different genes are shut down. This will reveal how the cell learns at a molecular level. At the same time, a simple mathematical model of behavior will be used to: a) predict genes that are involved in sensing when touched; b) identify genes that are involved in driving the contraction; and c) identify how the cell decides whether or not to contract. This project will show, for the first time, how a single cell is able to learn. Broader Impact activities will include the interdisciplinary training of students along with public outreach activities. Cells integrate multiple inputs and select between different behavioral responses, in some cases seeming to learn from experience. The computational processes by which cells process information to generate appropriate behaviors remain poorly understood. Learning is usually considered to be a feature of multicellular animals with some form of neuronal network, but the seeming ability of single cells to learn suggests it is a more general feature of life. One of the most tractable systems for studying learning by a single cell is Stentor coeruleus, a giant cell that shows quantifiable behaviors in response to mechanical stimulation. Repeated stimulation leads to habituation, in which the cell learns to ignore a stimulus of a particular magnitude. Habituation in Stentor has been well documented, but the mechanistic basis is unknown. In this project, an expert on the biology of Stentor coeruleus will team up with an expert on computational biology, to develop a quantitative understanding of how learning takes place in a single cell. The project will combine quantitative measurements of cell responses with a simple two-state mathematical model for cellular learning and molecular perturbations of gene function, to ask fundamental questions about how learning takes place, identify key molecular pathways that underlie learning and memory in a cell, and probe the computational complexity of cellular decision-making. Investigation of gene function will exploit proteomic and phosphoproteomic information to identify the sensory and effector molecules along with the signaling connections that link the stimulus to the response. Once these elements are known, it will then be possible to determine which aspects of the system (sensory, effector, or signaling) are modulated during the learning process.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.