Engineering human neural networks

工程人类神经网络

• 批准号: BB/H008608/1
• 负责人: Zhanfeng Cui
• 金额: $ 357.11万
• 依托单位: University of Oxford
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2010
• 资助国家: 英国
• 起止时间: 2010 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=BB%2FH008608%2F1
• 关键词: Engineering; human; neural; networks

Currently our knowledge of the human brain is restricted by ignorance of the most basic cellular functions of its fundamental unit, the neurone/astrocyte combination. Animal models of brain function are limited by their physical which differ in some vital respects to ours. In addition, the animal brains cannot be built from scratch in the manner that is necessary to understand any complex system or machine. The primary aim of the project is no less than to build a model of the most basic functions of the human brain to cast light on how living human neurones process and retain information and how they expand their memory of information-how they 'learn'. To begin we using a growth arrested human tumour cell line (NT2/D1 cells), which is then developed into an intimately connected mixture of neurones and astrocytes in similar proportions to those in our own brains using the 'differentiation' process which mimics human embryo development over the first few weeks of life and results in a miniature human neuronal tissue if grown in three dimensions. Once formed these cultures behave like brain slices of animals, as the astrocytes pulse at the same frequencies and the neurones can communicate in a network as the human brain does. We can fully manipulate the process of differentiation and even insert extra genes in the cells to visualize functions using multiphoton laser technology and calcium imaging. The key features of the system are that unlike an animal slice model, the cells are untraumatized and not slowly dying. The cultures can live for up to 6 months and so the most basic neuronal/astrocytic functions can be followed second by second as they happen in real time. The culture system has thus two considerable advantages over existing animal models, in that it is 'custom built' as required in order to understand the most basic steps of complex neuronal processes and it can be non-invasively monitored to study more advanced neuronal/astrocytic behaviour. To build a functioning model of the most basic functions of the human brain requires both cellular models, initially the NT2/D1 cells and then human stem cells, as well as the engineering and assembly of 3-D matrixes to construct biological neuronal networks (bNNs) which can should be able to store information as memory, as well as process data through a network of cells and then make a form of response. As even small areas of human brain tissue contain the ability to process and store and communicate with proximal and distal tissues, so we must build in a multi-compartmental approach where several basic biological networks are connected facilitating more advanced processing. Ultimately, the NT2.D1 cells (and eventually human stem cells) will be grown along 3-D neuronal networks which will have provable networking capability, which will be monitored using multi-photon laser technology as well as calcium imaging. Once the basic bNN networks have been formed, these will be challenged electronically to try to induce the cells to form new links (synapses) with each other to retain information, which is the basis of memory. This process, known as synaptic plasticity, is a vital step connection of multiple bNNs to determine if a more advanced processing such as visual memory development can take place. This is a very elementary form of 'training' where the linked bNNs can communicate, process and 'remember' information. If successful, this will be the most advanced model of basic human neuronal processing in existence and will pave the way for the adaptation to many other explorations of human neuronal activity, both compared with existing animal data and will facilitate the construction of bNNs which can be made to simulate different human conditions.

目前，我们对人脑的了解受到了对其基本单位--神经元/星形胶质细胞组合的最基本细胞功能的无知的限制。大脑功能的动物模型受到它们的生理功能的限制，它们在一些重要方面与我们的不同。此外，动物的大脑不能以理解任何复杂系统或机器所必需的方式从头开始构建。该项目的主要目标是建立一个人脑最基本功能的模型，以揭示活着的人类神经元如何处理和保留信息，以及它们如何扩大信息记忆--它们如何“学习”。首先，我们使用了一个生长受阻的人类肿瘤细胞系(NT2/D1细胞)，然后它被发育成紧密相连的神经元和星形胶质细胞的混合物，比例与我们自己的大脑中的细胞比例相似，这一过程模仿了人类胚胎在生命最初几周的发育，如果以三维方式生长，就会产生微型的人类神经元组织。一旦形成，这些文化的行为就像动物的脑片，因为星形胶质细胞以相同的频率跳动，神经元可以像人类大脑一样在网络中进行通信。使用多光子激光技术和钙成像技术，我们可以完全操纵分化过程，甚至在细胞中插入额外的基因，以可视化功能。该系统的关键特征是，与动物切片模型不同，细胞没有受到创伤，也不会缓慢死亡。这些培养物可以存活长达6个月，因此最基本的神经元/星形细胞功能可以在实时发生的时候一秒接着一秒地发生。因此，与现有的动物模型相比，该培养系统有两个相当大的优势，因为它是根据需要定制的，以便了解复杂神经元过程的最基本步骤，并且可以进行非侵入性监测，以研究更高级的神经元/星形细胞行为。要建立人脑最基本功能的功能模型，需要两个细胞模型，首先是NT2/D1细胞，然后是人类干细胞，以及设计和组装3-D矩阵，以构建生物神经元网络(BNN)，它应该能够将信息存储为记忆，以及通过细胞网络处理数据，然后做出某种形式的反应。由于即使是很小的人脑组织区域也包含处理和存储近端和远端组织并与之通信的能力，所以我们必须建立一种多隔室的方法，其中几个基本的生物网络连接在一起，以促进更高级的处理。最终，NT2.D1细胞(以及最终的人类干细胞)将沿着3D神经元网络生长，这种网络将具有可证明的联网能力，这将使用多光子激光技术和钙成像进行监测。一旦基本的BNN网络形成，这些网络将受到电子挑战，试图诱导细胞之间形成新的联系(突触)，以保持信息，这是记忆的基础。这一过程被称为突触可塑性，是连接多个bNN的关键步骤，以确定是否可以进行更高级的处理，如视觉记忆发育。这是一种非常基本的“训练”形式，其中链接的bNN可以交流、处理和“记住”信息。如果成功，这将是现有的最先进的基本人类神经元处理模型，并将为适应许多其他人类神经元活动的探索铺平道路，无论是与现有的动物数据相比，都将有助于构建可以模拟不同人类条件的bNN。

项目成果

Quantitative differences in developmental profiles of spontaneous activity in cortical and hippocampal cultures.

• DOI: 10.1186/s13064-014-0028-0
• 发表时间: 2015-01-28
• 期刊: Neural development
• 影响因子: 3.6
• 作者: Charlesworth P;Cotterill E;Morton A;Grant SG;Eglen SJ
• 通讯作者: Eglen SJ

Quantitative differences in developmental profiles of spontaneous activity in cortical and hippocampal cultures

皮质和海马培养物自发活动发育概况的数量差异

• DOI: 10.1101/009845
• 发表时间: 2014
• 作者: Charlesworth P

Improving characterisation of human Multipotent Stromal Cells cultured in 2D and 3D: Design and evaluation of primer sets for accurate gene expression normalisation

• DOI: 10.1371/journal.pone.0209772
• 发表时间: 2018-12-31
• 期刊: PLOS ONE
• 影响因子: 3.7
• 作者: Brinkhof, Bas;Jia, Huidong;Wang, Hui
• 通讯作者: Wang, Hui

A comparison of computational methods for detecting bursts in neuronal spike trains and their application to human stem cell-derived neuronal networks.

• DOI: 10.1152/jn.00093.2016
• 发表时间: 2016-08-01
• 期刊: Journal of neurophysiology
• 影响因子: 2.5
• 作者: Cotterill E;Charlesworth P;Thomas CW;Paulsen O;Eglen SJ
• 通讯作者: Eglen SJ

A closer look at neuron interaction with track-etched microporous membranes.

• DOI: 10.1038/s41598-018-33710-6
• 发表时间: 2018-10-19
• 期刊: Scientific reports
• 影响因子: 4.6
• 作者: George JH;Nagel D;Waller S;Hill E;Parri HR;Coleman MD;Cui Z;Ye H
• 通讯作者: Ye H