CRCNS: Multiresolution Modeling of Human Thalamocortical Upstates and Downstates

描述（由申请人提供）：哺乳动物皮层以两种根本不同的模式运作。其中一种在清醒时占主导地位，被称为上状态，因为神经元的放电率和突触活动相对较高。另一个，在最深的北部振荡 非快速眼动睡眠阶段的特征在于细胞放电的深度抑制，并被称为下状态。这种缓慢振荡（SO）已在具有细胞内记录的动物中进行了深入研究，特别是在麻醉下的体外和体内模型系统中。基本现象已经从通道特性和突触连接在现实的Hodgkin-Huxley（H-H）计算模型与有限数量的cells 4，6。最近在人类中的多微电极记录已经证明，SO对应于在第3和第4阶段睡眠EEG中突出的.5- 2 Hz δ活动2，并且进一步地，下状态可以相对孤立地发生为第2阶段睡眠的K复合体（KC）1。这些研究已经建立了上状态和下状态的基本局部机制，以及它们与在非侵入性记录中容易观察到的突出EEG现象的对应关系。然而，它们如何被触发和同步的重要方面仍然是未知的和有争议的。SO和KC是否发生在皮层的所有部位？如果是，它们是否优先发生在某些地区？不同的SO和KC涉及不同的皮质区吗？它们是同时发生在所有区域，还是遍布大脑皮层？如果它们扩散，是否有一个特征速度或起源点？上状态和下状态在触发或同步方式上是否不同？这些都是关于数十亿神经元如何协调的非常复杂的问题。尽管经验记录是提供线索所必需的，但这些记录必须用计算方法进行处理和解释，才能取得真实的进展。生物物理和统计的正向和反向计算是必要的，以将微电极数据与介观记录（ECOG-皮质电描记术）和非侵入性测量（MEG-脑磁图和EEG）相关联。神经模型是必要的，以测试特定的假设机制的起源和传播的上州和下州对应的微观和中观记录。需要结合神经建模和正向计算来将假设的机制与EEG和MEG记录相关联。拟议的研究将深入了解人类皮层的这些基本状态，在通道，神经元，电路，系统和非侵入性全脑水平上计算整合动物与人类记录。 虽然这项研究计划的具体目标是了解基本的皮质功能状态，但基于模型的进一步研究可以应用于睡眠障碍患者的异常EEG/MEG，以预测可能导致观察到的异常的机制。KC可能起着防止觉醒的作用;了解其神经基础可以更好地治疗失眠。大多数证据表明，SO是睡眠恢复过程的基本活动。SO似乎在巩固前一天获得的记忆方面也起着核心作用。睡眠障碍与神经认知功能降低以及各种不良生理和长期健康结局（包括全因死亡率、糖尿病和心血管疾病）存在因果关系。超过30%的人抱怨睡眠问题。睡眠障碍-特别是睡眠呼吸暂停，睡眠剥夺和嗜睡-影响7000万美国人，导致每年160亿美元的医疗费用和500亿美元的生产力损失。除了医疗保健带来的巨大经济效益外，教育效益还包括培训将参与研究的研究生和本科生。所有用于运行模型的软件将与其他研究人员共享，并将根据大学政策通过互联网提供。拟议的研究将建立新的跨学科合作，这将导致研究生和博士后研究员的跨学科培训，并将涉及代表性不足的群体和少数民族。除了在会议和讲座上的科学演示外，研究结果还将被纳入K-12教师可以使用的教学模块中，并与NSF赞助的学习科学中心一起由Sejnowski共同指导。 在马萨诸塞州综合医院（MGH- Cash）、纽约大学（NYU- Thesen）、马赛（Chauvel）和布达佩斯（Ulbert）进行人类颅内记录。MEG/EEG记录在UCSD（Halgren）进行。分析和建模发生在UC滨江（UCR- Bazhenov），巴黎（Destexhe）和UCSD（中心站点- Halgren，Sejnowski，Dale和Hagler）。

DESCRIPTION (provided by applicant): Mammalian cortex operates in two fundamentally different modes. One, dominant during waking, is termed the upstate because of relatively high neuronal firing rates and synaptic activity. The other, oscillating with the upstate in the deepest stages of non-rapid eye-movement sleep, is characterized by a profound suppression of cell-firing and is termed the downstate. This slow oscillation (SO) has been intensively studied in animals with intracellular recordings, especially in model systems in vitro and in vivo under anesthesia. The basic phenomena have been reproduced from channel properties and synaptic connectivity in realistic Hodgkin-Huxley (H-H) computational models with limited numbers of cells4,6. Recent multi-microelectrode recordings in humans have demonstrated that the SO corresponds to .5-2Hz delta activity prominent in the stage 3 and 4 sleep EEG2, and further, that the downstate can occur in relative isolation as the K-Complex (KC) of stage 2 sleep1. These studies have established the basic local mechanisms of upstates and downstates, and their correspondence to prominent EEG phenomena that are easily observable in non-invasive recordings. However, important aspects of how they are triggered and synchronized remain unknown and controversial. Do SO and KC occur in all parts of the cortex? If so, do they preferentially occur in some areas? Do different SO and KC involve different cortical areas? Do they occur in all areas simultaneously or do they spread across the cortex? If they spread, is there a characteristic speed or point of origin? Do upstates and downstates differ in how they are triggered or synchronized? These are very complex questions regarding how billions of neurons are coordinated. Although empirical recordings are necessary to provide clues, these must be processed and interpreted with computational methods to make real headway. Biophysical and statistical forward and inverse computations are necessary to relate the microelectrode data to mesoscopic recordings (ECOG- electrocorticography) and non-invasive measures (MEG- magnetoencephalography and EEG). Neural modeling is necessary to test if specific hypothesized mechanisms for the origin and spread of the upstate and downstate correspond to the microscopic and mesoscopic recordings. Combined neural modeling and forward computations are needed to relate hypothesized mechanisms to EEG and MEG recordings. The proposed studies will yield a deep understanding of these fundamental states of the human cortex, computationally integrating animal with human recordings made at the channel, neuronal, circuit, system, and non-invasive whole-brain levels. Although the specific goal of this research proposal is to understand fundamental cortical functional states, further research based on the models could be applied to abnormal EEG/MEG from patients with sleep disorders, to predict the mechanisms that may be responsible for the observed abnormalities. The KC may function to prevent awakening; knowing its neural basis could lead to better treatment of insomnia. Most evidence suggests that the SO is the essential activity underlying the restorative processes of sleep. The SO also appears to play a central role in the consolidation of memories acquired in the preceding day. Sleep disorders have a causal relationship with reduced neurocognitive functions as well as variety of adverse physiologic and long-term health outcomes including all-cause mortality, diabetes, and cardiovascular disease. Over 30% of the general population complains about sleep-related problems. Sleep disorders - notably sleep apnea, sleep deprivation and sleepiness - affect 70 million Americans, resulting in $16 billion in annual healthcare expenses and $50 billion in lost productivity. In addition to significant economic benefits from healthcare, educational benefits include the training of graduate students and undergraduates who will be participating in the research. All of the software for running the models will be shared with other researchers and will be available through the internet in accordance with University policies. The new cross-disciplinary collaborations that will be established by the proposed research will lead to cross-disciplinary training of graduate students and postdoctoral fellows and will involve underrepresented groups and minorities. In addition to scientific presentations at meetings and lectures, the results of th research will be incorporated into teaching modules that could be used by K-12 teachers, in conjunction with the NSF sponsored Science of Learning Center at UCSD co-directed by Sejnowski. Intracranial recordings from humans are performed at Massachusetts General Hospital (MGH- Cash), New York Univ. (NYU- Thesen), Marseille (Chauvel), and Budapest (Ulbert). MEG/EEG recordings occur at UCSD (Halgren). Analysis and modeling occur at UC Riverside (UCR- Bazhenov), Paris (Destexhe), and UCSD (the central site- Halgren, Sejnowski, Dale and Hagler).