VOLATILE ANESTHETIC ACTION IN A COMPUTATIONAL MODEL OF THALAMOCORTICAL NETWORKS

丘脑皮质网络计算模型中的挥发性麻醉作用

• 批准号: 8171810
• 负责人: ALLAN GOTTSCHALK
• 金额: $ 0.11万
• 依托单位: CARNEGIE-MELLON UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2010
• 资助国家: 美国
• 起止时间: 2010-08-01 至 2013-07-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/8171810
• 关键词: Amnesia; Anesthetics; Aplysia; Behavior; Brain; Calcium; Cell Nucleus; Cells; Cellular Morphology; Complex; Computer Retrieval of Information on Scientific Projects Database; Computer Simulation; Computer Systems; Conscious; Data; Development; Environment; Equation; Feedback; Functional Imaging; Funding; Gated Ion Channel; General Anesthesia; Goals; Grant; Hippocampus (Brain); Individual; Institution; Interneurons; Intervention; Ion Channel; Ligands; Link; Methods; Modeling; Morphology; Neurons; Pain; Phase; Population; Process; Publishing; Research; Research Personnel; Resources; Rest; Running; Services; Site; Sleep; Source; Stimulus; System; Thalamic structure; Unconscious State; United States National Institutes of Health; Work; base; cell behavior; cell type; computing resources; hippocampal pyramidal neuron; millisecond; network models; neural model; receptor; response; simulation; supercomputer; voltage

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Background Anesthetic agents modulate a number of voltage gated and ligand gated ion channels, but there does not seem to be specific channels that are the site of action for all anesthetics. Even when the site of action of a specific anesthetic is known, the process whereby anesthetic modulation of ion channels leads to the anesthetic state has yet to be established. The overall goal of this research is to use large scale computational models to elucidate the integrated systems level response of model neurons to anesthetics. The data generated from computing resources requested in this application will be used to support an application to NIH for additional resources to continue this work. To appreciate the importance of building a computational bridge between anesthetic action at the receptor level and the systems level, it is necessary to realize the fundamentally different responses to anesthetics in each domain. A single anesthetic can modulate the activity of one or more voltage gated and/or ligand gated ion channels. Typically, the concentration effect curves for a given channel are relatively shallow with midpoints occurring at anesthetic concentrations that are well above those used clinically. This graded modulation of ion channel behavior contrasts the abrupt changes that anesthetics induce at the systems level. For clinically useful anesthetic concentrations the brain is far from quiescent, as revealed by electroencephalographic and functional imaging studies. Over a population of subjects, the onset of the anesthetic state is relatively abrupt as anesthetic concentration is increased, and the concentration at which this occurs is considerably below the midpoint of the concentration effect curve for the putative ion channels. The systems level response to anesthetics is additionally complex because it encompasses a number of distinct features that emerge at distinct anesthetic concentrations. Minimally these include amnesia, loss of consciousness, blockade of painful stimuli, and immobility. Importantly, these effects can be produced by anesthetics that target entirely different sets of receptors. The qualitatively disparate behaviors described demonstrate the need for approaches linking anesthetic action at the receptor level with large scale systems level behavior. Although there is no specific network behavior that is definitively linked to general anesthesia, there is a growing appreciation that at least some aspects of general anesthesia are linked to the ability of the brain to generate coherent oscillations. These oscillations are thought to originate in the thalamic circuitry and the synchronization of these oscillations are dependent on the interaction established between the thalamus and the cortex. It is reasonable to hypothesize that these effects on the thalamus and cortex could be an important component of the anesthetic state since this region has already been shown to be closely associated with both sleep and consciousness. Furthermore, our preliminary results have demonstrated the ability of anesthetics to both synchronize and slow oscillations in a model of the thalamic relay and reticular nucleus neurons. The possible impact of these alterations on the rest of the thalamus and its interaction with the cortex are important considerations which have yet to be examined. To date, we have examined the effects of a variety of anesthetics on single cell and small network models of the reticular nucleus of the thalamus, thalamocortical neurons, hippocampal neurons, a fast spiking interneuron network, and Aplysia. We now seek to expand this effort to incorporate neurons which are more realistic with respect to types of ion channels and morphology. Preliminary results in small networks (2 cells of each neuron type) incorporating 4 different types of neurons, pyramidal neurons (PY) and interneurons (IN) in the cortex and thalamic relay (TC) and reticular nucleus (RE) neurons in the thalamus, have shown that the feedback between the thalamic neurons and the cortical neurons are important in understanding the behavior of the neurons under anesthetic effects. Being able to create large complex networks (100 cells of each neuron type) is essential to discerning the differences in anesthetics on the overall system level behavior and at the cellular level. Goals The following represent our immediate research goals with the requested support from PSC: 1. Continue our development of detailed pharmacologic models of anesthetic modulation of realistically complex biophysical neural models, where we now extend this work to the neurons of the thalamus and cortex. 2. Examine the ability of these networks to synchronize their activity as a function of the extent of pharmacologic intervention. We have already constructed several models of individual thalamic neurons and networks. The models have thus far been rather limited in complexity. It is clear that a greater level of model detail will permit the study of anesthetic effects that have heretofore been oversimplified. Increasing the level of detail in both ion channels and morphology is critical in assessing the integrative aspects of anesthetic action, as small and subtle changes in cell behavior can have large effects when incorporated into a larger network. A model endowed with sufficient detail in ion channel population, cell morphology and network interaction can be constructed from published models of thalamic cells and circuits. Each cell consists of multiple compartments representative of their morphology, contains both voltage gated and calcium controlled ion channels, sophisticated calcium concentration handling, and connects to the network through inhibitory GABAA channels. Supercomputer Use The goal in this phase of the research is to develop the network described above in a parallel environment using Parallel Genesis, a simulation environment already supported by the PSC. In the process of developing this network, preliminary study of the individual neuron and of the network in the presence of anesthetics will be conducted. Resources Needed Based on simulations of a network of 100 single compartment cells developed in Mathematica and run on a single 3.0 GHz processor, we estimate we will need 10,000 service units on the Terascale Computing System (TCS) to accomplish these goals. The use of implicit numerical methods in solving the equations permits a step size of .01 ms. With this step size, a 100 single compartment cell network requires approximately 20,000 cpu seconds to run for 10,000 simulation milliseconds (or 1*10^6 steps). As the TCS consists of 1 GHz processors, we use the approximation that 10,000 simulation seconds will require 20,000 cpu sec*(3 Ghz/1 GHz)/(3,600 sec./hr.) = 16.7 Service Units (S.U.). This is the standard duration for our simulations. Further development is intended to increase the complexity of the cells from a single compartment model to a three compartment model, as well as including 4 neuron types in the network. The following represents how we plan to use these resources: Adapting Scripts to Run Under PGENESIS 1,000 S.U. for development of existing models in parallel Genesis Simulation of Anesthetics in Thalamic Network 9,000 S.U. for 30 runs of 100 TC + 100 RE + 100 PE + 100 IN Three Compartment Cells 9,000 = (16.7 S.U.)*(4 Types of Cells)*(3 compartment scaling)*(30 runs)* (1.5 parallel scaling factor) = 1,000 + 9,000 = 10,000 S.U.

该子项目是利用该技术的众多研究子项目之一 资源由 NIH/NCRR 资助的中心拨款提供。子项目和 研究者 (PI) 可能已从 NIH 的另一个来源获得主要资金， 因此可以在其他 CRISP 条目中表示。列出的机构是 对于中心来说，它不一定是研究者的机构。 背景麻醉剂调节许多电压门控和配体门控离子通道，但似乎没有特定的通道是所有麻醉剂的作用位点。即使已知特定麻醉剂的作用位点，离子通道的麻醉调节导致麻醉状态的过程仍有待确定。这项研究的总体目标是使用大规模计算模型来阐明模型神经元对麻醉剂的综合系统级反应。本申请中请求的计算资源生成的数据将用于支持向 NIH 申请额外资源以继续这项工作。为了理解在受体水平和系统水平的麻醉作用之间建立计算桥梁的重要性，有必要认识到每个领域对麻醉剂的根本不同的反应。单一麻醉剂可以调节一个或多个电压门控和/或配体门控离子通道的活性。通常，给定通道的浓度效应曲线相对较浅，中点出现在远高于临床使用浓度的麻醉浓度处。这种离子通道行为的分级调制与麻醉剂在系统层面引起的突然变化形成鲜明对比。脑电图和功能成像研究表明，对于临床有用的麻醉浓度，大脑远未处于静止状态。在受试者群体中，随着麻醉剂浓度的增加，麻醉状态的开始相对突然，并且发生这种情况的浓度大大低于推定离子通道的浓度效应曲线的中点。对麻醉剂的系统级响应也非常复杂，因为它包含在不同麻醉剂浓度下出现的许多不同特征。至少包括健忘症、意识丧失、疼痛刺激的封锁和不动。重要的是，这些效应可以通过针对完全不同的受体组的麻醉剂来产生。所描述的性质不同的行为表明需要将受体水平的麻醉作用与大规模系统水平的行为联系起来的方法。尽管没有特定的网络行为与全身麻醉明确相关，但人们越来越认识到全身麻醉的至少某些方面与大脑产生相干振荡的能力有关。这些振荡被认为起源于丘脑电路，并且这些振荡的同步取决于丘脑和皮质之间建立的相互作用。可以合理地假设，这些对丘脑和皮质的影响可能是麻醉状态的重要组成部分，因为该区域已被证明与睡眠和意识密切相关。此外，我们的初步结果证明了麻醉剂能够在丘脑中继和网状核神经元模型中同步和减缓振荡。这些改变对丘脑其余部分的可能影响及其与皮质的相互作用是尚未得到研究的重要考虑因素。迄今为止，我们已经研究了各种麻醉剂对丘脑网状核、丘脑皮质神经元、海马神经元、快速尖峰中间神经元网络和海兔的单细胞和小网络模型的影响。我们现在寻求扩大这项工作，以纳入在离子通道类型和形态方面更现实的神经元。小网络（每种神经元类型 2 个细胞）包含 4 种不同类型的神经元、皮质中的锥体神经元 (PY) 和中间神经元 (IN) 以及丘脑中的丘脑中继 (TC) 和网状核 (RE) 神经元，初步结果表明，丘脑神经元和皮质神经元之间的反馈对于理解神经元的行为非常重要。 麻醉作用下的神经元。能够创建大型复杂网络（每种神经元类型 100 个细胞）对于辨别麻醉剂在整个系统水平行为和细胞水平上的差异至关重要。目标 以下是我们在 PSC 请求支持下的近期研究目标： 1. 继续开发实际复杂的生物物理神经模型的麻醉调节的详细药理学模型，我们现在将这项工作扩展到丘脑和皮质的神经元。 2. 检查这些网络同步其活动的能力，作为药物干预程度的函数。我们已经构建了多个单个丘脑神经元和网络的模型。迄今为止，这些模型的复杂性相当有限。显然，更高水平的模型细节将允许对迄今为止过于简单化的麻醉效果进行研究。增加离子通道和形态的细节水平对于评估麻醉作用的综合方面至关重要，因为细胞行为的微小而微妙的变化在纳入更大的网络时可能会产生巨大的影响。可以根据已发表的丘脑细胞和电路模型构建在离子通道群、细胞形态和网络相互作用方面具有足够细节的模型。每个细胞由代表其形态的多个隔室组成，包含电压门控和钙控制离子通道、复杂的钙浓度处理，并通过抑制性 GABAA 通道连接到网络。超级计算机的使用 本阶段研究的目标是使用 Parallel Genesis（PSC 已经支持的模拟环境）在并行环境中开发上述网络。在开发该网络的过程中，将对单个神经元和麻醉剂存在下的网络进行初步研究。所需资源 根据对在 Mathematica 中开发并在单个 3.0 GHz 处理器上运行的 100 个单室单元网络的模拟，我们估计需要兆级计算系统 (TCS) 上的 10,000 个服务单元才能实现这些目标。在求解方程时使用隐式数值方法允许步长为 0.01 ms。使用此步长大小，100 个单室单元网络需要大约 20,000 cpu 秒才能运行 10,000 模拟毫秒（或 1*10^6 步）。由于 TCS 由 1 GHz 处理器组成，我们使用近似值，即 10,000 秒仿真需要 20,000 cpu 秒*(3 Ghz/1 GHz)/(3,600 秒/小时) = 16.7 个服务单元 (S.U.)。这是我们模拟的标准持续时间。进一步的开发旨在增加细胞的复杂性，从单室模型增加到三室模型，并在网络中包含 4 种神经元类型。以下介绍了我们计划如何使用这些资源： 调整脚本以在 PGENESIS 1,000 S.U. 下运行用于在丘脑网络 9,000 S.U. 中并行开发现有模型 Genesis 麻醉模拟对于 100 TC + 100 RE + 100 PE + 100 IN 三室单元 30 次运行 9,000 = (16.7 S.U.)*(4 种单元)*(3 室缩放)*(30 次运行)* (1.5 并行缩放因子) = 1,000 + 9,000 = 10,000 S.U.