Computer simulations of populations of mammalian motor units

描述（由申请人提供）：运动单位是运动输出的基本元件，由运动神经元及其轴突支配的肌纤维组成。肌纤维抽搐通常与运动神经元动作电位1比1，因此运动单位是一个单一的功能实体。尽管如此，大多数研究，无论是实验还是模拟，都倾向于关注运动神经元或肌肉。这种焦点的分离极大地限制了对正常和病理状态下运动外流的理解。为了弥合这一差距，该提案旨在开发一个高度逼真的和彻底验证的计算机模拟的一组运动单位的单一肌肉。我们专注于猫的后肢伸肌，这是最完整的实验数据库。限制以前的努力在模拟运动单位的关键问题是缺乏了解的神经调质的突触输入转换为运动神经元的尖峰输出的影响。我们实验室和其他许多实验室的系统研究现在已经确定了这些神经调质效应，并发现它们在影响运动神经元兴奋性方面非常强大。其中最有效的是5-羟色胺（5-HT）和去甲肾上腺素（NE），它们在脊髓中由脑干中的轴突释放。5 HT和NE促进运动神经元树突中的持续性内向电流（PIC），其然后将突触输入放大多达5倍。我们已经成功地开发了一个初始模型的运动神经元与PIC。此外，我们已经成功地开发了一个很好的肌肉模型代表肌肉单位。在拟议工作的目标1中，这些初始模型被进一步开发，根据实验数据进行了仔细验证，并扩展到准确代表完整运动池和肌肉所需的200多个成员。在目标2中，我们使用模拟池/肌肉来研究运动外流的结构，重点关注神经调节输入如何改变整体系统增益以及影响运动神经元放电模式和噪声波动等细节。该模型在运动控制的广泛模拟中具有很大的潜力，但是涉及多组神经元和多块肌肉的模拟需要计算效率。因此，在目标3中，我们研究了几种不同的方法来简化100个运动单元的完整集合，以实现计算速度的大幅提高。这些目标的成功完成将提供一个运动输出的生物现实模型，可用于运动神经控制的广泛计算研究。这些模拟可用于深入了解运动命令的结构，并识别脊髓损伤等疾病状态下运动系统的缺陷。从长远来看，我们希望开发一个用户界面，让电机控制社区广泛使用。

DESCRIPTION (provided by applicant): The motor unit is the fundamental element of motor output and consists of a motoneuron and the muscle fibers that its axon innervates. Muscle fiber twitches are normally 1-to-1 with motoneuron action potentials and thus the motor unit is a single functional entity. Despite this, most studies, both experimental and simulation, tend to focus either on motoneurons or on muscle. This separation of focus has sharply limited understanding of motor outflow in both normal and pathological states. To bridge this gap, this proposal seeks to develop a highly realistic and thoroughly validated computer simulation of the set of motor units for a single muscle. We focus on hindlimb extensors in the cat, for which the most complete experimental database is available. The key issue limiting previous efforts at simulating motor units is the lack of understanding of the effects of neuromodulators on conversion of synaptic input to spiking outputs in motoneurons. Systematic studies in our lab and many others have now identified these neuromodulator effects, and found them to be remarkably strong in influencing motoneuron excitability. The most potent of all are serotonin (5HT) and norepinephrine (NE), which are released in the spinal cord by axons originating in the brainstem. 5HT and NE facilitate persistent inward currents (PICs) in the dendrites of motoneurons, which then amplify synaptic input by as much as 5-fold. We have successfully developed an initial model of the motoneuron with PICs. Moreover, we have successfully developed a good muscle model for representing muscle units. In Aim 1 of the proposed work, these initial models are further developed, carefully validated against experimental data and expanded into the set of more the 200 members needed to accurately represent the full motor pool and muscle. In Aim 2, we use the simulated pool/muscle to investigate the structure of motor outflow, focusing on how neuromodulatory inputs alter overall system gain as well as influence details like motoneuron firing patterns and noise fluctuations in force. This model has great potential for use in a wide range of simulations of motor control, but simulations that involved multiple sets of neurons and multiple muscles require computational efficiency. Thus in Aim 3, we investigate several different approaches for simplifying the full set of 100s of motor units to achieve great increases in computational speed. Successful completion of these aims will provide a biologically realistic model of motor output that can be used in a wide range of computational studies of the neural control of movement. These simulations can be used to generate deep insights into the structures of motor commands and to identify deficits in motor systems in disease states like spinal injury. In the long term, we hope to develop a user interface to allow widespread use by the motor control community.