CRCNS: Neural Populations, High Frequency Oscillations and EEG seizures

摘要 癫痫是一种神经系统疾病，全世界有6500万人患有癫痫。患有难治性癫痫和病灶清楚的患者，通常可以通过手术切除癫痫组织来成功地治疗。然而，定位的准确性是有限的，特别是在没有明显结构异常的情况下。高频振荡已被认为是一个关键的定位生物标志物。不幸的是，由于缺乏详细监测时间和空间领域的大型网络的测量技术，在理解癫痫电生理动力学方面的进展一直受到阻碍。我们认为，结合已知的病理机制的多尺度网络建模可以为产生高频振荡的情况提供有用的见解。我们提出了一种跨学科的方法，将建模与现有的独特的多尺度记录集相结合，从单个细胞到数百万个细胞的大型网络，在人类大脑皮层网络的癫痫活动期间。 我们的建议集中在几个重要的临床问题上。首先，我们试图准确地定义宽带临床记录中的高频成分如何反映特定于癫痫脑区域的病理活动。其次，我们将确定允许局部病理活动传播的神经元机制，以及这种传播如何影响宽带临床记录的特性。与这一领域的先前文献不同，我们的方法明确地结合了已知的在实验性癫痫发作期间发生的细胞动力学的网络效应，即阵发性去极化和特定细胞群的去极化阻断。在目标1中，我们使用单个网络节点的可扩展和详细的建模来将单个细胞的活动与介观网络规模相关联。这一目标的重点是确定网络交互如何产生高频振荡。在目标2中，我们独立地测试了这样的假设，即这些小的介观网络通过高频分量的传播对宏观电极信号中的高伽马振荡负责。我们通过使用单电池和小网络信号作为其输入的简单线性模型来产生宏电极信号来实现这一点。在目标3中，我们将由于抑制性否决权失败而观察到的癫痫传播与阵发性去极化阻滞相关的网络动力学联系起来。对于所有目标，我们将使用实验性癫痫发作期间在人类大脑皮层切片上记录的数据(单细胞和局部网络活动)以及人体癫痫发作期间的体内微电极记录(多单位和局部网络活动)来验证模拟结果。

Abstract Epilepsy is a neurological disease affecting 65 million people worldwide. Patients with medically intractable seizures and with a clearly localized focus can often be successfully treated by surgical resection of the epileptic tissue. However, localization accuracy is limited, especially if overt structural abnormalities are absent. High frequency oscillations have been proposed as a key localizing biomarker. Unfortunately, progress in understanding the dynamics of seizure electrophysiology has been impeded by lack of measurement techniques for detailed monitoring of large networks both the temporal and spatial domains. We propose that multiscale network modeling incorporating known pathological mechanisms can provide useful insights into the circumstances under which high frequency oscillations are generated. We propose an interdisciplinary approach coupling modeling with an existing dataset of unique multiscale recordings, ranging from single-cell to large networks of millions of cells, during seizure activity in human cortical networks. Our proposal is focused around several important clinical questions. First, we seek to define precisely how high frequency spectral components in broadband clinical recordings reflect the pathological activity specific to seizing brain areas. Second, we will identify the neuronal mechanisms permitting localized pathological activity to propagate, and how this propagation affects the properties· of broadband clinical recordings. In contrast to the prior literature in this area, our approach explicitly incorporates the network effects of cellular dynamics known to occur during experimental seizures, I.e. paroxysmal depolarization and depolarization block of specific cell populations. In Aim 1 we use scalable and detailed modeling of individual network nodes to relate single-cell activity to the mesoscopic network scale. The focus of this aim is to determine how network interactions generate high frequency oscillations. Independently, in Aim 2 we test the hypothesis that these small, mesoscopic networks are responsible, via propagation of the high frequency components, for the high-gamma oscillation in macroelectrode signals. We accomplish this by generating the macroelectrode signal with a simple linear model that employs single-cell and small network signals as its input. In Aim 3 we relate the observed seizure propagation, due to failure of the inhibitory veto, to network dynamics associated with the paroxysmal depolarization block. For all aims we will validate simulated results with data recorded during experimental seizures in slices of human cortex (single-cell and local network activity), and in vivo microelectrode recordings during human seizures (multi-unit as well as local network activities).