Electrical imaging of cortical and spinal axons using high-density microelectrode arrays

Mammalian axons are specialized for transmitting action potentials to targets within the central and peripheral nervous system. A growing body of evidence suggests that, besides signal conduction, axons play essential roles in neural information processing, and their malfunctions are common hallmarks of neurodegenerative diseases. The technologies available to study axonal function and structure integrally limit the comprehension of axon neurobiology. 

Complementary-metal-oxide-semiconductor (CMOS) -based high-density microelectrode arrays (HD-MEAs) have been designed to record extracellular action potentials from neuronal cultures and allow tracking axonal signals across hundreds of microelectrodes. Thanks to a low-noise CMOS-design, HD-MEAs enable detection of APs across entire arbors of cortical axons. HD-MEAs provide noninvasive access to axonal action potentials and impose no constraints on the duration of the recording sessions. Here we present extracellular electrical activity recorded from rat's primary neurons using CMOS-based HD-MEA system (MaxWell Biosystems AG). Datasets comprise electrophysiological recordings obtained from primary motor and cortical neurons. Recorded signals were up-sampled to 200 kHz following the Whitaker-Shannon interpolation formula. Recorded signals were spike-sorted and spike-trigger averaged across an entire array (26400 electrodes). Such signal representation is referred to as ‘axonal electrical image’. 

Methods
Array-wide averaging of voltage traces, synchronized with the initial trace (recorded from the axon initial segment), reveals the spatiotemporal distribution of extracellular action potentials across an entire axonal arbor. The first step in obtaining these data was selecting 9 electrodes that were closest to the putative axon initial segment. We next designed multiple recording configurations covering the entire array - in each configuration, 9 of the 1024 readout channels were routed to the 9 preselected electrodes, and remaining available channels were routed to randomly selected electrodes. Each configuration was used to sample neuronal activity during 2 minutes. Signals recorded by the 9 preselected electrodes in each configuration were sorted using the Spyking-Circus algorithm (1). Timestamps of the sorted signals were used to trigger the averaging of voltage traces recorded across all other electrodes in the array. The spatiotemporal distribution of averaged signals was reconstructed using a custom-designed Matlab code.

1. Yger, P. et al. A spike sorting toolbox for up to thousands of electrodes validated with ground truth recordings in vitro and in vivo. Elife 7, e34518 (2018).

哺乳动物轴突（mammalian axons）是特化用于向中枢和外周神经系统内的靶标传递动作电位（action potentials）的结构。越来越多的证据表明，除信号传导外，轴突在神经信息处理中也发挥着至关重要的作用，其功能异常是神经退行性疾病的常见标志性特征。当前用于完整研究轴突功能与结构的技术，仍制约着我们对轴突神经生物学的认知。

互补金属氧化物半导体（Complementary-metal-oxide-semiconductor, CMOS）基高密度微电极阵列（high-density microelectrode arrays, HD-MEA）被设计用于记录神经元培养物的细胞外动作电位，可实现对数百个微电极上的轴突信号进行追踪。得益于低噪声CMOS设计，HD-MEA能够检测皮层轴突整个分支树的动作电位。该阵列可无创获取轴突动作电位，且对记录时长无任何限制。本研究展示了使用CMOS基HD-MEA系统（MaxWell Biosystems AG）从大鼠原代神经元中记录的细胞外电活动。数据集包含来自原代运动神经元与皮层神经元的电生理记录。所记录的信号按照惠特克-香农插值公式（Whitaker-Shannon interpolation formula）被上采样至200 kHz，随后进行尖峰排序，并在整个阵列（26400个电极）范围内完成尖峰触发平均。这种信号表征方式被称为「轴突电成像（axonal electrical image）」。

方法
与轴突初始段记录的初始迹线同步的全阵列电压迹线平均处理，可揭示整个轴突分支树的细胞外动作电位时空分布。获取此类数据的第一步是选取最接近推定轴突初始段的9个电极。随后我们设计了覆盖整个阵列的多种记录配置：在每种配置中，1024个读出通道中的9个被路由至预先选定的9个电极，其余可用通道则被路由至随机选取的电极。每种配置均用于采集时长为2分钟的神经元活动。每种配置中预先选定的9个电极记录到的信号，通过Spyking-Circus算法（1）完成尖峰排序。排序后信号的时间戳被用于触发对阵列中其余所有电极记录的电压迹线进行平均。平均信号的时空分布通过自定义Matlab代码完成重构。

1. Yger P 等. 适用于数千电极的尖峰排序工具箱：经体外与体内真实记录验证. Elife 7, e34518 (2018).