Human neuronal networks on micro-electrode arrays are a highly robust tool to study disease-specific genotype-phenotype correlations in vitro
Description
Micro-electrode arrays (MEAs) are increasingly used to characterize neuronal network activity of human induced pluripotent stem-cell (hiPSC)-derived neurons. Despite their gain in popularity, MEA recordings from hiPSC-derived neuronal networks are not always used to their full potential in respect to experimental design, execution and data analysis. Therefore, we benchmarked the robustness and sensitivity of MEA-derived neuronal activity patterns derived from ten healthy individual control lines. We provide recommendations on experimental design and analysis to achieve standardization. With such standardization, MEAs can be used as a reliable platform to distinguish (disease-specific) network phenotypes. In conclusion, we show that MEAs are a powerful and robust tool to uncover functional neuronal network phenotypes from hiPSC-derived neuronal networks, and provide an important resource to advance the hiPSC field towards the use of MEAs for disease-phenotyping and drug discovery.
Files
Steps to reproduce
The dataset contains two zipped folders. The folder named Peak Trains.zip contains the filtered data export from each control and patient line discussed in the paper. Inside the patient or control folder you find all the recorded MEAs from that IPS line. Inside the well folder, you find the peak trains, which are .mat files containing the timing (collum A) and amplitude (collum B) of each detected spike for that specific electrode in that specific well (each MEA has 12 electrodes). All MEAs were recorded for 10 minutes (i.e. 600 sec). The raw signal was sampled at 10 kHz and filtered with a high-pass 2nd order Butterworth filter with a 100 Hz cut-off frequency and a low-pass 4th order Butterworth filter with a 3500 Hz cut-off frequency. The noise threshold for individual spike detection was set at ±4.5 standard deviations. For analysis settings, see paper. Control 6 was compared on human (hLAM) and mouse laminin (mLAM). The basic parameters and CV-NIBI were calculated using the Multiwell GUI and Calculate_CV.m scripts based on excel tables exported from Multiwell Analyser. Connectivity analysis was performed using previously published Spicodyn software (Pastore et al., 2018a; Bastle and Maze, 2019). Analysis of the peak trains can be performed using the previously published software called Spycode (Bologna et al., 2010). Codes for analyzing burst shapes is available upon request (Monica Frega).