Table 1.
Abbreviations used in text.
Fig 1.
A general scheme of an analysis workflow for several MEA recordings.
Fig 2.
Schemes for computing NBs and burst distributions.
A) Creating burst features distributions. First, burst feature histogram is calculated for each electrode (left panel). In this example, it is calculated for burst duration. Next, histograms are normalized to number of values, resulting in a 0–1 value. Last, all electrodes are averaged to create a normalized distribution plot (top right panel) and a cumulative plot (bottom right panel) for each tested treatment. B) Detecting synchronized bursts. Spike data from a raster plot (upper panel) showing the spikes (x-axis) for each active electrode (y-axis) is binned and combined through a weighted Gaussian kernel smoothing method to generate the fraction of active electrodes (blue lines in lower panel). The Otsu global thresholding algorithm [32] is then applied to identify intervals above the threshold (red horizontal line) as synchronized NBs.
Fig 3.
General information of plate activity.
A) A matrix representing a 48-well plate, where each well consists of 16 electrodes. aEs are represented by name consisting of column+row position in each well (“11” for first electrode to “44” for the last). B) MFR (Hz) for aEs per well in a 48-well plate. Title for each well has the genotype label of the well: +/+ (wild-type), -/- (homozygous), +/- (heterozygous), NA (Not available). C) Average MFR of all 16 electrodes of well A6, presented for each second of a 900s recording. D) Average number of electrodes participating in NSs around the peak of a network event. The x-axis represents user-defined time bins (default is 100 ms) before and after a NS peak (-10 equals 1s before the peak). Title for each well consists of well-name and number of identified NSs.
Fig 4.
Comparing features between treatments.
In panels A-C, treatment refers to genotype (heterozygous, +/- and homozygous, -/-). In panels D and E, treatment refers to drug treatment. A) Output tables are printed for every feature, holding average values (in this case, Spike Intensity per aEs) per well (rows) and for every recording analyzed (columns). B) Spike Intensity per aEs for two genotypes by DIV on a subset of five recordings. This graph corresponds to the full aggregated table of A). C) Same as B), but for a subset of eight recordings. D) Comparing the number of aEs for wells treated with a vehicle or one of two channel blocker treatments (0.1nM and 1nM) along a 27 DIV experiment. E) MFR for the same experiment as in D), showing differences in the effects of drug concentrations.
Fig 5.
A) Frequencies (y-axis) of spike-rate in bursts (x-axis) are calculated for four different cell density cultures in a single recording. A user defined maximum of 300 Hz is set. B) Frequencies (y-axis) of burst durations (x-axis) are presented when introducing two different microbial light sensitive membrane proteins vs. untreated cells in a single recording. A user defined maximum for burst duration was set to 3 seconds. C) Combining burst features distributions of number of spikes in bursts from 24 sequential same-plate recordings. Treatments were automatically tested for difference using the EMD test and results were displayed after permutations. D) Cumulative distributions of the same data as in C), treatments were automatically tested for difference using the MD test and results were displayed after permutations.
Fig 6.
Celf4-/- neurons show elevated network synchronization phenotypes.
A) Spike frequency in bursts for Celf4+/+ (blue) and Celf4-/- (red) neural networks. B) Mutual information between electrodes is increased in Celf4-/- networks (red). C) The percent of spikes participating in NBs is increased in Celf4-/- neurons (red), suggesting that ratio of spikes that participate in synchronized network events is higher in the Celf4-/- networks. D) Raster plots show NSs (vertical green lines) and bursts (horizontal red lines) for two adjacent wells: a Celf4-/- well (upper panel) and a Celf4+/+ well (lower panel). Celf4-/- exhibits few sporadic spikes between highly synchronized events, while Celf4+/+ exhibits less network events and more sporadic spikes. Raster plots present 60 seconds of 900 seconds recordings.