A simple Ca2+-imaging approach to neural network analyses in cultured neurons

BACKGROUND
Ca2+-imaging is a powerful tool to measure neuronal dynamics and network activity. To monitor network-level changes in cultured neurons, neuronal activity is often evoked by electrical or optogenetic stimulation and assessed using multi-electrode arrays or sophisticated imaging. Although such approaches allow detailed network analyses, multi-electrode arrays lack single-cell precision, whereas optical physiology generally requires advanced instrumentation that may not be universally available.


NEW METHOD
Here we developed a simple, stimulation-free protocol with associated Matlab algorithms that enables scalable analyses of spontaneous network activity in cultured human and mouse neurons. The approach allows analysis of the overall network activity and of single-neuron dynamics, and is amenable to screening purposes.


RESULTS
We validated the new protocol by assessing human neurons with a heterozygous conditional deletion of Munc18-1, and mouse neurons with a homozygous conditional deletion of neurexins. The approach described enabled identification of differential changes in these mutant neurons, allowing quantifications of the synchronous firing rate at the network level and of the amplitude and frequency of Ca2+-spikes at the single-neuron level. These results demonstrate the utility of the approach.


COMPARISION WITH EXISTING METHODS
Compared with current imaging platforms, our method is simple, scalable, accessible, and easy to implement. It enables quantification of more detailed parameters than multi-electrode arrays, but does not have the resolution and depth of more sophisticated yet labour-intensive methods, such as patch-clamp electrophysiology.


CONCLUSION
The method reported here is scalable for a rapid direct assessment of neuronal function in culture, and can be applied to both human and mouse neurons. Thus, the method can serve as a basis for phenotypical analysis of mutations and for drug discovery efforts.


Introduction
Ca 2+ -imaging provides a wide range of applications in neuroscience because it allows monitoring large populations of neurons in vivo and in culture (Grienberger and Konnerth, 2012). In culture, neurons exhibit two primary activity modes, sparse or burst spiking of individual neurons and synchronous repetitive firing of multiple neurons in networks. Depending on the maturity and density of cultured neurons, Ca 2+ -imaging typically reveals some spontaneous neuronal firing (Verstraelen et al., 2014;Verschuuren et al., 2019;Brewer et al., 2009). As neurons form more mature networks, synchronous firing of neurons via network activity increases in parallel with the synapse density, and can be visualized by Ca 2+ -imaging (Brewer et al., 2009). The properties of network activity in cultured neurons depends not only on the extent of synaptic communication between neurons, but also on the electrical properties and axonal and dendritic development of these neurons as well as on their cellular state (Verstraelen et al., 2014;Verschuuren et al., 2019;Brewer et al., 2009;Verstraelen et al., 2018). Therefore, network firing of neurons represents a proxy for a broad range of neuronal properties, including their developmental maturity, synaptic connectivity, intrinsic electrical properties, and signalling status.
A challenge in measuring the network activity of cultured neurons is that synchronicity occurs stochastically and depends on the culture quality (density, age etc) (Verschuuren et al., 2019;Virdee et al., 2017).
As a result, induction of network activity often requires external stimulation, such as electrical or optogenetic stimulation to induce responses (Wardill et al., 2013). Recently, imaging-based platforms have been developed for sophisticated all-optical 'electrophysiology', which allows network-level analyses by applying optical stimuli and monitoring individual cells via optical reporters (Fan et al., 2018;Williams et al., 2019). A major advantage of external stimulation in network analyses is that it enables observation of how synchronized firing is induced, i.e., provides causality. A disadvantage is that the delivery of stimuli introduces additional complexity to the imaging setup, for example the need for electrodes or installation of optical lightguides for stimulation. Moreover, variations in the stimulus efficacy may produce variable responses (Peled et al., 2014). An alternative approach to stimulating network activity in cultured neurons is the application of an activating pharmacological agent. This approach is particularly useful for neurons derived from human embryonic stem (ES) or induced pluripotent stem (iPS) cells that may be difficult to analyse because they are generally less mature than neurons in primary cultures and exhibit less spontaneous network activity (Fan et al., 2018). However, pharmacologically induced responses may differ from more physiological spontaneous responses (Reese and Kavalali, 2016).
The abovementioned issues underscore the need for less sophisticated Ca 2+ -imaging approaches that enable rapid analyses of sparse and synchronous neuronal activity and facilitate studies of stem cell-derived neurons. Here, we document such an approach. We describe a simple Ca 2+ -imaging protocol that induces spontaneous network activity in human and mouse neurons without external stimulation. Moreover, we provide an analysis pipeline that quantifies network activity and singleneuron dynamics. We offer procedures for the culture and imaging of human and mouse neurons, step-wise illustrations of the analysis pipeline, and protocols for the interpretations of the results. Network activity of cultured human neurons was robustly induced in the optimised Ca 2+imaging conditions, rendering it suitable for the study of diseaseassociated mutations in neurons. We tested the sensitivity and . At 0 days in vitro (DIV0), human ES or iPS cells were infected with lentiviruses expressing Ngn2 (from the tetO promoter activated by co-expressed rtTA) and with lentiviruses expressing GCaMP6m (under the synapsin promoter). At DIV4, cells were co-cultured with mouse glia. Cells were maintained by weekly feeding up to DIV35 for maturation of Ca 2+ -imaging. For image acquisition and analysis, an inverted epi-fluorescent microscope with a digital or CCD camera and a 10x or 20x objective connected to a PC or Macintosh equipped with the image analysis software ImageJ and with Matlab are required. A motorized microscope stage with robotic timing of imaging is optional. (B-E) Basic image processing pipeline. (B) Example of time-lapse images (yellow texts "t1, t1, t3 ′′ indicate time series). Images are 'stacked' to provide a maximal intensity projection (C). (C) Maximal intensity projected image superimposed with ROI boxes (coloured circles) at neuron soma. (D) Raw intensity of the Ca 2+ signal are extracted from selected ROIs. (E) Normalization of calcium signal. Coloured lines in both panels indicate individual Ca 2+ traces and the black thick line in the lower panel indicates the average intensity from all traces. (F & F') Illustration of network activity analysis. (F) The average intensity is calculated from Ca 2+ traces in all ROIs in one FOV (black thick line in lower panel in E) and is subsequently used for spike detection and deconvolution using a built-in Matlab spikedetection function. (F') Illustration of quantification of synchronous firing rate. Synchronous spikes are identified using a threshold based on predefined percentile (mean + 1.5-2 X standard deviation, s.d.). Synchronicity rate is quantified as the number of the detected synchronous spikes in one minute. (G-G') Illustration figure of the quantification of single-neuron dynamics. (G) Example of spikes detected in individual Ca 2+ traces. The spike detection and deconvolution is performed using the same method as illustrated in panel F'. (G') Quantification of single-neuron amplitude and frequency. The amplitude of peaks in each trace is defined as the mean value of △F/F 0 of individual peak. The frequency of peaks in each trace is defined as the number of detected peaks in one minute. Scale bars, 25 μm. Objective, 20 × . Z. Sun and T.C. Südhof robustness of this approach using mutants that were previously shown to exhibit defects in synaptic transmission (Patel et al., 2015;Chen et al., 2017), and validated our results using established Ca 2+ -analysis software programs.
Ca 2+ -imaging monitors intracellular Ca 2+ -fluxes that are induced by neuronal depolarization and action potential firing (Smetters et al., 1999). Ca 2+ -fluxes are generally monitored via the fluorescence signal of GCaMP-type Ca 2+ -indicators, which needs to be analysed to extract the features of individual Ca 2+ spikes and to infer the underlying neuronal activity (Grienberger and Konnerth, 2012). The image analysis algorithms we wrote quantifies multiple parameters, including the synchronicity rate of the network activity and the Ca 2+ -spike dynamics of individual neurons as well as the amplitude and frequency of these Ca 2+ -spikes. The goal of the present protocol is to establish an easily accessible and reproducible Ca 2+ -imaging approach that can be scaled up for mutagenesis or drug screening purposes.

Generation of mouse cortical primary neuron with GCaMP expression
Cortical primary neurons from newborn triple neurexin-1/2/3 conditional KO mice (Chen et al., 2017) were seeded on Matrigel-coated coverslips in a 24-well plate in MEM medium (ThermoFisher Scientific). On DIV2, the medium was changed to Neurobasal medium containing 2 μM AraC. On DIV3, half of the medium was exchanged, and the cells were infected with two lentiviruses that encoded (1) GCaMP6m and (2) △Cre-P2A-mCherry or Cre-P2A-mCherry (all driven by the human synapsin promoter). Cultures were maintained through DIV14-16, when Ca 2+ -imaging was performed.

Ca 2+ -imaging
Coverslips were gently washed twice in modified Tyrode solution (25 mM HEPES (Invitrogen), 140 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 10 mM glucose, 2 mM CaCl 2 , 10 μM glycine pH 7.2-7.4, pre-warmed to 37 • C), and placed into a glass-bottom 12-well plate (Cellvis) containing Ca 2+imaging buffer (25 mM HEPES, 140 mM NaCl, 8 mM KCl, 1 mM MgCl 2 , 10 mM glucose, 4 mM CaCl 2 , 10 μM glycine pH 7.2-7.4, pre-warmed to 37 • C). After 1− 2 min equilibration, Ca 2+ -imaging was performed on an inverted epi-fluorescence microscope (Nikon EclipseTS2R, DS-Qi2 digital camera) with the 488 nm filter at room temperature. GCaMP6m fluorescence was recorded for 2− 3 min at a frame rate of 4-10 frames/s. Approximately 800-3000 time-lapse images depending on the frame rate (536 × 536 pixel resolution, 14-bit grayscale depth for human and mouse cultures) were acquired at a 1 x digital zoom using either a 10x or 20x objective. Acquisition time of each frame can vary between 0.01− 0.1 s, depending on the camera used for detection and the exposure time subjected to the brightness of GCaMP expression; long-term recording can cause photobleaching and phototoxicity. Per coverslip, 2− 3 fields were imaged and a minimal of 3 coverslips were recorded for each biological batch. For each batch, all images were acquired using the same light intensity and exposure time. Time-lapse imaging was performed in a field of view (FOV) containing confluent neuron populations with non-overlapping soma.

Image pre-processing and network and spike analysis algorithms
Time-lapse image files were converted to .tiff /.tif format using ImageJ (https://imagej.nih.gov/ij/). Quantifications were performed using home-written Matlab functions and scripts included in Appendix A and B. Datasheets containing raw Ca 2+ traces are used as inputs for all algorithms. The datasheets should be organised such that rows correspond to times and columns to individual cells. To determine the network activity of the neuronal culture based on the synchronous firing rate of the entire cell population in the FOV, follow Section 3.2 (Appendix B3); to determine the single-neuron activity in terms of amplitude and frequency, follow Section 3.3 (Appendix B4). An example for the variable definition and the output can be found in the instruction (Appendix A). For measuring network activity, the algorithm in Appendix B3 provides the quantification of synchronous firing rate and synchronous peak amplitude as two outputs, and we only used the former for illustration here.

Statistical analyses
Matlab (MathWorks) was used to process images and quantifications from time-lapse tiff files. Data were exported from Matlab to. csv files which were imported for plotting into R software (http://www.r-pr oject.org, version 1.2.5001; R packages ggplot2, tidyverse, lubridate). The boxplots were plotted using Prism (GraphPad). The upper and lower hinges of the box indicate the 25th and 75th percentiles. The upper whisker and the lower whiskers extend from the hinge of the box indicate the minimum and maximum values. The thick line inside the boxplot indicates the median value and individual data points are indicated as grey circles. All P values were calculated using Student's t-test with P < 0.05 considered significant, unless otherwise stated.

Imaging of neurons and analyses workflow
Human neurons expressing GCaMP6m and co-cultured with mouse Z. Sun and T.C. Südhof  glia were imaged in a standard epi-fluorescence microscope, and the GCaMP6m fluorescence signal was recorded as a function of time (Fig. 1A). Regions of interests (ROIs) containing the soma region of neurons to be analysed were manually selected using the Matlab algorithm provided in Appendix B1. The algorithm first projects the timelapse stack to its maximum intensity image to allow a user to find and select a neuron (Fig. 1B). Upon selection of the center of the soma of a neuron, a circular ROI mask is superimposed onto the selected neuron (Fig. 1C). The size of the ROI can be specified in the algorithm according to the acquisition resolution (see Appendix A for detailed instructions and examples of variable definitions). Each ROI is defined as a node to extract the raw Ca 2+ intensity values from the ROI (the mean value of the GCaMP fluorescence intensity), which was used for all further analyses. After selection of desired ROIs, traces of the GCaMP6m fluorescence intensity per ROI over time were generated (Fig. 1D). The intensity profile was automatically saved into an .mat file named after the image name in an automatically created folder. The raw data are listed in the matrix ROI_intensity1, in which each row corresponds to a single frame and each column corresponds to an ROI and can be found by loading the .mat file in the command window. To plot the raw intensity of the Ca 2+ signal, load the ROI_intensity1 matrix generated from the above step and run the script in Appendix B2. This will return a figure showing all traces (upper panel in Fig. 1E). The baseline fluorescence level (F 0 ) of each trace is calculated as the average basal fluorescence intensity during the first, middle and last 100 frames of each trace (Patel et al., 2015;Piatkevich et al., 2019). The Ca 2+ signal intensity of each raw trace is then normalized by calculating △F/F 0 as the ratio of the increase in fluorescence (△F) to the baseline (F 0 ) (lower panel in Fig. 1E).

Analysis of network activity by inducing synchronous activity
In Ngn2-induced cultured human neurons trans-differentiated from ES cells, individual neurons exhibited occasional spontaneous Ca 2+spikes ( Fig. 2A-A'), but synchronous firing of two or more neurons was exceedingly rare ( Fig. 2A-A', C). Therefore we stimulated neuronal firing by increasing the concentrations of Ca 2+ and K + in the imaging solution to 4 mM and 8 mM, respectively. Under this condition, frequent synchronous bursting events were observed in cultured neurons ( Fig. 2B-B', D). To quantify the synchronous firing activity, we analysed network-wide synchronous Ca 2+ spikes. First, we obtained the global intensity profile of Ca 2+ transients by calculating the average intensity of the Ca 2+ signal from all ROIs in one FOV (thick black line in lower panel in Fig. 1E, Fig. 1F). Second, we used a built-in Matlab function for spike detection and defined Ca 2+ -spikes as all spikes exceeding a defined value above the average resting signal, usually spikes above 1.5-2.0 X the standard deviation of the background signal. Synchronous Ca 2+ peaks were quantified as all peaks detected by this method, and the synchronous firing rate was defined as the number of synchronous Ca 2+ peaks per minute (Fig. 1F'). The synchronicity rate and the amplitude of synchronous peaks were quantified by the script in Appendix B3. Here, we only include the synchronicity rate for discussion and illustration purposes.
We then compared the network activity of the cultured neurons in normal Tyrode medium versus the imaging solution that contains elevated Ca 2+ (4 mM) and K + (8 mM) concentrations. We found that the synchronous firing rate was significantly higher in the imaging solution than in Tyrode medium (Fig. 2G). In order to understand the variability of the Ca 2+ signals observed in cultured human neurons (iN cells), we quantified the coefficient of variation of the synchronous firing rate (defined as the ratio of the standard deviation of the synchronous firing rate to its mean value (i.e., s.d. / mean in a given FOV). Comparison of the coefficient of variation between neurons cultured in normal Tyrode medium versus the imaging solution did not uncover a significant difference (Fig. 2G'). Ngn2-induced human neurons are composed almost exclusively of excitatory neurons (Zhang et al., 2013). As a result, addition of CNQX (10 μM), an antagonist of AMPA-type glutamate receptors, suppressed the network activity of human neurons, although isolated Ca 2+ spikes could still be observed (Fig. 2E). Treatment of the cultured neurons with tetrodotoxin (TTX; 1 μM), a Na + -channel blocker, also abolished synchronous network activity of neurons, as evidenced by a complete loss of Ca 2+ spikes, which would be expected for network activity driven by action potential firing (Fig. 2F). Note that TTX also ablates sparse spontaneous Ca 2+ spikes, supporting the notion that spontaneous non-synchronous Ca 2+ spikes also depend on action potential firing (Fig. 2F).

Quantification of single-neuron amplitude and frequency
To analyse the single-neuron Ca 2+ dynamics, we computed the amplitude and frequency of Ca 2+ -spikes, which are common parameters quantified in Ca 2+ imaging analyses (Fig. 1G-G'). We used the spike detection method described above (Fig. 1G). The single-neuron amplitude was calculated from all detected spikes within an individual Ca 2+ trace after normalization (△F /F 0 , Fig. 1G') (Grienberger and Konnerth, 2012). The single-neuron frequency was calculated as the number of detected spikes in each Ca 2+ trace per minute (Fig. 1G'). The quantification program is provided in Appendix B4. Running the program produces two graphs: a plot of detected spikes (example in Fig. 1G) and a plot of the Ca 2+ traces after normalization (example in the lower panel of Fig. 1E). We compared the single-neuron Ca 2+ -spike amplitude and frequency of cultured neurons in normal and modified Tyrode solution. Both the amplitude and frequency were increased by the 4 mM Ca 2+ and 8 mM K + concentrations in the modified Tyrode solution as expected (Fig. 2H & I). The neuron-to-neuron variation was quantified as the coefficient of variation of the amplitude and frequency (i.e., s.d. / mean per neuron), revealing that the coefficient of variation of both decreased in the modified Tyrode solution as expected (Fig. 2H' & I').
Analysis of the neuronal Ca 2+ spikes demonstrated that the heterozygous Munc18-1 deletion caused a ~50% suppression of neuronal network activity, which was quantified as the synchronous firing rate (Fig. 3D). At the single-neuron level, the Munc18-1 deletion produced a ~30% decrease in Ca 2+ signal amplitude (Fig. 3G) and a significant increase in Ca 2+ signal frequency (Fig. 3H). It was noteworthy that the Ca 2+ spikes that we observed with an increased frequency in Munc18-1 mutant neurons are asynchronous (i.e., are not part of synchronous network activity; Fig. 3C'). The increase in the Ca 2+ spike frequency despite a decrease in network activity is due to the emergence of aberrant Ca 2+ spikes in the mutant neurons upon suppression of synchronous (caption on next page) Z. Sun and T.C. Südhof firing, consistent with the impairment in synaptic communication induced by the Munc18-1 deletion. Moreover, we observed a modest decrease of the coefficient of variation of the single-neuron Ca 2+ spike amplitude in the Munc18-1 mutant neurons compared with the control, presumably because the Ca 2+ spikes are reliably smaller in amplitude (Fig. 3G'). No significant difference in the coefficient of variation of the synchronous firing rate (Fig. 3D') and the single-neuron frequency (Fig. 3H') was detected. Together, these data suggest that the Munc18-1mutant neurons exhibit a major impairment in network properties and in single-neuron Ca 2+ dynamics. These results agree well with the previous finding that the heterozygous Munc18-1 deletion in human neurons causes a large decrease in synaptic strength (Patzke et al., 2015).
To validate the use of our analyses of the synchronous firing rate, we furthermore analysed the raw data by two alternative software packages, FluoroSNNAP (Patel et al., 2015) and PeakCaller (Artimovich et al., 2017). Both software quantify the synchronization index as a measure of network activity. Analysis of our data using FluoroSNNAP (Fig. 3E) and PeakCaller (Fig. 3 F) confirmed a decrease in the synchronization index in Munc18-1-mutant neurons compared to controls (Fig. 3D). Moreover, we also used the single-neuron module of Fluo-roSNNAP to quantify the Ca 2+ signalling amplitude and frequency. Consistently, these analysis showed that Munc18-1-mutant neurons exhibit a reduction in the amplitude and an increase in the frequency compared to control neurons (Fig. 3I & J), without a difference in the coefficient of variation of the amplitude (Fig. 3I') and frequency (Fig. 3J'). Together, these results validate the parameter quantifications we propose with our approach.

Validation of the proposed Ca 2+ imaging protocol using mouse neurons with a conditional deletion of all neurexins
In a second validation experiment, we examined cortical neurons cultured from newborn mice with a conditional deletion of all three neurexins, referred to as pan-neurexin deletion (Chen et al., 2017). Neurexins are presynaptic adhesion molecules that act as key regulators of synapse properties (Sudhof, 2017). Deletion of all neurexins has no effect on synapse numbers, but causes profound changes in the efficacy of synaptic transmission (Chen et al., 2017;Luo et al., 2020). We infected the cultured neurons at DIV3 with lentiviruses expressing GCaMP6m and △Cre (control) or Cre (deletion), and analysed them by Ca 2+ imaging at DIV14-16 as described in Section 2.2 (Fig. 4A).
The pan-neurexin deletion significantly decreased neuronal network activity, as shown by a decreased synchronous firing rate (Fig. 4B-D). The reduced network activity in triple neurexin-1/2/3 cKO neuron cultures was further confirmed by the decrease of synchronization index measured by FluoroSNNAP (Fig. 4E) as well as by PeakCaller (Fig. 4F). At the single-neuron level, interestingly, the pan-neurexin deletion caused a significant, somewhat paradoxical increase of the amplitude of Ca 2+ spikes (Fig. 4G), which was also confirmed by the single-neuron amplitude quantification using FluoroSNNAP (Fig. 4I). Opposite to the amplitude change, however, the frequency of Ca 2+ spikes in panneurexin deletion neurons significantly decreased as quantified using our program (Fig. 4H) and using FluoroSNNAP (Fig. 4J). The variability of the network activity was examined by quantification of the coefficient of variation of synchronous firing rate, but no differences were observed between control and pan-neurexin deletion neurons (Fig. 4D'). The neuron-to-neuron variability was compared by quantification of the coefficient of variation of single-neuron amplitude and frequency using our program (Fig. 4G' & H') and using FluoroSNNAP (Fig. 4I' & J'). Despite a slight increase of the coefficient of variation of the amplitude in the pan-neurexin deletion neurons measured by our program (Fig. 4G'), all other data detected little or no significant difference in pan-neurexin deletion neurons compared to controls (Fig. 4H', Fig. 4I' & J').

Discussion
In this study, we developed a Ca 2+ imaging approach that enables economical and rapid functional analyses of the network activity and Ca 2+ dynamics of cultured human and rodent neurons. The mechanisms underlying neuronal network activity have been studied extensively by patch-clamp electrophysiology, which provides an exquisite amount of precise information on individual neurons, but is only suitable for lowthroughput applications. We demonstrated that the simple approach for the analysis of neuronal network activity and Ca 2+ signalling that we describe yields reproducible quantitative parameters. Among others, these parameters include the synchronicity of the overall network activity and the amplitude and frequency of Ca 2+ spikes of individual neurons in larger populations. Since the neuronal network activity and the Ca 2+ signals depend on a large number of neuronal properties, including the synaptic connectivity and the intrinsic electrical properties of neurons, the parameters that emerge from analyses of this network activity cannot be directly ascribed to a particular neuronal property, but are indicators of overall neuronal function. These parameters provide a useful measure for screening purposes but cannot be mechanistically interpreted without additional experimentation.
In the approach proposed here, we induced synchronous network activity in cultured human and mouse neurons by raising the ambient Ca 2+ and K + concentrations, which enabled routine analyses of network activity without external stimulation. The increased Ca 2+ and K + concentrations stimulate network activity because they elevate the neurotransmitter release probability and excitability of neurons. Although these conditions are not physiological, they trigger normal action potentials in cultured neurons, which in themselves are not a physiological preparation, but represent a reduced system amenable to mechanistic analyses. We validated our Ca 2+ -imaging approach with human and mouse neurons that carry defined conditional mutations in synaptic genes, human heterozygous Munc18-1 cKO neurons and mouse homozygous pan-neurexin cKO neurons. The results we obtained with these mutant neurons confirmed that the simple Ca 2+ -imaging approach we tests were used for all groups (*, p < 0.05; ***, p < 0.001; ****, p < 0.0001; n.s., not significant). Scale bars, 25 μm. Objective, 20 × .

Z. Sun and T.C. Südhof
describe is not only capable of robustly detecting mutant phenotypes, but can also differentiate between phenotypes (Figs. 3 and 4). Moreover, we validated our approach by analysing the Ca 2+ -imaging data we obtained with the mutant neurons using two publicly available software packages, FluoroSNNAP (Patel et al., 2015) and PeakCaller (Artimovich et al., 2017). This parallel analysis demonstrated that our approach is, at least in some instances, more sensitive and versatile than previously described methods (Figs. 3 and 4; see also Table 1).
In our approach, we adopted a semi-automated image analysis approach. For the manual part, users select neurons of interest to be analysed (Fig. 1B), while the data extraction and computation of the synchronicity rate, single-neuron amplitude and frequencies, coefficients of variation are automatic. To test the robustness of the ROI selection as well as the parameter quantification methods in our protocol, we analysed our data with two independently alternative software packages mentioned above, FluoroSNNAP (Patel et al., 2015) and PeakCaller (Artimovich et al., 2017). FluoroSNNAP provides automated ROI detection using time-lapse files as input and PeakCaller uses Ca 2+ traces as input. Analysis of network activity as well as single-neuron dynamics using these software showed consistent differences between control and mutant neurons similar to those revealed by our approach (Fig. 3D-J', Fig. 4D-J', Supplementary Fig. 1E-H, Supplementary Fig. 2E-H). Additionally, because raw time-lapse movies were imported into FluoroSNNAP for analysis ( Supplementary Fig. 1A-B', Supplementary Fig. 2A-B') while ROIs containing raw Ca 2+ traces extracted using our programs were imported into PeakCaller for analysis ( Supplementary Fig. 1C-D, Supplementary Fig. 2C-D), the consistent results of the synchronization index measured from the two software ( Fig. 3E & F, Fig. 4E & F) suggested that the ROI detection method we proposed is robust. Overall, comparisons of the parameter values obtained (Figs. 3 and 4) suggest that our approach compares favourably with that of other software.
At present, Ca 2+ -imaging analysis packages are aiming to provide evaluations of diverse features at multiple levels and at different scales (see a partial list of available packages in Table 1) (Patel et al., 2015;Kaifosh et al., 2014;Giovannucci et al., 2019;Oh et al., 2019;Zhou et al., 2018). The increasing size of the imaging datasets demands that analyses should be fully automated, ultimately towards standardization and automation to leverage phenotypic assays especially in human neurons that can be used for drug screening purposes for neurodegenerative or neuropsychiatric disorders (Grienberger and Konnerth, 2012;Verstraelen et al., 2014;Verschuuren et al., 2019;Verstraelen et al., 2018;Mackay et al., 2016). Our Ca 2+ imaging approach is relatively simple compared to these more sophisticated packages. However, our approach uses basic instrumentation and is thus useful for non-industrial applications, and can be readily expanded to monitor additional parameters. For example, our protocol examines network activity as the frequency of synchronous peaks, yet in some cases the percentage of synchronously firing neurons might decrease without changes of the synchronous firing frequency (Verschuuren et al., 2019). To address this possibility, the algorithms we describe could be readily expanded to measure the percentage of neurons contributing to synchronous firing of the population. Furthermore, we quantify the Ca 2+ signal from the soma of a neuron, which integrates the signals of hundreds of synaptic inputs. To measure Ca 2+ signals directly at the level of synapses, GCaMPs could be targeted to pre-or to postsynaptic specializations by fusion to synaptophysin (Dreosti et al., 2009) or to PSD95 (Reese and Kavalali, 2016), respectively. Thus, the present study aims to enable a straightforward implementation of measurements of Ca 2+ signals in neuronal populations. By helping to implement such measurements as an important basic tool, our approach aims to facilitate widespread application of such measurements and to allow their expansion to other neuronal parameters as needed.

Credit author statement
Z.S. and T.C.S. jointly conceived the experiments, Z.S. performed the experiments, and Z.S. and T.C.S. analysed the results and wrote the manuscript.

Declaration of Competing Interest
The authors declare no competing of interests. Z. Sun and T.C. Südhof