A toolbox for navigating and analyzing the spatiotemporal properties of retinal waves
The precise development of the visual system is driven by retinal waves, which are bursts of spontaneous activity that propagate across retinal neurons in a wave-like fashion. In mice, retinal waves begin embryonically and continue until eye opening at the end of the second postnatal week. During this time, the mechanisms for generating and propagating retinal waves are changing, thus causing retinal waves to exhibit highly dynamic spatiotemporal properties from one day to the next. Critically, the spatiotemporal properties of retinal waves have been shown to instruct the development of the visual system, including eye-specific segregation, retinotopic mapping, direction selectivity, and potentially retinal vascularization. Currently, there is no method for the automatic detection and high-throughput quantitative analysis of the spatiotemporal properties of retinal waves. To overcome this barrier, we developed WaveMiner, an automated, high-throughput toolbox for detecting, segregating, and quantifying retinal waves from microelectrode-array (MEA) and calcium-imaging recordings. After first validating our toolbox, we use it to uncover novel dynamic spatiotemporal properties of retinal waves in the first two postnatal weeks. We also use this toolbox to analyze ultra long physiological recordings, revealing that waves exhibit both stable and dynamic spatiotemporal properties on an hourly basis. Finally, we demonstrate that this toolbox can detect waves in the presence of pharmacological agents that increase the baseline firing of neurons, enabling the discovery of novel factors that perturb the spatiotemporal properties of retinal waves and visual development. In summary, WaveMiner is a platform to standardize the detection and quantification of retinal waves across recording modalities and experimental conditions, enabling novel discoveries about their dynamic spatiotemporal properties and the factors that govern them.