Barbu, A. | Search bioRxiv

bioRxiv2026

GABAergic TH2 Amacrine Cells Participate in Spontaneous Wave Activity in the Developing Retina

Amacrine cells (ACs) are retinal interneurons that regulate synaptic transmission from bipolar cells to retinal ganglion cells (RGCs) and play essential roles in object motion detection, contrast sensitivity, and light adaptation. A subtype of GABAergic ACs identified using a tyrosine hydroxylase (TH) promoter-driven green fluorescent protein (GFP) mouse line has been termed TH2 amacrine cells (TH2-ACs). Although TH2-ACs contribute to the feature selectivity of object-motion signals in the adult retina, their functional properties during early postnatal development remain unclear. Using genetic mouse models, electrophysiology, immunohistochemistry, and calcium imaging, we show that TH2-ACs exhibit spontaneous rhythmic depolarizations during development. In the first postnatal week, these depolarizations were abolished by acetylcholine receptor antagonists, indicating that TH2-ACs are excited by starburst amacrine cells (SACs) via spontaneous cholinergic retinal waves. During the second postnatal week, rhythmic depolarizations persisted but were blocked by glutamate receptor antagonists, demonstrating that TH2-ACs are subsequently driven by bipolar cells through glutamatergic waves. Calcium imaging further revealed that this activity propagates across the TH2-AC network in a wave-like manner, potentially resulting in spatially and temporally patterned GABA release. Pharmacological blockades of GABAA receptors significantly enhanced glutamatergic wave activity in SACs and RGCs, indicating that GABAergic signaling from TH2-ACs participates in exerting inhibitory control over retinal waves. Together, these findings identify TH2-ACs as active participants in the development of retinal wave circuits and suggest that this participation via GABA signaling could contribute to activity-dependent refinement of retinal circuits underlying object motion processing. Key pointsO_LITH2 amacrine cells are excited by starburst amacrine cells through cholinergic retinal wave activity during the first postnatal week. C_LIO_LIDuring the second postnatal week, TH2 amacrine cells are driven by bipolar cells via glutamatergic retinal wave activity. C_LIO_LIThe dense dendritic arborization of TH2 amacrine cells enables their participation in the propagation of both cholinergic and glutamatergic waves. C_LIO_LIWave-like GABA release from TH2 amacrine cells contributes to the modulation of retinal wave activity through activation of GABAA receptors. C_LI

neuroscience↗

bioRxiv2021

Neural Regression, Representational Similarity, Model Zoology & Neural Taskonomy at Scale in Rodent Visual Cortex

How well do deep neural networks fare as models of mouse visual cortex? A majority of research to date suggests results far more mixed than those produced in the modeling of primate visual cortex. Here, we perform a large-scale benchmarking of dozens of deep neural network models in mouse visual cortex with both representational similarity analysis and neural regression. Using the Allen Brain Observatorys 2-photon calcium-imaging dataset of activity in over 6,000 reliable rodent visual cortical neurons recorded in response to natural scenes, we replicate previous findings and resolve previous discrepancies, ultimately demonstrating that modern neural networks can in fact be used to explain activity in the mouse visual cortex to a more reasonable degree than previously suggested. Using our benchmark as an atlas, we offer preliminary answers to overarching questions about levels of analysis (e.g. do models that better predict the representations of individual neurons also predict representational similarity across neural populations?); questions about the properties of models that best predict the visual system overall (e.g. is convolution or category-supervision necessary to better predict neural activity?); and questions about the mapping between biological and artificial representations (e.g. does the information processing hierarchy in deep nets match the anatomical hierarchy of mouse visual cortex?). Along the way, we catalogue a number of models (including vision transformers, MLP-Mixers, normalization free networks, Taskonomy encoders and self-supervised models) outside the traditional circuit of convolutional object recognition. Taken together, our results provide a reference point for future ventures in the deep neural network modeling of mouse visual cortex, hinting at novel combinations of mapping method, architecture, and task to more fully characterize the computational motifs of visual representation in a species so central to neuroscience, but with a perceptual physiology and ecology markedly different from the ones we study in primates.

neuroscience↗