Circuit mechanisms underlying embryonic retinal waves

Emergence of input selective recurrent dynamics via information transfer maximization

Network structures of the brain have wiring patterns specialized for specific functions. These patterns are partially determined genetically or evolutionarily based on the type of task or stimulus. These wiring patterns are important in information processing; however, their organizational principles are not fully understood. This study frames the maximization of information transmission alongside the reduction of maintenance costs as a multi-objective optimization challenge, utilizing information theory and evolutionary computing algorithms with an emphasis on the visual system. The goal is to understand the underlying principles of circuit formation by exploring the patterns of wiring and information processing. The study demonstrates that efficient information transmission necessitates sparse circuits with internal modular structures featuring distinct wiring patterns. Significant trade-offs underscore the necessity of balance in wiring pattern development. The dynamics of effective circuits exhibit moderate flexibility in response to stimuli, in line with observations from prior visual system studies. Maximizing information transfer may allow for the self-organization of information processing functions similar to actual biological circuits, without being limited by modality. This study offers insights into neuroscience and the potential to improve reservoir computing performance.

Development and organization of the retinal orientation selectivity map

Orientation or axial selectivity, the property of neurons in the visual system to respond preferentially to certain angles of visual stimuli, plays a pivotal role in our understanding of visual perception and information processing. This computation is performed as early as the retina, and although much work has established the cellular mechanisms of retinal orientation selectivity, how this computation is organized across the retina is unknown. Using a large dataset collected across the mouse retina, we demonstrate functional organization rules of retinal orientation selectivity. First, we identify three major functional classes of retinal cells that are orientation selective and match previous descriptions. Second, we show that one orientation is predominantly represented in the retina and that this predominant orientation changes as a function of retinal location. Third, we demonstrate that neural activity plays little role on the organization of retinal orientation selectivity. Lastly, we use in silico modeling followed by validation experiments to demonstrate that the overrepresented orientation aligns along concentric axes. These results demonstrate that, similar to direction selectivity, orientation selectivity is organized in a functional map as early as the retina.

Conservation of neuron-astrocyte coordinated activity among sensory processing centers of the developing brain

Afferent neurons in developing sensory organs exhibit a prolonged period of burst firing prior to the onset of sensory experience. This intrinsically generated activity propagates from the periphery through central processing centers to promote the survival and physiological maturation of neurons and refine their synaptic connectivity. Recent studies in the auditory system indicate that these bursts of action potentials also trigger metabotropic glutamate receptor-mediated calcium increases within astrocytes that are spatially and temporally correlated with neuronal events; however, it is not known if this phenomenon occurs in other sensory modalities. Here we show using in vivo simultaneous imaging of neuronal and astrocyte calcium activity in awake mouse pups that waves of retinal ganglion cell activity induce spatially and temporally correlated waves of astrocyte activity in the superior colliculus that depend on metabotropic glutamate receptors mGluR5 and mGluR3. Astrocyte calcium transients reliably occurred with each neuronal wave, but peaked more than one second after neuronal events. Despite differences in the temporal features of spontaneous activity in auditory and visual processing regions, individual astrocytes exhibited similar overall calcium activity patterns, providing a conserved mechanism to synchronize neuronal and astrocyte maturation within discrete sensory domains.

Development and Organization of the Retinal Orientation Selectivity Map

Orientation or axial selectivity, the property of neurons in the visual system to respond preferentially to certain angles of a visual stimuli, plays a pivotal role in our understanding of visual perception and information processing. This computation is performed as early as the retina, and although much work has established the cellular mechanisms of retinal orientation selectivity, how this computation is organized across the retina is unknown. Using a large dataset collected across the mouse retina, we demonstrate functional organization rules of retinal orientation selectivity. First, we identify three major functional classes of retinal cells that are orientation selective and match previous descriptions. Second, we show that one orientation is predominantly represented in the retina and that this predominant orientation changes as a function of retinal location. Third, we demonstrate that neural activity plays little role on the organization of retinal orientation selectivity. Lastly, we use in silico modeling followed by validation experiments to demonstrate that the overrepresented orientation aligns along concentric axes. These results demonstrate that, similar to direction selectivity, orientation selectivity is organized in a functional map as early as the retina.

One Sentence Summary

Development and organization of retinal orientation selectivity.

Sequentially activated discrete modules appear as traveling waves in neuronal measurements with limited spatiotemporal sampling

Numerous studies have identified traveling waves in the cortex and suggested they play important roles in brain processing. These waves are most often measured using macroscopic methods that are unable to assess the local spiking activity underlying wave dynamics. Here, we investigated the possibility that waves may not be traveling at the single neuron scale. We first show that sequentially activating two discrete brain areas can appear as traveling waves in EEG simulations. We next reproduce these results using an analytical model of two sequentially activated regions. Using this model, we were able to generate wave-like activity with variable directions, velocities, and spatial patterns, and to map the discriminability limits between traveling waves and modular sequential activations. Finally, we investigated the link between field potentials and single neuron excitability using large-scale measurements from turtle cortex ex vivo. We found that while field potentials exhibit wave-like dynamics, the underlying spiking activity was better described by consecutively activated spatially adjacent groups of neurons. Taken together, this study suggests caution when interpreting phase delay measurements as continuously propagating wavefronts in two different spatial scales. A careful distinction between modular and wave excitability profiles across scales will be critical for understanding the nature of cortical computations.

Mapping the Retina onto the Brain

Vision begins in the retina, which extracts salient features from the environment and encodes them in the spike trains of retinal ganglion cells (RGCs), the output neurons of the eye. RGC axons innervate diverse brain areas (>50 in mice) to support perception, guide behavior, and mediate influences of light on physiology and internal states. In recent years, complete lists of RGC types (∼45 in mice) have been compiled, detailed maps of their dendritic connections drawn, and their light responses surveyed at scale. We know less about the RGCs' axonal projection patterns, which map retinal information onto the brain. However, some organizing principles have emerged. Here, we review the strategies and mechanisms that govern developing RGC axons and organize their innervation of retinorecipient brain areas.

Unsupervised learning on spontaneous retinal activity leads to efficient neural representation geometry

Prior to the onset of vision, neurons in the developing mammalian retina spontaneously fire in correlated activity patterns known as retinal waves. Experimental evidence suggests that retinal waves strongly influence the emergence of sensory representations before visual experience. We aim to model this early stage of functional development by using movies of neurally active developing retinas as pre-training data for neural networks. Specifically, we pre-train a ResNet-18 with an unsupervised contrastive learning objective (SimCLR) on both simulated and experimentally-obtained movies of retinal waves, then evaluate its performance on image classification tasks. We find that pre-training on retinal waves significantly improves performance on tasks that test object invariance to spatial translation, while slightly improving performance on more complex tasks like image classification. Notably, these performance boosts are realized on held-out natural images even though the pre-training procedure does not include any natural image data. We then propose a geometrical explanation for the increase in network performance, namely that the spatiotemporal characteristics of retinal waves facilitate the formation of separable feature representations. In particular, we demonstrate that networks pre-trained on retinal waves are more effective at separating image manifolds than randomly initialized networks, especially for manifolds defined by sets of spatial translations. These findings indicate that the broad spatiotemporal properties of retinal waves prepare networks for higher order feature extraction.

Acetylcholine Promotes Directionally Biased Glutamatergic Retinal Waves

Spontaneous retinal waves are a critical driving force for the self-organization of the mouse visual system prior to eye-opening. Classically characterized as taking place in three distinct stages defined by their primary excitatory drive, Stage II waves during the first postnatal week are propagated through the volume transmission of acetylcholine while Stage III retinal waves during the second postnatal week depend on glutamatergic transmission from bipolar cells. However, both late Stage II and early Stage III retinal waves share a defining propagation bias toward the temporal-to-nasal direction despite developmental changes in the underlying cholinergic and glutamatergic retinal networks. Here, we leverage genetic and pharmacological manipulations to investigate the relationship between cholinergic and glutamatergic neurotransmission during the transition between Stage II and Stage III waves in vivo. We find that the cholinergic network continues to play a vital role in the propagation of waves during Stage III after the primary mode of neurotransmission changes to glutamate. In the absence of glutamatergic waves, compensatory cholinergic activity persists but lacks the propagation bias typically observed in Stage III waves. In the absence of cholinergic waves, gap junction-mediated activity typically associated with Stage I waves persists throughout the developmental window in which Stage III waves usually emerge and lacks the spatiotemporal profile of normal Stage III waves, including a temporal-to-nasal propagation bias. Finally, we show that cholinergic signaling through β2 subunit-containing nicotinic acetylcholine receptors, essential for Stage II wave propagation, is also critical for Stage III wave directionality.

Ethanol Causes Cell Death and Neuronal Differentiation Defect During Initial Neurogenesis of the Neural Retina by Disrupting Calcium Signaling in Human Retinal Organoids

Fetal Alcohol Syndrome (FAS) affects a significant proportion, exceeding 90%, of afflicted children, leading to severe ocular aberrations such as microphthalmia and optic nerve hypoplasia. During the early stages of pregnancy, the commencement of neural retina neurogenesis represents a critical period for human eye development, concurrently exposing the developing retinal structures to the highest risk of prenatal ethanol exposure due to a lack of awareness. Despite the paramount importance of this period, the precise influence and underlying mechanisms of short-term ethanol exposure on the developmental process of the human neural retina have remained largely elusive. In this study, we utilize the human embryonic stem cells derived retinal organoids (hROs) to recapitulate the initial retinal neurogenesis and find that 1% (v/v) ethanol slows the growth of hROs by inducing robust cell death and retinal ganglion cell differentiation defect. Bulk RNA-seq analysis and two-photon microscope live calcium imaging reveal altered calcium signaling dynamics derived from ethanol-induced down-regulation of RYR1 and CACNA1S. Moreover, the calcium-binding protein RET, one of the downstream effector genes of the calcium signaling pathway, synergistically integrates ethanol and calcium signals to abort neuron differentiation and cause cell death. To sum up, our study illustrates the effect and molecular mechanism of ethanol on the initial neurogenesis of the human embryonic neural retina, providing a novel interpretation of the ocular phenotype of FAS and potentially informing preventative measures for susceptible populations.

Circuit mechanisms underlying embryonic retinal waves

Spontaneous activity is a hallmark of developing neural systems. In the retina, spontaneous activity comes in the form of retinal waves, comprised of three stages persisting from embryonic day 16 (E16) to eye opening at postnatal day 14 (P14). Though postnatal retinal waves have been well characterized, little is known about the spatiotemporal properties or the mechanisms mediating embryonic retinal waves, designated stage 1 waves. Using a custom-built macroscope to record spontaneous calcium transients from whole embryonic retinas, we show that stage 1 waves are initiated at several locations across the retina and propagate across a broad range of areas. Blocking gap junctions reduced the frequency and size of stage 1 waves, nearly abolishing them. Global blockade of nAChRs similarly nearly abolished stage 1 waves. Thus, stage 1 waves are mediated by a complex circuitry involving subtypes of nAChRs and gap junctions. Stage 1 waves in mice lacking the β2 subunit of the nAChRs (β2-nAChR-KO) persisted with altered propagation properties and were abolished by a gap junction blocker. To assay the impact of stage 1 waves on retinal development, we compared the spatial distribution of a subtype of retinal ganglion cells, intrinsically photosensitive retinal ganglion cells (ipRGCs), which undergo a significant amount of cell death, in WT and β2-nAChR-KO mice. We found that the developmental decrease in ipRGC density is preserved between WT and β2-nAChR-KO mice, indicating that processes regulating ipRGC numbers and distributions are not influenced by spontaneous activity.