Retinal connectomics: towards complete, accurate networks

Retinal Connectomics: A Review

The retina is an ideal model for understanding the fundamental rules for how neural networks are constructed. The compact neural networks of the retina perform all of the initial processing of visual information before transmission to higher visual centers in the brain. The field of retinal connectomics uses high-resolution electron microscopy datasets to map the intricate organization of these networks and further our understanding of how these computations are performed by revealing the fundamental topologies and allowable networks behind retinal computations. In this article, we review some of the notable advances that retinal connectomics has provided in our understanding of the specific cells and the organization of their connectivities within the retina, as well as how these are shaped in development and break down in disease. Using these anatomical maps to inform modeling has been, and will continue to be, instrumental in understanding how the retina processes visual signals.

Visual hallucinations originating in the retinofugal pathway under clinical and psychedelic conditions

Psychedelics like LSD (Lysergic acid diethylamide) and psilocybin are known to modulate perceptual modalities due to the activation of mostly serotonin receptors in specific cortical (e.g., visual cortex) and subcortical (e.g., thalamus) regions of the brain. In the visual domain, these psychedelic modulations often result in peculiar disturbances of viewed objects and light and sometimes even in hallucinations of non-existent environments, objects, and creatures. Although the underlying processes are poorly understood, research conducted over the past twenty years on the subjective experience of psychedelics details theories that attempt to explain these perceptual alterations due to a disruption of communication between cortical and subcortical regions. However, rare medical conditions in the visual system like Charles Bonnet syndrome that cause perceptual distortions may shed new light on the additional importance of the retinofugal pathway in psychedelic subjective experiences. Interneurons in the retina called amacrine cells could be the first site of visual psychedelic modulation and aid in disrupting the hierarchical structure of how humans perceive visual information. This paper presents an understanding of how the retinofugal pathway communicates and modulates visual information in psychedelic and clinical conditions. Therefore, we elucidate a new theory of psychedelic modulation in the retinofugal pathway.

Morphology and Fractal-Based Classifications of Neurons and Microglia in Two and Three Dimensions

Microglia and neurons live physically intertwined, intimately related structurally and functionally in a dynamic relationship in which microglia change continuously over a much shorter timescale than do neurons. Although microglia may unwind and depart from the neurons they attend under certain circumstances, in general, together both contribute to the fractal topology of the brain that defines its computational capabilities. Both neuronal and microglial morphologies are well-described using fractal analysis complementary to more traditional measures. For neurons, the fractal dimension has proved valuable for classifying dendritic branching and other neuronal features relevant to pathology and development. For microglia, fractal geometry has substantially contributed to classifying functional categories, where, in general, the more pathological the biological status, the lower the fractal dimension for individual cells, with some exceptions, including hyper-ramification. This chapter provides a review of the intimate relationships between neurons and microglia, by introducing 2D and 3D fractal analysis methodology and its applications in neuron-microglia function in health and disease.

Spatial Cluster Patterns of Retinal Sensitivity Loss in Intermediate Age-Related Macular Degeneration Features

Purpose

To examine spatial patterns of retinal sensitivity loss in the three key features of intermediate age-related macular degeneration (iAMD).

Methods

One-hundred individuals (53 iAMD, 47 normal) underwent 10-2 mesopic microperimetry testing in one eye. Pointwise sensitivities (dB) were corrected for age, sex, iAMD status, and co-presence of co-localized key iAMD features: drusen load, pigmentary abnormalities, and reticular pseudodrusen (RPD). Clusters (labeled by ranks of magnitude C-2, C-1, C0) were derived from pointwise sensitivities and then assessed by quadrants and eccentricity/rings.

Results

Two clusters of decreased sensitivities were evident in iAMD versus normal: C-2, -1.67 dB (95% CI (confidence intervals), -2.36 to -0.98; P < 0.0001); C-1, -0.93 dB (95% CI, -1.5 to -0.36; P < 0.01). One cluster of decreased sensitivity was independently associated each with increased drusen load (13.57 µm increase per -1 dB; P < 0.0001), pigmentary abnormalities (C-1: -2.23 dB; 95% CI, -3.36 to -1.1; P < 0.01), and RPD (C-1: -1.07 dB; 95% CI, -2 to -0.14; P < 0.01). Sensitivity loss in iAMD was biased toward the superior and central macula (P = 0.16 to <0.0001), aligning with structural distributions of features. However, sensitivity loss associated with drusen load also extended to the peripheral macula (P < 0.0001) with paracentral sparing, which was discordant with the central distribution of drusen.

Conclusions

Drusen load, pigmentary abnormalities, and RPD are associated with patterns of retinal sensitivity loss commonly demonstrating superior and central bias. Results highlighted that a clinical focus on these three key iAMD features using structural measures alone does not capture the complex, spatial extent of vision-related functional impairment in iAMD.

Translational Relevance

Defining the spatial patterns of retinal sensitivity loss in iAMD can facilitate a targeted visual field protocol for iAMD assessment.

Distinctive synaptic structural motifs link excitatory retinal interneurons to diverse postsynaptic partner types

Neurons make converging and diverging synaptic connections with distinct partner types. Whether synapses involving separate partners demonstrate similar or distinct structural motifs is not yet well understood. We thus used serial electron microscopy in mouse retina to map output synapses of cone bipolar cells (CBCs) and compare their structural arrangements across bipolar types and postsynaptic partners. Three presynaptic configurations emerge-single-ribbon, ribbonless, and multiribbon synapses. Each CBC type exploits these arrangements in a unique combination, a feature also found among rabbit ON CBCs. Though most synapses are dyads, monads and triads are also seen. Altogether, mouse CBCs exhibit at least six motifs, and each CBC type uses these in a stereotypic pattern. Moreover, synapses between CBCs and particular partner types appear biased toward certain motifs. Our observations reveal synaptic strategies that diversify the output within and across CBC types, potentially shaping the distinct functions of retinal microcircuits.

Retinal Pathoconnectomics: A Window into Neurodegeneration

Over the past decade, the field of retinal connectomics has made huge strides in describing the precise topologies underlying retinal visual processing. The same techniques that allowed these advancements are also applicable to understanding the progression of rewiring in retinal remodeling: retinal pathoconnectomics. Pathoconnectomics is unique in its unbiased approach to understanding the impacts of deafferentation on the remaining network components and identifying aberrant connectivities leading to visual processing defects. Pathoconnectomics also paves the way for identifying underlying rules of rewiring that may be recapitulated throughout the nervous system in other neurodegenerative diseases.

Neural System Identification With Spike-Triggered Non-Negative Matrix Factorization

Neuronal circuits formed in the brain are complex with intricate connection patterns. Such complexity is also observed in the retina with a relatively simple neuronal circuit. A retinal ganglion cell (GC) receives excitatory inputs from neurons in previous layers as driving forces to fire spikes. Analytical methods are required to decipher these components in a systematic manner. Recently a method called spike-triggered non-negative matrix factorization (STNMF) has been proposed for this purpose. In this study, we extend the scope of the STNMF method. By using retinal GCs as a model system, we show that STNMF can detect various computational properties of upstream bipolar cells (BCs), including spatial receptive field, temporal filter, and transfer nonlinearity. In addition, we recover synaptic connection strengths from the weight matrix of STNMF. Furthermore, we show that STNMF can separate spikes of a GC into a few subsets of spikes, where each subset is contributed by one presynaptic BC. Taken together, these results corroborate that STNMF is a useful method for deciphering the structure of neuronal circuits.

Using methanol to preserve retinas for immunostaining

BACKGROUND

Experimental studies on retinal vasculature and retinal ganglion cells (RGCs) investigating the developmental and pathological conditions of the retina mainly rely on whole-mount retinal immunostaining. Methanol, an auxiliary fixed medium for retinal whole-mount preparations, has been used in some studies; however, its application in short- and long-term storage of retinas for further study has not been well described. We aimed to evaluate methanol use as a preservation treatment for further immunostaining of the retina.

METHODS

We generated oxygen-induced retinopathy (OIR) and optic nerve crush (ONC) mouse models and used their retinas for analysis. We pipetted cold methanol (-20°C) on the surface of the retina to help fix the tissues while promoting permeability, after which the retinas were stored in cold methanol (-20°C) for 1, 6, or 12 months before being evaluated using various optical techniques. Thereafter, retinal whole-mount immunostaining was performed to analyse retinal neovascularisation and retinal hypoxia in OIR model, and retinal ganglion cell survival rate in ONC model.

RESULTS

Quantitative analysis revealed no significant differences in the fixed retinas after long-term storage in terms of retinal vasculature or retinal hypoxia in the OIR model. Similarly, no significant difference was found in RGC survival rate after long-term storage in methanol. These results suggest that methanol can be used as a storage medium when preserving retinal whole-mount samples.

CONCLUSIONS

Cold (-20°C) methanol can serve as an effective medium for long-term storage of fixed retinas, which is useful for further research.

Transience of the Retinal Output Is Determined by a Great Variety of Circuit Elements

Retinal ganglion cells (RGCs) encrypt stimulus features of the visual scene in action potentials and convey them toward higher visual centers in the brain. Although there are many visual features to encode, our recent understanding is that the ~46 different functional subtypes of RGCs in the retina share this task. In this scheme, each RGC subtype establishes a separate, parallel signaling route for a specific visual feature (e.g., contrast, the direction of motion, luminosity), through which information is conveyed. The efficiency of encoding depends on several factors, including signal strength, adaptational levels, and the actual efficacy of the underlying retinal microcircuits. Upon collecting inputs across their respective receptive field, RGCs perform further analysis (e.g., summation, subtraction, weighting) before they generate the final output spike train, which itself is characterized by multiple different features, such as the number of spikes, the inter-spike intervals, response delay, and the rundown time (transience) of the response. These specific kinetic features are essential for target postsynaptic neurons in the brain in order to effectively decode and interpret signals, thereby forming visual perception. We review recent knowledge regarding circuit elements of the mammalian retina that participate in shaping RGC response transience for optimal visual signaling.

Revealing Fine Structures of the Retinal Receptive Field by Deep-Learning Networks

Deep convolutional neural networks (CNNs) have demonstrated impressive performance on many visual tasks. Recently, they became useful models for the visual system in neuroscience. However, it is still not clear what is learned by CNNs in terms of neuronal circuits. When a deep CNN with many layers is used for the visual system, it is not easy to compare the structure components of CNNs with possible neuroscience underpinnings due to highly complex circuits from the retina to the higher visual cortex. Here, we address this issue by focusing on single retinal ganglion cells with biophysical models and recording data from animals. By training CNNs with white noise images to predict neuronal responses, we found that fine structures of the retinal receptive field can be revealed. Specifically, convolutional filters learned are resembling biological components of the retinal circuit. This suggests that a CNN learning from one single retinal cell reveals a minimal neural network carried out in this cell. Furthermore, when CNNs learned from different cells are transferred between cells, there is a diversity of transfer learning performance, which indicates that CNNs are cell specific. Moreover, when CNNs are transferred between different types of input images, here white noise versus natural images, transfer learning shows a good performance, which implies that CNNs indeed capture the full computational ability of a single retinal cell for different inputs. Taken together, these results suggest that CNNs could be used to reveal structure components of neuronal circuits, and provide a powerful model for neural system identification.

Dissecting cascade computational components in spiking neural networks

Finding out the physical structure of neuronal circuits that governs neuronal responses is an important goal for brain research. With fast advances for large-scale recording techniques, identification of a neuronal circuit with multiple neurons and stages or layers becomes possible and highly demanding. Although methods for mapping the connection structure of circuits have been greatly developed in recent years, they are mostly limited to simple scenarios of a few neurons in a pairwise fashion; and dissecting dynamical circuits, particularly mapping out a complete functional circuit that converges to a single neuron, is still a challenging question. Here, we show that a recent method, termed spike-triggered non-negative matrix factorization (STNMF), can address these issues. By simulating different scenarios of spiking neural networks with various connections between neurons and stages, we demonstrate that STNMF is a persuasive method to dissect functional connections within a circuit. Using spiking activities recorded at neurons of the output layer, STNMF can obtain a complete circuit consisting of all cascade computational components of presynaptic neurons, as well as their spiking activities. For simulated simple and complex cells of the primary visual cortex, STNMF allows us to dissect the pathway of visual computation. Taken together, these results suggest that STNMF could provide a useful approach for investigating neuronal systems leveraging recorded functional neuronal activity.

It is well known that the computation of neuronal circuits is carried out through the staged and cascade structure of different types of neurons. Nevertheless, the information, particularly sensory information, is processed in a network primarily with feedforward connections through different pathways. A peculiar example is the early visual system, where light is transcoded by the retinal cells, routed by the lateral geniculate nucleus, and reached the primary visual cortex. One meticulous interest in recent years is to map out these physical structures of neuronal pathways. However, most methods so far are limited to taking snapshots of a static view of connections between neurons. It remains unclear how to obtain a functional and dynamical neuronal circuit beyond the simple scenarios of a few randomly sampled neurons. Using simulated spiking neural networks of visual pathways with different scenarios of multiple stages, mixed cell types, and natural image stimuli, we demonstrate that a recent computational tool, named spike-triggered non-negative matrix factorization, can resolve these issues. It enables us to recover the entire structural components of neural networks underlying the computation, together with the functional components of each individual neuron. Utilizing it for complex cells of the primary visual cortex allows us to reveal every underpinning of the nonlinear computation. Our results, together with other recent experimental and computational efforts, show that it is possible to systematically dissect neural circuitry with detailed structural and functional components.

Generating and Using Transcriptomically Based Retinal Cell Atlases

It has been known for over a century that the basic organization of the retina is conserved across vertebrates. It has been equally clear that retinal cells can be classified into numerous types, but only recently have methods been devised to explore this diversity in unbiased, scalable, and comprehensive ways. Advances in high-throughput single-cell RNA-sequencing (scRNA-seq) have played a pivotal role in this effort. In this article, we outline the experimental and computational components of scRNA-seq and review studies that have used them to generate retinal atlases of cell types in several vertebrate species. These atlases have enabled studies of retinal development, responses of retinal cells to injury, expression patterns of genes implicated in retinal disease, and the evolution of cell types. Recently, the inquiry has expanded to include the entire eye and visual centers in the brain. These studies have enhanced our understanding of retinal function and dysfunction and provided tools and insights for exploring neural diversity throughout the brain. Expected final online publication date for the Annual Review of Vision Science, Volume 7 is September 2021. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.

Model-based comparison of current flow in rod bipolar cells of healthy and early-stage degenerated retina

Retinal degenerative diseases, such as retinitis pigmentosa, are generally thought to initiate with the loss of photoreceptors, though recent work suggests that plasticity and remodeling occurs prior to photoreceptor cell loss. This degeneration subsequently leads to death of other retinal neurons, creating functional alterations and extensive remodeling of retinal networks. Retinal prosthetic devices stimulate the surviving retinal cells by applying external current using implanted electrodes. Although these devices restore partial vision, the quality of restored vision is limited. Further knowledge about the precise changes in degenerated retina as the disease progresses is essential to understand how current flows in retinas undergoing degenerative disease and to improve the performance of retinal prostheses. We developed computational models that describe current flow from rod photoreceptors to rod bipolar cells (RodBCs) in the healthy and early-stage degenerated retina. Morphologically accurate models of retinal cells with their synapses are constructed based on retinal connectome datasets, created using serial section transmission electron microscopy (TEM) images of 70 nm-thick slices of either healthy (RC1) or early-stage degenerated (RPC1) rabbit retina. The passive membrane and active ion currents of each cell are implemented using conductance-based models in the Neuron simulation environment. In response to photocurrent input at rod photoreceptors, the simulated membrane potential at RodBCs in early degenerate tissue is approximately 10-20 mV lower than that of RodBCs of that observed in wild type retina. Results presented here suggest that although RodBCs in RPC1 show early, altered morphology compared to RC1, the lower membrane potential is primarily a consequence of reduced rod photoreceptor input to RodBCs in the degenerated retina. Frequency response and step input analyses suggest that individual cell responses of RodBCs in either healthy or early-degenerated retina, prior to substantial photoreceptor cell loss, do not differ significantly.

An Update on Connexin Gap Junction and Hemichannels in Diabetic Retinopathy

Diabetic retinopathy (DR) is one of the main causes of vision loss in the working age population. It is characterized by a progressive deterioration of the retinal microvasculature, caused by long-term metabolic alterations inherent to diabetes, leading to a progressive loss of retinal integrity and function. The mammalian retina presents an orderly layered structure that executes initial but complex visual processing and analysis. Gap junction channels (GJC) forming electrical synapses are present in each retinal layer and contribute to the communication between different cell types. In addition, connexin hemichannels (HCs) have emerged as relevant players that influence diverse physiological and pathological processes in the retina. This article highlights the impact of diabetic conditions on GJC and HCs physiology and their involvement in DR pathogenesis. Microvascular damage and concomitant loss of endothelial cells and pericytes are related to alterations in gap junction intercellular communication (GJIC) and decreased connexin 43 (Cx43) expression. On the other hand, it has been shown that the expression and activity of HCs are upregulated in DR, becoming a key element in the establishment of proinflammatory conditions that emerge during hyperglycemia. Hence, novel connexin HCs blockers or drugs to enhance GJIC are promising tools for the development of pharmacological interventions for diabetic retinopathy, and initial in vitro and in vivo studies have shown favorable results in this regard.

Dense Circuit Reconstruction to Understand Neuronal Computation: Focus on Zebrafish

The dense reconstruction of neuronal wiring diagrams from volumetric electron microscopy data has the potential to generate fundamentally new insights into mechanisms of information processing and storage in neuronal circuits. Zebrafish provide unique opportunities for dynamical connectomics approaches that combine reconstructions of wiring diagrams with measurements of neuronal population activity and behavior. Such approaches have the power to reveal higher-order structure in wiring diagrams that cannot be detected by sparse sampling of connectivity and that is essential for neuronal computations. In the brain stem, recurrently connected neuronal modules were identified that can account for slow, low-dimensional dynamics in an integrator circuit. In the spinal cord, connectivity specifies functional differences between premotor interneurons. In the olfactory bulb, tuning-dependent connectivity implements a whitening transformation that is based on the selective suppression of responses to overrepresented stimulus features. These findings illustrate the potential of dynamical connectomics in zebrafish to analyze the circuit mechanisms underlying higher-order neuronal computations.

Creation of virtual channels in the retina using synchronous and asynchronous stimulation - a modelling study

OBJECTIVE

Implantable retinal prostheses aim to provide artificial vision to those suffering from retinal degenerative diseases by electrically stimulating the remaining retinal neurons using a multi-electrode array. The spatial resolution of these devices can be improved by creation of so-called virtual channels (VCs) that are commonly achieved through synchronized stimulation of multiple electrodes. It is largely unclear though if VCs can be created using asynchronous stimulation, which was the primary aim of this study.

APPROACH

A computational model of multi-layered retina and epi-retinal dual-electrode stimulation was developed to simulate the neural activity of populations of retinal ganglion cells (RGCs) using the VC strategy under both synchronous and asynchronous stimulation conditions.

MAIN RESULTS

Our simulation suggests that VCs can be created using asynchronous stimulation. VC performance under both synchronous and asynchronous stimulation conditions can be improved by optimizing stimulation parameters such as current intensity, current ratio (α) between two electrodes, electrode spacing and the stimulation waveform. In particular, two VC performance measures; (1) linear displacement of the centroid of RGC activation, and (2) the RGC activation size consistency as a function of different current ratios α, have comparable performance under asynchronous and synchronous stimulation with appropriately selected stimulation parameters.

SIGNIFICANCE

Our findings support the possibility of creating VCs in the retina under both synchronous and asynchronous stimulation conditions. The results provide theoretical evidence for future retinal prosthesis designs with higher spatial resolution and power efficiency whilst reducing the number of current sources required to achieve these outcomes.

A pathoconnectome of early neurodegeneration: Network changes in retinal degeneration

Connectomics has demonstrated that synaptic networks and their topologies are precise and directly correlate with physiology and behavior. The next extension of connectomics is pathoconnectomics: to map neural network synaptology and circuit topologies corrupted by neurological disease in order to identify robust targets for therapeutics. In this report, we characterize a pathoconnectome of early retinal degeneration. This pathoconnectome was generated using serial section transmission electron microscopy to achieve an ultrastructural connectome with 2.18nm/px resolution for accurate identification of all chemical and gap junctional synapses. We observe aberrant connectivity in the rod-network pathway and novel synaptic connections deriving from neurite sprouting. These observations reveal principles of neuron responses to the loss of network components and can be extended to other neurodegenerative diseases.

Stimulus waveform design for decreasing charge and increasing stimulation selectivity in retinal prostheses

Retinal degenerative diseases, such as retinitis pigmentosa, begin with damage to the photoreceptor layer of the retina. In the absence of presynaptic input from photoreceptors, networks of electrically coupled AII amacrine and cone bipolar cells have been observed to exhibit oscillatory behaviour and result in spontaneous firing of ganglion cells. This ganglion cell activity could interfere with external stimuli provided by retinal prosthetic devices and potentially degrade their performance. In this work, the authors computationally investigate stimulus waveform designs, which can improve the performance of retinal prostheses by suppressing undesired spontaneous firing of ganglion cells and generating precise temporal spiking patterns. They utilise a multi-scale computational model for electrical stimulation of degenerated retina based on the admittance method and NEURON simulation environments. They present a class of asymmetric biphasic pulses that can generate precise ganglion cell firing patterns with up to 55% lower current requirements compared to traditional symmetric biphasic pulses. This lower current results in activation of only proximal ganglion cells, provides more focused stimulation and lowers the risk of tissue damage.

Retinal Drug Delivery: Rethinking Outcomes for the Efficient Replication of Retinal Behavior

The retina is a highly organized structure that is considered to be "an approachable part of the brain." It is attracting the interest of development scientists, as it provides a model neurovascular system. Over the last few years, we have been witnessing significant development in the knowledge of the mechanisms that induce the shape of the retinal vascular system, as well as knowledge of disease processes that lead to retina degeneration. Knowledge and understanding of how our vision works are crucial to creating a hardware-adaptive computational model that can replicate retinal behavior. The neuronal system is nonlinear and very intricate. It is thus instrumental to have a clear view of the neurophysiological and neuroanatomic processes and to take into account the underlying principles that govern the process of hardware transformation to produce an appropriate model that can be mapped to a physical device. The mechanistic and integrated computational models have enormous potential toward helping to understand disease mechanisms and to explain the associations identified in large model-free data sets. The approach used is modulated and based on different models of drug administration, including the geometry of the eye. This work aimed to review the recently used mathematical models to map a directed retinal network.

Network Architecture of Gap Junctional Coupling among Parallel Processing Channels in the Mammalian Retina

Gap junctions are ubiquitous throughout the nervous system, mediating critical signal transmission and integration, as well as emergent network properties. In mammalian retina, gap junctions within the Aii amacrine cell-ON cone bipolar cell (CBC) network are essential for night vision, modulation of day vision, and contribute to visual impairment in retinal degenerations, yet neither the extended network topology nor its conservation is well established. Here, we map the network contribution of gap junctions using a high-resolution connectomics dataset of an adult female rabbit retina. Gap junctions are prominent synaptic components of ON CBC classes, constituting 5%-25% of all axonal synaptic contacts. Many of these mediate canonical transfer of rod signals from Aii cells to ON CBCs for night vision, and we find that the uneven distribution of Aii signals to ON CBCs is conserved in rabbit, including one class entirely lacking direct Aii coupling. However, the majority of gap junctions formed by ON CBCs unexpectedly occur between ON CBCs, rather than with Aii cells. Such coupling is extensive, creating an interconnected network with numerous lateral paths both within, and particularly across, these parallel processing streams. Coupling patterns are precise with ON CBCs accepting and rejecting unique combinations of partnerships according to robust rulesets. Coupling specificity extends to both size and spatial topologies, thereby rivaling the synaptic specificity of chemical synapses. These ON CBC coupling motifs dramatically extend the coupled Aii-ON CBC network, with implications for signal flow in both scotopic and photopic retinal networks during visual processing and disease.SIGNIFICANCE STATEMENT Electrical synapses mediated by gap junctions are fundamental components of neural networks. In retina, coupling within the Aii-ON CBC network shapes visual processing in both the scotopic and photopic networks. In retinal degenerations, these same gap junctions mediate oscillatory activity that contributes to visual impairment. Here, we use high-resolution connectomics strategies to identify gap junctions and cellular partnerships. We describe novel, pervasive motifs both within and across classes of ON CBCs that dramatically extend the Aii-ON CBC network. These motifs are highly specific with implications for both signal processing within the retina and therapeutic interventions for blinding conditions. These findings highlight the underappreciated contribution of coupling motifs in retinal circuitry and the necessity of their detection in connectomics studies.

Application of machine learning in ophthalmic imaging modalities

In clinical ophthalmology, a variety of image-related diagnostic techniques have begun to offer unprecedented insights into eye diseases based on morphological datasets with millions of data points. Artificial intelligence (AI), inspired by the human multilayered neuronal system, has shown astonishing success within some visual and auditory recognition tasks. In these tasks, AI can analyze digital data in a comprehensive, rapid and non-invasive manner. Bioinformatics has become a focus particularly in the field of medical imaging, where it is driven by enhanced computing power and cloud storage, as well as utilization of novel algorithms and generation of data in massive quantities. Machine learning (ML) is an important branch in the field of AI. The overall potential of ML to automatically pinpoint, identify and grade pathological features in ocular diseases will empower ophthalmologists to provide high-quality diagnosis and facilitate personalized health care in the near future. This review offers perspectives on the origin, development, and applications of ML technology, particularly regarding its applications in ophthalmic imaging modalities.

Optic cup morphogenesis requires neural crest-mediated basement membrane assembly

Organogenesis requires precise interactions between a developing tissue and its environment. In vertebrates, the developing eye is surrounded by a complex extracellular matrix as well as multiple mesenchymal cell populations. Disruptions to either the matrix or periocular mesenchyme can cause defects in early eye development, yet in many cases the underlying mechanism is unknown. Here, using multidimensional imaging and computational analyses in zebrafish, we establish that cell movements in the developing optic cup require neural crest. Ultrastructural analysis reveals that basement membrane formation around the developing eye is also dependent on neural crest, but only specifically around the retinal pigment epithelium. Neural crest cells produce the extracellular matrix protein nidogen: impairing nidogen function disrupts eye development, and, strikingly, expression of nidogen in the absence of neural crest partially restores optic cup morphogenesis. These results demonstrate that eye formation is regulated in part by extrinsic control of extracellular matrix assembly.This article has an associated 'The people behind the papers' interview.

Synaptic inputs from identified bipolar and amacrine cells to a sparsely branched ganglion cell in rabbit retina

There are more than 30 distinct types of mammalian retinal ganglion cells, each sensitive to different features of the visual environment. In rabbit retina, they can be grouped into four classes according to their morphology and stratification of their dendrites in the inner plexiform layer (IPL). The goal of this study was to describe the synaptic inputs to one type of Class IV ganglion cell, the third member of the sparsely branched Class IV cells (SB3). One cell of this type was partially reconstructed in a retinal connectome developed using automated transmission electron microscopy (ATEM). It had slender, relatively straight dendrites that ramify in the sublamina a of the IPL. The dendrites of the SB3 cell were always postsynaptic in the IPL, supporting its identity as a ganglion cell. It received 29% of its input from bipolar cells, a value in the middle of the range for rabbit retinal ganglion cells studied previously. The SB3 cell typically received only one synapse per bipolar cell from multiple types of presumed OFF bipolar cells; reciprocal synapses from amacrine cells at the dyad synapses were infrequent. In a few instances, the bipolar cells presynaptic to the SB3 ganglion cell also provided input to an amacrine cell presynaptic to the ganglion cell. There was apparently no crossover inhibition from narrow-field ON amacrine cells. Most of the amacrine cell inputs were from axons and dendrites of GABAergic amacrine cells, likely providing inhibitory input from outside the classical receptive field.

Persistent remodeling and neurodegeneration in late-stage retinal degeneration

Retinal remodeling is a progressive series of negative plasticity revisions that arise from retinal degeneration, and are seen in retinitis pigmentosa, age-related macular degeneration and other forms of retinal disease. These processes occur regardless of the precipitating event leading to degeneration. Retinal remodeling then culminates in a late-stage neurodegeneration that is indistinguishable from progressive central nervous system (CNS) proteinopathies. Following long-term deafferentation from photoreceptor cell death in humans, and long-lived animal models of retinal degeneration, most retinal neurons reprogram, then die. Glial cells reprogram into multiple anomalous metabolic phenotypes. At the same time, survivor neurons display degenerative inclusions that appear identical to progressive CNS neurodegenerative disease, and contain aberrant α-synuclein (α-syn) and phosphorylated α-syn. In addition, ultrastructural analysis indicates a novel potential mechanism for misfolded protein transfer that may explain how proteinopathies spread. While neurodegeneration poses a barrier to prospective retinal interventions that target primary photoreceptor loss, understanding the progression and time-course of retinal remodeling will be essential for the establishment of windows of therapeutic intervention and appropriate tuning and design of interventions. Finally, the development of protein aggregates and widespread neurodegeneration in numerous retinal degenerative diseases positions the retina as a ideal platform for the study of proteinopathies, and mechanisms of neurodegeneration that drive devastating CNS diseases.

Dynamic assembly of ribbon synapses and circuit maintenance in a vertebrate sensory system

Ribbon synapses transmit information in sensory systems, but their development is not well understood. To test the hypothesis that ribbon assembly stabilizes nascent synapses, we performed simultaneous time-lapse imaging of fluorescently-tagged ribbons in retinal cone bipolar cells (BCs) and postsynaptic densities (PSD95-FP) of retinal ganglion cells (RGCs). Ribbons and PSD95-FP clusters were more stable when these components colocalized at synapses. However, synapse density on ON-alpha RGCs was unchanged in mice lacking ribbons (ribeye knockout). Wildtype BCs make both ribbon-containing and ribbon-free synapses with these GCs even at maturity. Ribbon assembly and cone BC-RGC synapse maintenance are thus regulated independently. Despite the absence of synaptic ribbons, RGCs continued to respond robustly to light stimuli, although quantitative examination of the responses revealed reduced frequency and contrast sensitivity.

A Brief History of Simulation Neuroscience

Our knowledge of the brain has evolved over millennia in philosophical, experimental and theoretical phases. We suggest that the next phase is simulation neuroscience. The main drivers of simulation neuroscience are big data generated at multiple levels of brain organization and the need to integrate these data to trace the causal chain of interactions within and across all these levels. Simulation neuroscience is currently the only methodology for systematically approaching the multiscale brain. In this review, we attempt to reconstruct the deep historical paths leading to simulation neuroscience, from the first observations of the nerve cell to modern efforts to digitally reconstruct and simulate the brain. Neuroscience began with the identification of the neuron as the fundamental unit of brain structure and function and has evolved towards understanding the role of each cell type in the brain, how brain cells are connected to each other, and how the seemingly infinite networks they form give rise to the vast diversity of brain functions. Neuronal mapping is evolving from subjective descriptions of cell types towards objective classes, subclasses and types. Connectivity mapping is evolving from loose topographic maps between brain regions towards dense anatomical and physiological maps of connections between individual genetically distinct neurons. Functional mapping is evolving from psychological and behavioral stereotypes towards a map of behaviors emerging from structural and functional connectomes. We show how industrialization of neuroscience and the resulting large disconnected datasets are generating demand for integrative neuroscience, how the scale of neuronal and connectivity maps is driving digital atlasing and digital reconstruction to piece together the multiple levels of brain organization, and how the complexity of the interactions between molecules, neurons, microcircuits and brain regions is driving brain simulation to understand the interactions in the multiscale brain.

Functional Diversity in the Retina Improves the Population Code

Within a given brain region, individual neurons exhibit a wide variety of different feature selectivities. Here, we investigated the impact of this extensive functional diversity on the population neural code. Our approach was to build optimal decoders to discriminate among stimuli using the spiking output of a real, measured neural population and compare its performance against a matched, homogeneous neural population with the same number of cells and spikes. Analyzing large populations of retinal ganglion cells, we found that the real, heterogeneous population can yield a discrimination error lower than the homogeneous population by several orders of magnitude and consequently can encode much more visual information. This effect increases with population size and with graded degrees of heterogeneity. We complemented these results with an analysis of coding based on the Chernoff distance, as well as derivations of inequalities on coding in certain limits, from which we can conclude that the beneficial effect of heterogeneity occurs over a broad set of conditions. Together, our results indicate that the presence of functional diversity in neural populations can enhance their coding fidelity appreciably. A noteworthy outcome of our study is that this effect can be extremely strong and should be taken into account when investigating design principles for neural circuits.

Heterocellular Coupling Between Amacrine Cells and Ganglion Cells

All superclasses of retinal neurons, including bipolar cells (BCs), amacrine cells (ACs) and ganglion cells (GCs), display gap junctional coupling. However, coupling varies extensively by class. Heterocellular AC coupling is common in many mammalian GC classes. Yet, the topology and functions of coupling networks remains largely undefined. GCs are the least frequent superclass in the inner plexiform layer and the gap junctions mediating GC-to-AC coupling (GC::AC) are sparsely arrayed amidst large cohorts of homocellular AC::AC, BC::BC, GC::GC and heterocellular AC::BC gap junctions. Here, we report quantitative coupling for identified GCs in retinal connectome 1 (RC1), a high resolution (2 nm) transmission electron microscopy-based volume of rabbit retina. These reveal that most GC gap junctions in RC1 are suboptical. GC classes lack direct cross-class homocellular coupling with other GCs, despite opportunities via direct membrane contact, while OFF alpha GCs and transient ON directionally selective (DS) GCs are strongly coupled to distinct AC cohorts. Integrated small molecule immunocytochemistry identifies these as GABAergic ACs (γ+ ACs). Multi-hop synaptic queries of RC1 connectome further profile these coupled γ+ ACs. Notably, OFF alpha GCs couple to OFF γ+ ACs and transient ON DS GCs couple to ON γ+ ACs, including a large interstitial amacrine cell, revealing matched ON/OFF photic drive polarities within coupled networks. Furthermore, BC input to these γ+ ACs is tightly matched to the GCs with which they couple. Evaluation of the coupled versus inhibitory targets of the γ+ ACs reveals that in both ON and OFF coupled GC networks these ACs are presynaptic to GC classes that are different than the classes with which they couple. These heterocellular coupling patterns provide a potential mechanism for an excited GC to indirectly inhibit nearby GCs of different classes. Similarly, coupled γ+ ACs engaged in feedback networks can leverage the additional gain of BC synapses in shaping the signaling of downstream targets based on their own selective coupling with GCs. A consequence of coupling is intercellular fluxes of small molecules. GC::AC coupling involves primarily γ+ cells, likely resulting in GABA diffusion into GCs. Surveying GABA signatures in the GC layer across diverse species suggests the majority of vertebrate retinas engage in GC::γ+ AC coupling.

Increasing Electrical Stimulation Efficacy in Degenerated Retina: Stimulus Waveform Design in a Multiscale Computational Model

A computational model of electrical stimulation of the retina is proposed for investigating current waveforms used in prosthetic devices for restoring partial vision lost to retinal degenerative diseases. The model framework combines a connectome-based neural network model characterized by accurate morphological and synaptic properties with an admittance method model of bulk tissue and prosthetic electronics. In this model, the retina was computationally "degenerated," considering cellular death and anatomical changes that occur early in disease, as well as altered neural behavior that develops throughout the neurodegeneration and is likely interfering with current attempts at restoring vision. A resulting analysis of stimulation range and threshold of ON ganglion cells within the retina that are either healthy or in beginning stages of degeneration is presented for currently used stimulation waveforms, and an asymmetric biphasic current stimulation for subduing spontaneous firing to allow increased control over ganglion cell firing patterns in degenerated retina is proposed. Results show that stimulation thresholds of retinal ganglion cells do not notably vary after beginning stages of retina degeneration. In addition, simulation of proposed asymmetric waveforms showed the ability to enhance the control of ganglion cell firing via electrical stimulation.

Reexamination of Dopaminergic Amacrine Cells in the Rabbit Retina: Confocal Analysis with Double- and Triple-labeling Immunohistochemistry

Dopaminergic amacrine cells (DACs) are among the most well-characterized neurons in the mammalian retina, and their connections to AII amacrine cells have been described in detail. However, the stratification of DAC dendrites differs based on their location in the inner plexiform layer (IPL), raising the question of whether all AII lobules are modulated by dopamine release from DACs. The present study aimed to clarify the relationship between DACs and AII amacrine cells, and to further elucidate the role of dopamine at synapses with AII amacrine cell. In the rabbit retina, DAC dendrites were observed in strata 1, 3, and 5 of the IPL. In stratum 1, most DAC dendritic varicosities—the presumed sites of neurotransmitter release—made contact with the somata and lobular appendages of AII amacrine cells. However, most lobular appendages of AII amacrine cells localized within stratum 2 of the IPL exhibited little contact with DAC varicosities. In addition, double- or triple-labeling experiments revealed that DACs did not express the GABAergic neuronal markers anti-GABA, vesicular GABA transporter, or glutamic acid decarboxylase. These findings suggest that the lobular appendages of AII amacrine cells are involved in at least two different circuits. We speculate that the circuit associated with stratum 1 of the IPL is modulated by DACs, while that associated with stratum 2 is modulated by unknown amacrine cells expressing a different neuroactive substance. Our findings further indicate that DACs in the rabbit retina do not use GABA as a neurotransmitter, in contrast to those in other mammals.

Connectome studies on Drosophila: a short perspective on a tiny brain

The brain is a network of neurons, one that generates behaviour, and knowing the former is crucial to understanding the latter. Identifying the exact network of synaptic connections, or connectome, of the fly's central nervous system is now a major objective in Drosophila neurobiology, one that has been initiated in several laboratories, especially the Janelia Research Campus of the Howard Hughes Medical Institute. Progress is most advanced in the optic neuropiles of the visual system. The effort to derive a connectome from these and other neuropile regions is proceeding by various methods of electron microscopy, especially focused-ion beam milling scanning electron microscopy, and relies upon - but is to be carefully distinguished from - published light microscopic methods that reveal the projections of genetically labelled cell types. The latter reveal those neurons that come into close proximity and are therefore candidate synaptic partners. Synaptic partnerships are not in fact reliably revealed by such candidate pairs, anatomical connections often revealing unexpected pathways. Synaptic partnerships identified from ultrastructural features provide a strong heuristic basis to interpret not only functional interactions between identified neurons, but also a powerful means to predict such interactions, and suggest functional pathways not readily predicted from existing experimental evidence. The analysis of circuit function may proceed cell by cell, by examining the behavioural outcome of either interrupting or restoring function to any one element in an anatomically defined circuit, but can be foiled by degeneracy in pathway elements. Circuit information can also be used to identify and analyse circuit motifs, and their role in higher-order network properties. These attempts in Drosophila anticipate parallel attempts in other systems, notably the inner plexiform layer of the vertebrate retina, and augment the one complete connectome already available to us, that available for 30 years in the nematode Caenorhabditis elegans.

Bipolar cell gap junctions serve major signaling pathways in the human retina

Connexin36 (Cx36) constituent gap junctions (GJ) throughout the brain connect neurons into functional syncytia. In the retina they underlie the transmission, averaging and correlation of signals prior conveying visual information to the brain. This is the first study that describes retinal bipolar cell (BC) GJs in the human inner retina, whose function is enigmatic even in the examined animal models. Furthermore, a number of unique features (e.g. fovea, trichromacy, midget system) necessitate a reexamination of the animal model results in the human retina. Well-preserved postmortem human samples of this study are allowed to identify Cx36 expressing BCs neurochemically. Results reveal that both rod and cone pathway interneurons display strong Cx36 expression. Rod BC inputs to AII amacrine cells (AC) appear in juxtaposition to AII GJs, thus suggesting a strategic AII cell targeting by rod BCs. Cone BCs serving midget, parasol or koniocellular signaling pathways display a wealth of Cx36 expression to form homologously coupled arrays. In addition, they also establish heterologous GJ contacts to serve an exchange of information between parallel signaling streams. Interestingly, a prominent Cx36 expression was exhibited by midget system BCs that appear to maintain intimate contacts with bistratified BCs serving other pathways. These findings suggest that BC GJs in parallel signaling streams serve both an intra- and inter-pathway exchange of signals in the human retina.

Localization and diagnosis framework for pediatric cataracts based on slit-lamp images using deep features of a convolutional neural network

Slit-lamp images play an essential role for diagnosis of pediatric cataracts. We present a computer vision-based framework for the automatic localization and diagnosis of slit-lamp images by identifying the lens region of interest (ROI) and employing a deep learning convolutional neural network (CNN). First, three grading degrees for slit-lamp images are proposed in conjunction with three leading ophthalmologists. The lens ROI is located in an automated manner in the original image using two successive applications of Candy detection and the Hough transform, which are cropped, resized to a fixed size and used to form pediatric cataract datasets. These datasets are fed into the CNN to extract high-level features and implement automatic classification and grading. To demonstrate the performance and effectiveness of the deep features extracted in the CNN, we investigate the features combined with support vector machine (SVM) and softmax classifier and compare these with the traditional representative methods. The qualitative and quantitative experimental results demonstrate that our proposed method offers exceptional mean accuracy, sensitivity and specificity: classification (97.07%, 97.28%, and 96.83%) and a three-degree grading area (89.02%, 86.63%, and 90.75%), density (92.68%, 91.05%, and 93.94%) and location (89.28%, 82.70%, and 93.08%). Finally, we developed and deployed a potential automatic diagnostic software for ophthalmologists and patients in clinical applications to implement the validated model.

Rod-cone crossover connectome of mammalian bipolar cells

The basis of cross-suppression between rod and cone channels has long been an enigma. Using rabbit retinal connectome RC1, we show that all cone bipolar cell (BC) classes inhibit rod BCs via amacrine cell (AC) motifs (C1-6); that all cone BC classes are themselves inhibited by AC motifs (R1-5, R25) driven by rod BCs. A sparse symmetric AC motif (CR) is presynaptic and postsynaptic to both rod and cone BCs. ON cone BCs of all classes drive inhibition of rod BCs via motif C1 wide-field GABAergic ACs (γACs) and motif C2 narrow field glycinergic ON ACs (GACs). Each rod BC receives ≈10 crossover AC synapses and each ON cone BC can target ≈10 or more rod BCs via separate AC processes. OFF cone BCs mediate monosynaptic inhibition of rod BCs via motif C3 driven by OFF γACs and GACs and disynaptic inhibition via motifs C4 and C5 driven by OFF wide-field γACs and narrow-field GACs, respectively. Motifs C4 and C5 form halos of 60-100 inhibitory synapses on proximal dendrites of AI γACs. Rod BCs inhibit surrounding arrays of cone BCs through AII GAC networks that access ON and OFF cone BC patches via motifs R1, R2, R4, R5 and a unique ON AC motif R3 that collects rod BC inputs and targets ON cone BCs. Crossover synapses for motifs C1, C4, C5, and R3 are 3-4× larger than typical feedback synapses, which may be a signature for synaptic winner-take-all switches. J. Comp. Neurol. 527:87-116, 2019. © 2016 The Authors The Journal of Comparative Neurology Published by Wiley Periodicals, Inc.

Towards high-resolution retinal prostheses with direct optical addressing and inductive telemetry

OBJECTIVE

Despite considerable advances in retinal prostheses over the last two decades, the resolution of restored vision has remained severely limited, well below the 20/200 acuity threshold of blindness. Towards drastic improvements in spatial resolution, we present a scalable architecture for retinal prostheses in which each stimulation electrode is directly activated by incident light and powered by a common voltage pulse transferred over a single wireless inductive link.

APPROACH

The hybrid optical addressability and electronic powering scheme provides separate spatial and temporal control over stimulation, and further provides optoelectronic gain for substantially lower light intensity thresholds than other optically addressed retinal prostheses using passive microphotodiode arrays. The architecture permits the use of high-density electrode arrays with ultra-high photosensitive silicon nanowires, obviating the need for excessive wiring and high-throughput data telemetry. Instead, the single inductive link drives the entire array of electrodes through two wires and provides external control over waveform parameters for common voltage stimulation.

MAIN RESULTS

A complete system comprising inductive telemetry link, stimulation pulse demodulator, charge-balancing series capacitor, and nanowire-based electrode device is integrated and validated ex vivo on rat retina tissue.

SIGNIFICANCE

Measurements demonstrate control over retinal neural activity both by light and electrical bias, validating the feasibility of the proposed architecture and its system components as an important first step towards a high-resolution optically addressed retinal prosthesis.

Store-Operated Calcium Entry in Müller Glia Is Controlled by Synergistic Activation of TRPC and Orai Channels

The endoplasmic reticulum (ER) is at the epicenter of astrocyte Ca(2+) signaling. We sought to identify the molecular mechanism underlying store-operated calcium entry that replenishes ER stores in mouse Müller cells. Store depletion, induced through blockade of sequestration transporters in Ca(2+)-free saline, induced synergistic activation of canonical transient receptor potential 1 (TRPC1) and Orai channels. Store-operated TRPC1 channels were identified by their electrophysiological properties, pharmacological blockers, and ablation of the Trpc1 gene. Ca(2+) release-activated currents (ICRAC) were identified by ion permeability, voltage dependence, and sensitivity to selective Orai antagonists Synta66 and GSK7975A. Depletion-evoked calcium influx was initiated at the Müller end-foot and apical process, triggering centrifugal propagation of Ca(2+) waves into the cell body. EM analysis of the end-foot compartment showed high-density ER cisternae that shadow retinal ganglion cell (RGC) somata and axons, protoplasmic astrocytes, vascular endothelial cells, and ER-mitochondrial contacts at the vitreal surface of the end-foot. The mouse retina expresses transcripts encoding both Stim and all known Orai genes; Müller glia predominantly express stromal interacting molecule 1 (STIM1), whereas STIM2 is mainly confined to the outer plexiform and RGC layers. Elimination of TRPC1 facilitated Müller gliosis induced by the elevation of intraocular pressure, suggesting that TRPC channels might play a neuroprotective role during mechanical stress. By characterizing the properties of store-operated signaling pathways in Müller cells, these studies expand the current knowledge about the functional roles these cells play in retinal physiology and pathology while also providing further evidence for the complexity of calcium signaling mechanisms in CNS astroglia.

SIGNIFICANCE STATEMENT

Store-operated Ca(2+) signaling represents a major signaling pathway and source of cytosolic Ca(2+) in astrocytes. Here, we show that the store-operated response in Müller cells, radial glia that perform key structural, signaling, osmoregulatory, and mechanosensory functions within the retina, is mediated through synergistic activation of transient receptor potential and Orai channels. The end-foot disproportionately expresses the depletion sensor stromal interacting molecule 1, which contains an extraordinarily high density of endoplasmic reticulum cisternae that shadow neuronal, astrocytic, vascular, and axonal structures; interface with mitochondria; but also originate store-operated Ca(2+) entry-induced transcellular Ca(2+) waves that propagate glial excitation into the proximal retina. These results identify a molecular mechanism that underlies complex interactions between the plasma membrane and calcium stores, and contributes to astroglial function, regulation, and response to mechanical stress.

Network Analysis and Visualization of Mouse Retina Connectivity Data

The largest available cellular level connectivity map, of a 0.1 mm sample of the mouse retina Inner Plexiform Layer, was analysed using network models and visualized using spectral graph layouts and observed cell coordinates. This allows key nodes in the network to be identified with retinal neurons. Their strongest synaptic links can trace pathways in the network, elucidating possible circuits. Modular decomposition of the network, by sampling signal flows over nodes and links using the InfoMap method, shows discrete modules of cone bipolar cells that form a tiled mosaic in the retinal plane. The highest flow nodes, calculated by InfoMap, proved to be the most useful landmarks for elucidating possible circuits. Their dominant links to high flow amacrine cells reveal possible circuits linking bipolar through to ganglion cells and show an Off-On discrimination between the Left-Right sections of the sample. Circuits suggested by this analysis confirm known roles for some cells and point to roles for others.

Progress Towards Mammalian Whole-Brain Cellular Connectomics

Neurons are the fundamental structural units of the nervous system-i.e., the Neuron Doctrine-as the pioneering work of Santiago Ramón y Cajal in the 1880's clearly demonstrated through careful observation of Golgi-stained neuronal morphologies. However, at that time sample preparation, imaging methods and computational tools were either nonexistent or insufficiently developed to permit the precise mapping of an entire brain with all of its neurons and their connections. Some measure of the "mesoscopic" connectional organization of the mammalian brain has been obtained over the past decade by alignment of sparse subsets of labeled neurons onto a reference atlas or via MRI-based diffusion tensor imaging. Neither method, however, provides data on the complete connectivity of all neurons comprising an individual brain. Fortunately, whole-brain cellular connectomics now appears within reach due to recent advances in whole-brain sample preparation and high-throughput electron microscopy (EM), though substantial obstacles remain with respect to large volume electron microscopic acquisitions and automated neurite reconstructions. This perspective examines the current status and problems associated with generating a mammalian whole-brain cellular connectome and argues that the time is right to launch a concerted connectomic attack on a small mammalian whole-brain.

On the computation of a retina resistivity profile for applications in multi-scale modeling of electrical stimulation and absorption

This study proposes a methodology for computationally estimating resistive properties of tissue in multi-scale computational models, used for studying the interaction of electromagnetic fields with neural tissue, with applications to both dosimetry and neuroprosthetics. Traditionally, models at bulk tissue- and cellular-level scales are solved independently, linking resulting voltage from existing resistive tissue-scale models as extracellular sources to cellular models. This allows for solving the effects that external electric fields have on cellular activity. There are two major limitations to this approach: first, the resistive properties of the tissue need to be chosen, of which there are contradicting measurements in literature; second, the measurements of resistivity themselves may be inaccurate, leading to the mentioned contradicting results found across different studies. Our proposed methodology allows for constructing computed resistivity profiles using knowledge of only the neural morphology within the multi-scale model, resulting in a practical implementation of the effective medium theory; this bypasses concerns regarding the choice of resistive properties and accuracy of measurement setups. A multi-scale model of retina is constructed with an external electrode to serve as a test bench for analyzing existing and resulting resistivity profiles, and validation is presented through the reconstruction of a published resistivity profile of retina tissue. Results include a computed resistivity profile of retina tissue for use with a retina multi-scale model used to analyze effects of external electric fields on neural activity.

Functional Circuitry of the Retina

The mammalian retina is an important model system for studying neural circuitry: Its role in sensation is clear, its cell types are relatively well defined, and its responses to natural stimuli-light patterns-can be studied in vitro. To solve the retina, we need to understand how the circuits presynaptic to its output neurons, ganglion cells, divide the visual scene into parallel representations to be assembled and interpreted by the brain. This requires identifying the component interneurons and understanding how their intrinsic properties and synapses generate circuit behaviors. Because the cellular composition and fundamental properties of the retina are shared across species, basic mechanisms studied in the genetically modifiable mouse retina apply to primate vision. We propose that the apparent complexity of retinal computation derives from a straightforward mechanism-a dynamic balance of synaptic excitation and inhibition regulated by use-dependent synaptic depression-applied differentially to the parallel pathways that feed ganglion cells.

A multi-scale computational model for the study of retinal prosthetic stimulation

An implantable retinal prosthesis has been developed to restore vision to patients who have been blinded by degenerative diseases that destroy photoreceptors. By electrically stimulating the surviving retinal cells, the damaged photoreceptors may be bypassed and limited vision can be restored. While this has been shown to restore partial vision, the understanding of how cells react to this systematic electrical stimulation is largely unknown. Better predictive models and a deeper understanding of neural responses to electrical stimulation is necessary for designing a successful prosthesis. In this work, a computational model of an epi-retinal implant was built and simulated, spanning multiple spatial scales, including a large-scale model of the retina and implant electronics, as well as underlying neuronal networks.

Fixation strategies for retinal immunohistochemistry

Immunohistochemical and ex vivo anatomical studies have provided many glimpses of the variety, distribution, and signaling components of vertebrate retinal neurons. The beauty of numerous images published to date, and the qualitative and quantitative information they provide, indicate that these approaches are fundamentally useful. However, obtaining these images entailed tissue handling and exposure to chemical solutions that differ from normal extracellular fluid in composition, temperature, and osmolarity. Because the differences are large enough to alter intercellular and intracellular signaling in neurons, and because retinae are susceptible to crush, shear, and fray, it is natural to wonder if immunohistochemical and anatomical methods disturb or damage the cells they are designed to examine. Tissue fixation is typically incorporated to guard against this damage and is therefore critically important to the quality and significance of the harvested data. Here, we describe mechanisms of fixation; advantages and disadvantages of using formaldehyde and glutaraldehyde as fixatives during immunohistochemistry; and modifications of widely used protocols that have recently been found to improve cell shape preservation and immunostaining patterns, especially in proximal retinal neurons.

Knowing a synapse when you see one

Recent years have seen a rapidly growing recognition of the complexity and diversity of the myriad individual synaptic connections that define brain synaptic networks. It has also become increasingly apparent that the synapses themselves are a major key to understanding the development, function and adaptability of those synaptic networks. In spite of this growing appreciation, the molecular, structural and functional characteristics of individual synapses and the patterning of their diverse characteristics across functional networks have largely eluded quantitative study with available imaging technologies. Here we offer an overview of new computational imaging methods that promise to bring single-synapse analysis of synaptic networks to the fore. We focus especially on the challenges and opportunities associated with quantitative detection of individual synapses and with measuring individual synapses across network scale populations in mammalian brain.

The types of retinal ganglion cells: current status and implications for neuronal classification

In the retina, photoreceptors pass visual information to interneurons, which process it and pass it to retinal ganglion cells (RGCs). Axons of RGCs then travel through the optic nerve, telling the rest of the brain all it will ever know about the visual world. Research over the past several decades has made clear that most RGCs are not merely light detectors, but rather feature detectors, which send a diverse set of parallel, highly processed images of the world on to higher centers. Here, we review progress in classification of RGCs by physiological, morphological, and molecular criteria, making a particular effort to distinguish those cell types that are definitive from those for which information is partial. We focus on the mouse, in which molecular and genetic methods are most advanced. We argue that there are around 30 RGC types and that we can now account for well over half of all RGCs. We also use RGCs to examine the general problem of neuronal classification, arguing that insights and methods from the retina can guide the classification enterprise in other brain regions.

Large-scale automatic reconstruction of neuronal processes from electron microscopy images

Automated sample preparation and electron microscopy enables acquisition of very large image data sets. These technical advances are of special importance to the field of neuroanatomy, as 3D reconstructions of neuronal processes at the nm scale can provide new insight into the fine grained structure of the brain. Segmentation of large-scale electron microscopy data is the main bottleneck in the analysis of these data sets. In this paper we present a pipeline that provides state-of-the art reconstruction performance while scaling to data sets in the GB-TB range. First, we train a random forest classifier on interactive sparse user annotations. The classifier output is combined with an anisotropic smoothing prior in a Conditional Random Field framework to generate multiple segmentation hypotheses per image. These segmentations are then combined into geometrically consistent 3D objects by segmentation fusion. We provide qualitative and quantitative evaluation of the automatic segmentation and demonstrate large-scale 3D reconstructions of neuronal processes from a 27,000 μm(3) volume of brain tissue over a cube of 30 μm in each dimension corresponding to 1000 consecutive image sections. We also introduce Mojo, a proofreading tool including semi-automated correction of merge errors based on sparse user scribbles.

Developmental changes in the expression level of connexin36 in the rat retina

Connexin36 (Cx36) is the major gap junction forming protein in the brain and the retina; thus, alterations in its expression indicate changes in the corresponding circuitry. Many structural changes occur in the early postnatal retina before functional neuronal circuits are finalized, including those that incorporate gap junctions. To reveal the time-lapse formation of inner retinal gap junctions, we examine the developing postnatal rat retina from birth (P0) to young adult age (P20) and follow the expression of Cx36 in the mRNA and protein levels. We found a continuous elevation in the expression of both the Cx36 transcript and protein between P0 and P20 and a somewhat delayed Cx36 plaque formation throughout the inner plexiform layer (IPL) starting at P10. By using tristratificated calretinin positive (CaR(+)) fibers in the IPL as a guide, we detected a clear preference of Cx36 plaques for the ON sublamina from the earliest time of detection. This distributional preference became more pronounced at P15 and P20 due to the emergence and widespread expression of large (>0.1 μm(2)) Cx36 plaques in the ON sublamina. Finally, we showed that parvalbumin-positive (PV(+)) AII amacrine cell dendrites colocalize with Cx36 plaques as early as P10 in strata 3 and 4, whereas colocalizations in stratum 5 became characteristic only around P20. We conclude that Cx36 expression in the rat IPL displays a characteristic succession of changes during retinogenesis reflecting the formation of the underlying electrical synaptic circuitry. In particular, AII cell gap junctions, first formed with ON cone bipolar cells and later with other AII amacrine cells, accounted for the observed Cx36 expressional changes.