Retinal synaptic pathways underlying the response of the rabbit local edge detector

Defining morphologically and genetically distinct GABAergic/cholinergic amacrine cell subtypes in the vertebrate retina

In mammals, retinal direction selectivity originates from GABAergic/cholinergic amacrine cells (ACs) specifically expressing the sox2 gene. However, the cellular diversity of GABAergic/cholinergic ACs of other vertebrate species remains largely unexplored. Here, we identified 2 morphologically and genetically distinct GABAergic/cholinergic AC types in zebrafish, a previously undescribed bhlhe22+ type and a mammalian counterpart sox2+ type. Notably, while sole sox2 disruption removed sox2+ type, the codisruption of bhlhe22 and bhlhe23 was required to remove bhlhe22+ type. Also, both types significantly differed in dendritic arbors, lamination, and soma position. Furthermore, in vivo two-photon calcium imaging and the behavior assay suggested the direction selectivity of both AC types. Nevertheless, the 2 types showed preferential responses to moving bars of different sizes. Thus, our findings provide new cellular diversity and functional characteristics of GABAergic/cholinergic ACs in the vertebrate retina.

Evolution of Brains and Computers: The Roads Not Taken

When computers started to become a dominant part of technology around the 1950s, fundamental questions about reliable designs and robustness were of great relevance. Their development gave rise to the exploration of new questions, such as what made brains reliable (since neurons can die) and how computers could get inspiration from neural systems. In parallel, the first artificial neural networks came to life. Since then, the comparative view between brains and computers has been developed in new, sometimes unexpected directions. With the rise of deep learning and the development of connectomics, an evolutionary look at how both hardware and neural complexity have evolved or designed is required. In this paper, we argue that important similarities have resulted both from convergent evolution (the inevitable outcome of architectural constraints) and inspiration of hardware and software principles guided by toy pictures of neurobiology. Moreover, dissimilarities and gaps originate from the lack of major innovations that have paved the way to biological computing (including brains) that are completely absent within the artificial domain. As it occurs within synthetic biocomputation, we can also ask whether alternative minds can emerge from A.I. designs. Here, we take an evolutionary view of the problem and discuss the remarkable convergences between living and artificial designs and what are the pre-conditions to achieve artificial intelligence.

Visual properties of human retinal ganglion cells

The retinal output is the sole source of visual information for the brain. Studies in non-primate mammals estimate that this information is carried by several dozens of retinal ganglion cell types, each informing the brain about different aspects of a visual scene. Even though morphological studies of primate retina suggest a similar diversity of ganglion cell types, research has focused on the function of only a few cell types. In human retina, recordings from individual cells are anecdotal or focus on a small subset of identified types. Here, we present the first systematic ex-vivo recording of light responses from 342 ganglion cells in human retinas obtained from donors. We find a great variety in the human retinal output in terms of preferences for positive or negative contrast, spatio-temporal frequency encoding, contrast sensitivity, and speed tuning. Some human ganglion cells showed similar response behavior as known cell types in other primate retinas, while we also recorded light responses that have not been described previously. This first extensive description of the human retinal output should facilitate interpretation of primate data and comparison to other mammalian species, and it lays the basis for the use of ex-vivo human retina for in-vitro analysis of novel treatment approaches.

Inhibitory components of retinal bipolar cell receptive fields are differentially modulated by dopamine D1 receptors

During adaptation to an increase in environmental luminance, retinal signaling adjustments are mediated by the neuromodulator dopamine. Retinal dopamine is released with light and can affect center-surround receptive fields, the coupling state between neurons, and inhibitory pathways through inhibitory receptors and neurotransmitter release. While the inhibitory receptive field surround of bipolar cells becomes narrower and weaker during light adaptation, it is unknown how dopamine affects bipolar cell surrounds. If dopamine and light have similar effects, it would suggest that dopamine could be a mechanism for light-adapted changes. We tested the hypothesis that dopamine D1 receptor activation is sufficient to elicit the magnitude of light-adapted reductions in inhibitory bipolar cell surrounds. Surrounds were measured from OFF bipolar cells in dark-adapted mouse retinas while stimulating D1 receptors, which are located on bipolar, horizontal, and inhibitory amacrine cells. The D1 agonist SKF-38393 narrowed and weakened OFF bipolar cell inhibitory receptive fields but not to the same extent as with light adaptation. However, the receptive field surround reductions differed between the glycinergic and GABAergic components of the receptive field. GABAergic inhibitory strength was reduced only at the edges of the surround, while glycinergic inhibitory strength was reduced across the whole receptive field. These results expand the role of retinal dopamine to include modulation of bipolar cell receptive field surrounds. Additionally, our results suggest that D1 receptor pathways may be a mechanism for the light-adapted weakening of glycinergic surround inputs and the furthest wide-field GABAergic inputs to bipolar cells. However, remaining differences between light-adapted and D1 receptor-activated inhibition demonstrate that non-D1 receptor mechanisms are necessary to elicit the full effect of light adaptation on inhibitory surrounds.

Evolutionary aspects of reservoir computing

Reservoir computing (RC) is a powerful computational paradigm that allows high versatility with cheap learning. While other artificial intelligence approaches need exhaustive resources to specify their inner workings, RC is based on a reservoir with highly nonlinear dynamics that does not require a fine tuning of its parts. These dynamics project input signals into high-dimensional spaces, where training linear readouts to extract input features is vastly simplified. Thus, inexpensive learning provides very powerful tools for decision-making, controlling dynamical systems, classification, etc. RC also facilitates solving multiple tasks in parallel, resulting in a high throughput. Existing literature focuses on applications in artificial intelligence and neuroscience. We review this literature from an evolutionary perspective. RC's versatility makes it a great candidate to solve outstanding problems in biology, which raises relevant questions. Is RC as abundant in nature as its advantages should imply? Has it evolved? Once evolved, can it be easily sustained? Under what circumstances? (In other words, is RC an evolutionarily stable computing paradigm?) To tackle these issues, we introduce a conceptual morphospace that would map computational selective pressures that could select for or against RC and other computing paradigms. This guides a speculative discussion about the questions above and allows us to propose a solid research line that brings together computation and evolution with RC as test model of the proposed hypotheses. This article is part of the theme issue 'Liquid brains, solid brains: How distributed cognitive architectures process information'.

Synaptic Contributions to Receptive Field Structure and Response Properties in the Rodent Lateral Geniculate Nucleus of the Thalamus

Comparative physiological and anatomical studies have greatly advanced our understanding of sensory systems. Many lines of evidence show that the murine lateral geniculate nucleus (LGN) has unique attributes, compared with other species such as cat and monkey. For example, in rodent, thalamic receptive field structure is markedly diverse, and many cells are sensitive to stimulus orientation and direction. To explore shared and different strategies of synaptic integration across species, we made whole-cell recordings in vivo from the murine LGN during the presentation of visual stimuli, analyzed the results with different computational approaches, and compared our findings with those from cat. As for carnivores, murine cells with classical center-surround receptive fields had a "push-pull" structure of excitation and inhibition within a given On or Off subregion. These cells compose the largest single population in the murine LGN (∼40%), indicating that push-pull is key in the form vision pathway across species. For two cell types with overlapping On and Off responses, which recalled either W3 or suppressed-by-contrast ganglion cells in murine retina, inhibition took a different form and was most pronounced for spatially extensive stimuli. Other On-Off cells were selective for stimulus orientation and direction. In these cases, retinal inputs were tuned and, for oriented cells, the second-order subunit of the receptive field predicted the preferred angle. By contrast, suppression was not tuned and appeared to sharpen stimulus selectivity. Together, our results provide new perspectives on the role of excitation and inhibition in retinothalamic processing.

SIGNIFICANCE STATEMENT

We explored the murine lateral geniculate nucleus from a comparative physiological perspective. In cat, most retinal cells have center-surround receptive fields and push-pull excitation and inhibition, including neurons with the smallest (highest acuity) receptive fields. The same is true for thalamic relay cells. In mouse retina, the most numerous cell type has the smallest receptive fields but lacks push-pull. The most common receptive field in rodent thalamus, however, is center-surround with push-pull. Thus, receptive field structure supersedes size per se for form vision. Further, for many orientation-selective cells, the second-order component of the receptive field aligned with stimulus preference, whereas suppression was untuned. Thus, inhibition may improve spatial resolution and sharpen other forms of selectivity in rodent lateral geniculate nucleus.

GlyRα2, not GlyRα3, modulates the receptive field surround of OFF retinal ganglion cells

Receptive fields (RFs) of most retinal ganglion cells (RGCs) consist of an excitatory center and suppressive surround. The RF center arises from the summation of excitatory bipolar cell glutamatergic inputs, whereas the surround arises from lateral inhibitory inputs. In the retina, both gamma amino butyric acid (GABA) and glycine are inhibitory neurotransmitters. A clear role for GABAergic inhibition modulating the RGC RF surround has been demonstrated across species. Glycinergic inhibition is more commonly associated with RF center modulation, although there is some evidence that it may contribute to the RF surround. The synaptic glycinergic chloride channels are formed by three homomeric β and two homomeric α subunits that can be glycine receptor (GlyR) α1, α2, α3, or α4. GlyRα composition is responsible for currents with distinct decay kinetics. Their expression within the inner plexiform laminae and neuronal subtypes also differ. We studied the role of GlyR subunit selective modulation of RGC RF surrounds, using mice lacking GlyRα2 (Glra2 -/-), GlyRα3 (Glra3 -/-), or both (Glra2/3 -/-). We chose this molecular genetic approach instead of pharmacological manipulation because there are no subunit selective antagonists and strychnine blocks all GlyRs. Comparisons of annulus-evoked responses among wild type (WT) and GlyRα knockouts (Glra2 -/-, Glra3 -/- and Glra2/3 -/-) show that GlyRα2 inhibition enhances RF surround suppression and post-stimulus excitation in only WT OFF RGCs. Similarities in the responses in Glra2 -/- and Glra2/3 -/- RGCs verify these conclusions. Based on previous and current data, we propose that GlyRα2-mediated input uses a crossover inhibitory circuit. Further, we suggest that GlyRα2 modulates the OFF RGC RF center and surround independently. In summary, our results define a selective GlyR subunit-specific control of RF surround suppression in OFF RGCs.

Light adaptation alters inner retinal inhibition to shape OFF retinal pathway signaling

The retina adjusts its signaling gain over a wide range of light levels. A functional result of this is increased visual acuity at brighter luminance levels (light adaptation) due to shifts in the excitatory center-inhibitory surround receptive field parameters of ganglion cells that increases their sensitivity to smaller light stimuli. Recent work supports the idea that changes in ganglion cell spatial sensitivity with background luminance are due in part to inner retinal mechanisms, possibly including modulation of inhibition onto bipolar cells. To determine how the receptive fields of OFF cone bipolar cells may contribute to changes in ganglion cell resolution, the spatial extent and magnitude of inhibitory and excitatory inputs were measured from OFF bipolar cells under dark- and light-adapted conditions. There was no change in the OFF bipolar cell excitatory input with light adaptation; however, the spatial distributions of inhibitory inputs, including both glycinergic and GABAergic sources, became significantly narrower, smaller, and more transient. The magnitude and size of the OFF bipolar cell center-surround receptive fields as well as light-adapted changes in resting membrane potential were incorporated into a spatial model of OFF bipolar cell output to the downstream ganglion cells, which predicted an increase in signal output strength with light adaptation. We show a prominent role for inner retinal spatial signals in modulating the modeled strength of bipolar cell output to potentially play a role in ganglion cell visual sensitivity and acuity.

The Synaptic and Morphological Basis of Orientation Selectivity in a Polyaxonal Amacrine Cell of the Rabbit Retina

Much of the computational power of the retina derives from the activity of amacrine cells, a large and diverse group of GABAergic and glycinergic inhibitory interneurons. Here, we identify an ON-type orientation-selective, wide-field, polyaxonal amacrine cell (PAC) in the rabbit retina and demonstrate how its orientation selectivity arises from the structure of the dendritic arbor and the pattern of excitatory and inhibitory inputs. Excitation from ON bipolar cells and inhibition arising from the OFF pathway converge to generate a quasi-linear integration of visual signals in the receptive field center. This serves to suppress responses to high spatial frequencies, thereby improving sensitivity to larger objects and enhancing orientation selectivity. Inhibition also regulates the magnitude and time course of excitatory inputs to this PAC through serial inhibitory connections onto the presynaptic terminals of ON bipolar cells. This presynaptic inhibition is driven by graded potentials within local microcircuits, similar in extent to the size of single bipolar cell receptive fields. Additional presynaptic inhibition is generated by spiking amacrine cells on a larger spatial scale covering several hundred microns. The orientation selectivity of this PAC may be a substrate for the inhibition that mediates orientation selectivity in some types of ganglion cells. Significance statement: The retina comprises numerous excitatory and inhibitory circuits that encode specific features in the visual scene, such as orientation, contrast, or motion. Here, we identify a wide-field inhibitory neuron that responds to visual stimuli of a particular orientation, a feature selectivity that is primarily due to the elongated shape of the dendritic arbor. Integration of convergent excitatory and inhibitory inputs from the ON and OFF visual pathways suppress responses to small objects and fine textures, thus enhancing selectivity for larger objects. Feedback inhibition regulates the strength and speed of excitation on both local and wide-field spatial scales. This study demonstrates how different synaptic inputs are regulated to tune a neuron to respond to specific features in the visual scene.

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.

Response Properties of a Newly Identified Tristratified Narrow Field Amacrine Cell in the Mouse Retina

Amacrine cells were targeted for whole cell recording using two-photon fluorescence microscopy in a transgenic mouse line in which the promoter for dopamine receptor 2 drove expression of green fluorescent protein in a narrow field tristratified amacrine cell (TNAC) that had not been studied previously. Light evoked a multiphasic response that was the sum of hyperpolarizing and depolarization synaptic inputs consistent with distinct dendritic ramifications in the off and on sublamina of the inner plexiform layer. The amplitude and waveform of the response, which consisted of an initial brief hyperpolarization at light onset followed by recovery to a plateau potential close to dark resting potential and a hyperpolarizing response at the light offset varied little over an intensity range from 0.4 to ~10^6 Rh*/rod/s. This suggests that the cell functions as a differentiator that generates an output signal (a transient reduction in inhibitory input to downstream retina neurons) that is proportional to the derivative of light input independent of its intensity. The underlying circuitry appears to consist of rod and cone driven on and off bipolar cells that provide direct excitatory input to the cell as well as to GABAergic amacrine cells that are synaptically coupled to TNAC. Canonical reagents that blocked excitatory (glutamatergic) and inhibitory (GABA and glycine) synaptic transmission had effects on responses to scotopic stimuli consistent with the rod driven component of the proposed circuit. However, responses evoked by photopic stimuli were paradoxical and could not be interpreted on the basis of conventional thinking about the neuropharmacology of synaptic interactions in the retina.

Sidekick 2 directs formation of a retinal circuit that detects differential motion

In the mammalian retina, processes of approximately 70 types of interneurons form specific synapses on roughly 30 types of retinal ganglion cells (RGCs) in a neuropil called the inner plexiform layer. Each RGC type extracts salient features from visual input, which are sent deeper into the brain for further processing. The specificity and stereotypy of synapses formed in the inner plexiform layer account for the feature-detecting ability of RGCs. Here we analyse the development and function of synapses on one mouse RGC type, called the W3B-RGC. These cells have the remarkable property of responding when the timing of the movement of a small object differs from that of the background, but not when they coincide. Such cells, known as local edge detectors or object motion sensors, can distinguish moving objects from a visual scene that is also moving. We show that W3B-RGCs receive strong and selective input from an unusual excitatory amacrine cell type known as VG3-AC (vesicular glutamate transporter 3). Both W3B-RGCs and VG3-ACs express the immunoglobulin superfamily recognition molecule sidekick 2 (Sdk2), and both loss- and gain-of-function studies indicate that Sdk2-dependent homophilic interactions are necessary for the selectivity of the connection. The Sdk2-specified synapse is essential for visual responses of W3B-RGCs: whereas bipolar cells relay visual input directly to most RGCs, the W3B-RGCs receive much of their input indirectly, via the VG3-ACs. This non-canonical circuit introduces a delay into the pathway from photoreceptors in the centre of the receptive field to W3B-RGCs, which could improve their ability to judge the synchrony of local and global motion.

APLP2 Regulates Refractive Error and Myopia Development in Mice and Humans

Myopia is the most common vision disorder and the leading cause of visual impairment worldwide. However, gene variants identified to date explain less than 10% of the variance in refractive error, leaving the majority of heritability unexplained (“missing heritability”). Previously, we reported that expression of APLP2 was strongly associated with myopia in a primate model. Here, we found that low-frequency variants near the 5’-end of APLP2 were associated with refractive error in a prospective UK birth cohort (n = 3,819 children; top SNP rs188663068, p = 5.0 × 10⁻⁴) and a CREAM consortium panel (n = 45,756 adults; top SNP rs7127037, p = 6.6 × 10⁻³). These variants showed evidence of differential effect on childhood longitudinal refractive error trajectories depending on time spent reading (gene x time spent reading x age interaction, p = 4.0 × 10⁻³). Furthermore, Aplp2 knockout mice developed high degrees of hyperopia (+11.5 ± 2.2 D, p < 1.0 × 10⁻⁴) compared to both heterozygous (-0.8 ± 2.0 D, p < 1.0 × 10⁻⁴) and wild-type (+0.3 ± 2.2 D, p < 1.0 × 10⁻⁴) littermates and exhibited a dose-dependent reduction in susceptibility to environmentally induced myopia (F(2, 33) = 191.0, p < 1.0 × 10⁻⁴). This phenotype was associated with reduced contrast sensitivity (F(12, 120) = 3.6, p = 1.5 × 10⁻⁴) and changes in the electrophysiological properties of retinal amacrine cells, which expressed Aplp2. This work identifies APLP2 as one of the “missing” myopia genes, demonstrating the importance of a low-frequency gene variant in the development of human myopia. It also demonstrates an important role for APLP2 in refractive development in mice and humans, suggesting a high level of evolutionary conservation of the signaling pathways underlying refractive eye development.

Gene variants identified by GWAS studies to date explain only a small fraction of myopia cases because myopia represents a complex disorder thought to be controlled by dozens or even hundreds of genes. The majority of genetic variants underlying myopia seems to be of small effect and/or low frequency, which makes them difficult to identify using classical genetic approaches, such as GWAS, alone. Here, we combined gene expression profiling in a monkey model of myopia, human GWAS, and a gene-targeted mouse model of myopia to identify one of the “missing” myopia genes, APLP2. We found that a low-frequency risk allele of APLP2 confers susceptibility to myopia only in children exposed to large amounts of daily reading, thus, providing an experimental example of the long-hypothesized gene-environment interaction between nearwork and genes underlying myopia. Functional analysis of APLP2 using an APLP2 knockout mouse model confirmed functional significance of APLP2 in refractive development and implicated a potential role of synaptic transmission at the level of glycinergic amacrine cells of the retina for the development of myopia. Furthermore, mouse studies revealed that lack of Aplp2 has a dose-dependent suppressive effect on susceptibility to form-deprivation myopia, providing a potential gene-specific target for therapeutic intervention to treat myopia.

High Accuracy Decoding of Dynamical Motion from a Large Retinal Population

Motion tracking is a challenge the visual system has to solve by reading out the retinal population. It is still unclear how the information from different neurons can be combined together to estimate the position of an object. Here we recorded a large population of ganglion cells in a dense patch of salamander and guinea pig retinas while displaying a bar moving diffusively. We show that the bar's position can be reconstructed from retinal activity with a precision in the hyperacuity regime using a linear decoder acting on 100+ cells. We then took advantage of this unprecedented precision to explore the spatial structure of the retina's population code. The classical view would have suggested that the firing rates of the cells form a moving hill of activity tracking the bar's position. Instead, we found that most ganglion cells in the salamander fired sparsely and idiosyncratically, so that their neural image did not track the bar. Furthermore, ganglion cell activity spanned an area much larger than predicted by their receptive fields, with cells coding for motion far in their surround. As a result, population redundancy was high, and we could find multiple, disjoint subsets of neurons that encoded the trajectory with high precision. This organization allows for diverse collections of ganglion cells to represent high-accuracy motion information in a form easily read out by downstream neural circuits.

Broad thorny ganglion cells: a candidate for visual pursuit error signaling in the primate retina

Functional analyses exist only for a few of the morphologically described primate ganglion cell types, and their correlates in other mammalian species remain elusive. Here, we recorded light responses of broad thorny cells in the whole-mounted macaque retina. They showed ON-OFF-center light responses that were strongly suppressed by stimulation of the receptive field surround. Spike responses were delayed compared with parasol ganglion cells and other ON-OFF cells, including recursive bistratified ganglion cells and A1 amacrine cells. The receptive field structure was shaped by direct excitatory synaptic input and strong presynaptic and postsynaptic inhibition in both ON and OFF pathways. The cells responded strongly to dark or bright stimuli moving either in or out of the receptive field, independent of the direction of motion. However, they did not show a maintained spike response either to a uniform background or to a drifting plaid pattern. These properties could be ideally suited for guiding movements involved in visual pursuit. The functional characteristics reported here permit the first direct cross-species comparison of putative homologous ganglion cell types. Based on morphological similarities, broad thorny ganglion cells have been proposed to be homologs of rabbit local edge detector ganglion cells, but we now show that the two cells have quite distinct physiological properties. Thus, our data argue against broad thorny cells as the homologs of local edge detector cells.

An excitatory amacrine cell detects object motion and provides feature-selective input to ganglion cells in the mouse retina

Retinal circuits detect salient features of the visual world and report them to the brain through spike trains of retinal ganglion cells. The most abundant ganglion cell type in mice, the so-called W3 ganglion cell, selectively responds to movements of small objects. Where and how object motion sensitivity arises in the retina is incompletely understood. In this study, we use 2-photon-guided patch-clamp recordings to characterize responses of vesicular glutamate transporter 3 (VGluT3)-expressing amacrine cells (ACs) to a broad set of visual stimuli. We find that these ACs are object motion sensitive and analyze the synaptic mechanisms underlying this computation. Anatomical circuit reconstructions suggest that VGluT3-expressing ACs form glutamatergic synapses with W3 ganglion cells, and targeted recordings show that the tuning of W3 ganglion cells' excitatory input matches that of VGluT3-expressing ACs' responses. Synaptic excitation of W3 ganglion cells is diminished, and responses to object motion are suppressed in mice lacking VGluT3. Object motion, thus, is first detected by VGluT3-expressing ACs, which provide feature-selective excitatory input to W3 ganglion cells.

DOI:http://dx.doi.org/10.7554/eLife.08025.001

Animals can use their eyes to detect moving objects, which helps them to avoid predators and other threats, and to spot potential prey or allies. Visual information from the eyes is sent to the brain, which processes the information to form a coherent picture of how the objects are moving. This processing has to be able to account for movements of the head, eyes, and body—which can cause the image of an object on the retina within the eye to move even if the object itself remains stationary.

Within the retina, light is converted into electrical signals by cells called rods and cones. A layer of cells called bipolar cells relay these signals to the ‘ganglion’ cells, which in turn pass them on to the brain. In mice, a type of ganglion cell called the W3 ganglion cell has been shown to respond selectively to small moving objects, but exactly how these cells acquire their motion sensitivity remained unclear.

Kim et al. now reveal that cells called amacrine cells, which regulate the transfer of signals from the bipolar cells to ganglion cells, supply the information needed for motion detection. The mouse eye contains up to 50 different types of amacrine cells. One of these—called the VG3-amacrine cell—increases its activity whenever an object moves relative to its background, but decreases its activity whenever the object and background move together. The overall effect is that the cells respond selectively to the presence of small moving objects.

Most amacrine cells regulate the transfer of signals within the retina by inhibiting the activity of ganglion cells. But, Kim et al. show that VG3-amacrine cells release a molecule called glutamate to activate W3 ganglion cells when a moving object is detected. These unusual and specialized cells are, thus, an essential component of a circuit in the nervous system that supports motion detection. It is possible that some other types of amacrine cells may also play specialized roles in the detection of other features in the visual world.

DOI:http://dx.doi.org/10.7554/eLife.08025.002

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.

GABAergic neurotransmission and retinal ganglion cell function

Ganglion cells are the output retinal neurons that convey visual information to the brain. There are ~20 different types of ganglion cells, each encoding a specific aspect of the visual scene as spatial and temporal contrast, orientation, direction of movement, presence of looming stimuli; etc. Ganglion cell functioning depends on the intrinsic properties of ganglion cell's membrane as well as on the excitatory and inhibitory inputs that these cells receive from other retinal neurons. GABA is one of the most abundant inhibitory neurotransmitters in the retina. How it modulates the activity of different types of ganglion cells and what is its significance in extracting the basic features from visual scene are questions with fundamental importance in visual neuroscience. The present review summarizes current data concerning the types of membrane receptors that mediate GABA action in proximal retina; the effects of GABA and its antagonists on the ganglion cell light-evoked postsynaptic potentials and spike discharges; the action of GABAergic agents on centre-surround organization of the receptive fields and feature related ganglion cell activity. Special emphasis is put on the GABA action regarding the ON-OFF and sustained-transient ganglion cell dichotomy in both nonmammalian and mammalian retina.

The neural circuit mechanisms underlying the retinal response to motion reversal

To make up for delays in visual processing, retinal circuitry effectively predicts that a moving object will continue moving in a straight line, allowing retinal ganglion cells to anticipate the object's position. However, a sudden reversal of motion triggers a synchronous burst of firing from a large group of ganglion cells, possibly signaling a violation of the retina's motion prediction. To investigate the neural circuitry underlying this response, we used a combination of multielectrode array and whole-cell patch recordings to measure the responses of individual retinal ganglion cells in the tiger salamander to reversing stimuli. We found that different populations of ganglion cells were responsible for responding to the reversal of different kinds of objects, such as bright versus dark objects. Using pharmacology and designed stimuli, we concluded that ON and OFF bipolar cells both contributed to the reversal response, but that amacrine cells had, at best, a minor role. This allowed us to formulate an adaptive cascade model (ACM), similar to the one previously used to describe ganglion cell responses to motion onset. By incorporating the ON pathway into the ACM, we were able to reproduce the time-varying firing rate of fast OFF ganglion cells for all experimentally tested stimuli. Analysis of the ACM demonstrates that bipolar cell gain control is primarily responsible for generating the synchronized retinal response, as individual bipolar cells require a constant time delay before recovering from gain control.

A dedicated visual pathway for prey detection in larval zebrafish

Zebrafish larvae show characteristic prey capture behavior in response to small moving objects. The neural mechanism used to recognize objects as prey remains largely unknown. We devised a machine learning behavior classification system to quantify hunting kinematics in semi-restrained animals exposed to a range of virtual stimuli. Two-photon calcium imaging revealed a small visual area, AF7, that was activated specifically by the optimal prey stimulus. This pretectal region is innervated by two types of retinal ganglion cells, which also send collaterals to the optic tectum. Laser ablation of AF7 markedly reduced prey capture behavior. We identified neurons with arbors in AF7 and found that they projected to multiple sensory and premotor areas: the optic tectum, the nucleus of the medial longitudinal fasciculus (nMLF) and the hindbrain. These findings indicate that computations in the retina give rise to a visual stream which transforms sensory information into a directed prey capture response.

DOI:http://dx.doi.org/10.7554/eLife.04878.001

Our ability to recognize objects, and to respond instinctively to them, is something that is not fully understood. For example, seeing your favorite dessert could trigger an irresistible urge to eat it. Yet precisely how the image of the dessert could trigger an inner desire to indulge is a question that has so far eluded scientists. This compelling question also applies to the animal kingdom. Predators often demonstrate a typical hunting behavior upon seeing their prey from a distance. But just how the image of the prey triggers this hunting behavior is not known.

Semmelhack et al. have now investigated this question by looking at the hunting behavior of zebrafish larvae. The larvae's prey is a tiny microbe that resembles a small moving dot. When the larvae encounter something that looks like their prey, they demonstrate a hardwired hunting response towards it. The hunting behavior consists of a series of swimming maneuvers to help the larvae successfully capture their prey.

Semmelhack et al. used prey decoys to lure the zebrafish larvae, and video recordings to monitor the larvae's response. During the recordings, the larvae were embedded in a bed of jelly with only their tails free to move. The larvae's tail movements were recorded, and because the larvae are completely transparent, their brain activity could be visually monitored at the same time using calcium dyes.

Using this approach, Semmelhack et al. identified a specific area of the brain that is responsible for triggering the larvae's hunting behavior. It turns out that this brain region forms a circuit that directly connects the retina at the back of the eye to nerve centers that control hunting maneuvers. So when the larva sees its prey, this circuit could directly trigger the larva's hunting behavior. When the circuit was specifically destroyed with a laser, this instinctive hunting response was impaired.

These findings suggest that predators have a distinct brain circuit that hardwires their hunting response to images of their prey. Future studies would involve understanding precisely how this circuit coordinates the larvae's complex hunting behavior.

DOI:http://dx.doi.org/10.7554/eLife.04878.002

Adeno-associated virus-RNAi of GlyRα1 and characterization of its synapse-specific inhibition in OFF alpha transient retinal ganglion cells

In the central nervous system, inhibition shapes neuronal excitation. In spinal cord glycinergic inhibition predominates, whereas GABAergic inhibition predominates in the brain. The retina uses GABA and glycine in approximately equal proportions. Glycinergic crossover inhibition, initiated in the On retinal pathway, controls glutamate release from presynaptic OFF cone bipolar cells (CBCs) and directly shapes temporal response properties of OFF retinal ganglion cells (RGCs). In the retina, four glycine receptor (GlyR) α-subunit isoforms are expressed in different sublaminae and their synaptic currents differ in decay kinetics. GlyRα1, expressed in both On and Off sublaminae of the inner plexiform layer, could be the glycinergic isoform that mediates On-to-Off crossover inhibition. However, subunit-selective glycine contributions remain unknown because we lack selective antagonists or cell class-specific subunit knockouts. To examine the role of GlyRα1 in direct inhibition in mature RGCs, we used retrogradely transported adeno-associated virus (AAV) that performed RNAi and eliminated almost all glycinergic spontaneous and visually evoked responses in PV5 (OFFα(Transient)) RGCs. Comparisons of responses in PV5 RGCs infected with AAV-scrambled-short hairpin RNA (shRNA) or AAV-Glra1-shRNA confirm a role for GlyRα1 in crossover inhibition in cone-driven circuits. Our results also define a role for direct GlyRα1 inhibition in setting the resting membrane potential of PV5 RGCs. The absence of GlyRα1 input unmasked a serial and a direct feedforward GABA(A)ergic modulation in PV5 RGCs, reflecting a complex interaction between glycinergic and GABA(A)ergic inhibition.

NMDA and AMPA receptors contribute similarly to temporal processing in mammalian retinal ganglion cells

Postsynaptic AMPA- and NMDA-type glutamate receptors (AMPARs, NMDARs) are commonly expressed at the same synapses. AMPARs are thought to mediate the majority of fast excitatory neurotransmission whereas NMDARs, with their relatively slower kinetics and higher Ca(2+) permeability, are thought to mediate synaptic plasticity, especially in neural circuits devoted to learning and memory. In sensory neurons, however, the roles of AMPARs and NMDARs are less well understood. Here, we tested in the in vitro guinea pig retina whether AMPARs and NMDARs differentially support temporal contrast encoding by two ganglion cell types. In both OFF Alpha and Delta ganglion cells, contrast stimulation evoked an NMDAR-mediated response with a characteristic J-shaped I-V relationship. In OFF Delta cells, AMPAR- and NMDAR-mediated responses could be modulated at low frequencies but were suppressed during 10 Hz stimulation, when responses were instead shaped by synaptic inhibition. With inhibition blocked, both AMPAR- and NMDAR-mediated responses could be modulated at 10 Hz, indicating that NMDAR kinetics do not limit temporal encoding. In OFF Alpha cells, NMDAR-mediated responses followed stimuli at frequencies up to ∼18 Hz. In both cell types, NMDAR-mediated responses to contrast modulation at 9-18 Hz showed delays of <10 ms relative to AMPAR-mediated responses. Thus, NMDARs combine with AMPARs to encode rapidly modulated glutamate release, and NMDAR kinetics do not limit temporal coding by OFF Alpha and Delta ganglion cells substantially. Furthermore, glutamatergic transmission is differentially regulated across bipolar cell pathways: in some, release is suppressed at high temporal frequencies by presynaptic inhibition.

Distinct roles for inhibition in spatial and temporal tuning of local edge detectors in the rabbit retina

This paper examines the role of inhibition in generating the receptive-field properties of local edge detector (LED) ganglion cells in the rabbit retina. We confirm that the feed-forward inhibition is largely glycinergic but, contrary to a recent report, our data demonstrate that the glycinergic inhibition contributes to temporal tuning for the OFF and ON inputs to the LEDs by delaying the onset of spiking; this delay was more pronounced for the ON inputs (∼340 ms) than the OFF inputs (∼12 ms). Blocking glycinergic transmission reduced the delay to spike onset and increased the responses to flickering stimuli at high frequencies. Analysis of the synaptic conductances indicates that glycinergic amacrine cells affect temporal tuning through both postsynaptic inhibition of the LEDs and presynaptic modulation of the bipolar cells that drive the LEDs. The results also confirm that presynaptic GABAergic transmission contributes significantly to the concentric surround antagonism in LEDs; however, unlike presumed LEDs in the mouse retina, the surround is only partly generated by spiking amacrine cells.

Transformation of visual signals by inhibitory interneurons in retinal circuits

One of the largest mysteries of the brain lies in understanding how higher-level computations are implemented by lower-level operations in neurons and synapses. In particular, in many brain regions inhibitory interneurons represent a diverse class of cells, the individual functional roles of which are unknown. We discuss here how the operations of inhibitory interneurons influence the behavior of a circuit, focusing on recent results in the vertebrate retina. A key role in this understanding is played by a common representation of the visual stimulus that can be applied at different stages. By considering how this stimulus representation changes at each location in the circuit, we can understand how neuron-level operations such as thresholds and inhibition yield circuit-level computations such as how stimulus selectivity and gain are controlled by local and peripheral visual stimuli.

Diverse visual features encoded in mouse lateral geniculate nucleus

The thalamus is crucial in determining the sensory information conveyed to cortex. In the visual system, the thalamic lateral geniculate nucleus (LGN) is generally thought to encode simple center-surround receptive fields, which are combined into more sophisticated features in cortex, such as orientation and direction selectivity. However, recent evidence suggests that a more diverse set of retinal ganglion cells projects to the LGN. We therefore used multisite extracellular recordings to define the repertoire of visual features represented in the LGN of mouse, an emerging model for visual processing. In addition to center-surround cells, we discovered a substantial population with more selective coding properties, including direction and orientation selectivity, as well as neurons that signal absence of contrast in a visual scene. The direction and orientation selective neurons were enriched in regions that match the termination zones of direction selective ganglion cells from the retina, suggesting a source for their tuning. Together, these data demonstrate that the mouse LGN contains a far more elaborate representation of the visual scene than current models posit. These findings should therefore have a significant impact on our understanding of the computations performed in mouse visual cortex.

Inhibitory mechanisms that generate centre and surround properties in ON and OFF brisk-sustained ganglion cells in the rabbit retina

Lateral inhibition produces the centre-surround organization of retinal receptive fields, in which inhibition driven by the mean luminance enhances the sensitivity of ganglion cells to spatial and temporal contrast. Surround inhibition is generated in both synaptic layers; however, the synaptic mechanisms within the inner plexiform layer are not well characterized within specific classes of retinal ganglion cell. Here, we compared the synaptic circuits generating concentric centre-surround receptive fields in ON and OFF brisk-sustained ganglion cells (BSGCs) in the rabbit retina. We first characterized the synaptic inputs to the centre of ON BSGCs, for comparison with previous results from OFF BSGCs. Similar to wide-field ganglion cells, the spatial extent of the excitatory centre and inhibitory surround was larger for the ON than the OFF BSGCs. The results indicate that the surrounds of ON and OFF BSGCs are generated in both the outer and the inner plexiform layers. The inner plexiform layer surround inhibition comprised GABAergic suppression of excitatory inputs from bipolar cells. However, ON and OFF BSGCs displayed notable differences. Surround suppression of excitatory inputs was weaker in ON than OFF BSGCs, and was mediated largely by GABA(C) receptors in ON BSGCs, and by both GABA(A) and GABA(C) receptors in OFF BSGCs. Large ON pathway-mediated glycinergic inputs to ON and OFF BSGCs also showed surround suppression, while much smaller GABAergic inputs showed weak, if any, spatial tuning. Unlike OFF BSGCs, which receive strong glycinergic crossover inhibition from the ON pathway, the ON BSGCs do not receive crossover inhibition from the OFF pathway. We compare and discuss possible roles for glycinergic inhibition in the two cell types.

The most numerous ganglion cell type of the mouse retina is a selective feature detector

The retina reports the visual scene to the brain through many parallel channels, each carried by a distinct population of retinal ganglion cells. Among these, the population with the smallest and densest receptive fields encodes the neural image with highest resolution. In human retina, and those of cat and macaque, these high-resolution ganglion cells act as generic pixel encoders: They serve to represent many different visual inputs and convey a neural image of the scene downstream for further processing. Here we identify and analyze high-resolution ganglion cells in the mouse retina, using a transgenic line in which these cells, called "W3", are labeled fluorescently. Counter to the expectation, these ganglion cells do not participate in encoding generic visual scenes, but remain silent during most common visual stimuli. A detailed study of their response properties showed that W3 cells pool rectified excitation from both On and Off bipolar cells, which makes them sensitive to local motion. However, they also receive unusually strong lateral inhibition, both pre- and postsynaptically, triggered by distant motion. As a result, the W3 cell can detect small moving objects down to the receptive field size of bipolar cells, but only if the background is featureless or stationary--an unusual condition. A survey of naturalistic stimuli shows that W3 cells may serve as alarm neurons for overhead predators.

Receptor targets of amacrine cells

Amacrine cells are a morphologically and functionally diverse group of inhibitory interneurons. Morphologically, they have been divided into approximately 30 types. Although this diversity is probably important to the fine structure and function of the retinal circuit, the amacrine cells have been more generally divided into two subclasses. Glycinergic narrow-field amacrine cells have dendrites that ramify close to their somas, cross the sublaminae of the inner plexiform layer, and create cross talk between its parallel ON and OFF pathways. GABAergic wide-field amacrine cells have dendrites that stretch long distances from their soma but ramify narrowly within an inner plexiform layer sublamina. These wide-field cells are thought to mediate inhibition within a sublamina and thus within the ON or OFF pathway. The postsynaptic targets of all amacrine cell types include bipolar, ganglion, and other amacrine cells. Almost all amacrine cells use GABA or glycine as their primary neurotransmitter, and their postsynaptic receptor targets include the most common GABA(A), GABA(C), and glycine subunit receptor configurations. This review addresses the diversity of amacrine cells, the postsynaptic receptors on their target cells in the inner plexiform layer of the retina, and some of the inhibitory mechanisms that arise as a result. When possible, the effects of GABAergic and glycinergic inputs on the visually evoked responses of their postsynaptic targets are discussed.

Selective glycine receptor α2 subunit control of crossover inhibition between the on and off retinal pathways

In the retina, the receptive fields (RFs) of almost all ganglion cells (GCs) are comprised of an excitatory center and a suppressive surround. The RF center arises from local excitatory bipolar cell (BC) inputs and the surround from lateral inhibitory inputs. Selective antagonists have been used to define the roles of GABA(A) and GABA(C) receptor-mediated input in RF organization. In contrast, the role of glycine receptor (GlyR) subunit-specific inhibition is less clear because the only antagonist, strychnine, blocks all GlyR subunit combinations. We used mice lacking the GlyRα2 (Glra2(-/-)) and GlyRα3 (Glra3(-/-)) subunits, or both (Glra2/3(-/-)), to explore their roles in GC RF organization. By comparing spontaneous and visually evoked responses of WT with Glra2(-/-), Glra3(-/-) and Glra2/3(-/-) ON- and OFF-center GCs, we found that both GlyRα2 and GlyRα3 modulate local RF interactions. In the On pathway, both receptors enhance the excitatory center response; however, the underlying inhibitory mechanisms differ. GlyRα2 participates in crossover inhibition, whereas GlyRα3 mediates serial inhibition. In the Off pathway, GlyRα2 plays a similar role, again using crossover inhibition and enhancing excitatory responses within the RF center. Comparisons of single and double KOs indicate that GlyRα2 and GlyRα3 inhibition are independent and additive, consistent with the finding that they use different inhibitory circuitry. These findings are the first to define GlyR subunit-specific control of visual function and GlyRα2 subunit-specific control of crossover inhibition in the retina.

Amacrine-to-amacrine cell inhibition: Spatiotemporal properties of GABA and glycine pathways

We measured the spatial and temporal properties of GABAergic and glycinergic inhibition to amacrine cells in the whole-mount rabbit retina. The amacrine cells were parsed into two morphological classes: narrow-field cells with processes spreading less than 200 μm and wide-field cells with processes extending more than 300 μm. The inhibition was also parsed into two types: sustained glycine and transient GABA. Narrow-field amacrine cells receive 1) very transient GABAergic inhibition with a fast onset latency of 140 ± 16 ms decaying to 30% of the peak level within 208 ± 27 ms elicited broadly over a lateral distance of up to 1500 μm and 2) sustained glycinergic inhibition with a medium onset latency of 286 ± 23 ms that was elicited over a spatial area often broader than the processes of the narrow-field amacrine cells. Wide-field amacrine cells received sustained glycinergic inhibition but no broad transient GABAergic inhibition. Surprisingly, neither of these amacrine cell classes received sustained local GABAergic inhibition, commonly found in an earlier study of ganglion cells.

Trigger features and excitation in the retina

This review focuses on recent advances in our understanding of how neural divergence and convergence give rise to complex encoding properties of retinal ganglion cells. We describe the apparent mismatch between the number of cone bipolar cell types, and the diversity of excitatory input to retinal ganglion cells, and outline two possible solutions. One proposal is for diversity in the excitatory pathways to be generated within axon terminals of cone bipolar cells, and the second invokes narrow-field glycinergic amacrine cells that can apparently act like bipolar cells by providing excitatory drive to ganglion cells. Finally we highlight two advances in technique that promise to provide future insights; automation of electron microscope data collection and analysis, and the use of the ideal observer to quantitatively compare neural performance at all levels.

The retinal hypercircuit: a repeating synaptic interactive motif underlying visual function

Abstract

The vertebrate retina generates a stack of about a dozen different movies that represent the visual world as dynamic neural images or movies. The stack is embodied as separate strata that span the inner plexiform layer (IPL). At each stratum, ganglion cell dendrites reach up to read out inhibitory interactions between three different amacrine cell classes that shape bipolar-to-ganglion cell transmission. The nexus of these five cell classes represents a functional module, a retinal ‘hypercircuit’, that is repeated across the surface of each of the dozen strata that span the depth of the IPL. Individual differences in the characteristics of each cell class at each stratum lead to the unique processing characteristics of each neural image throughout the stack. This review shows how the interactions between the morphological and physiological characteristics of each cell class generate many of the known retinal visual functions including motion detection, directional selectivity, local edge detection, looming detection, object motion and looming detection.