Computational model of the on-alpha ganglion cell receptive field based on bipolar cell circuitry

Synaptic Origins of the Complex Receptive Field Structure in Primate Smooth Monostratified Retinal Ganglion Cells

Considerable progress has been made in studying the receptive fields of the most common primate retinal ganglion cell (RGC) types, such as parasol RGCs. Much less is known about the rarer primate RGC types and the circuitry that gives rise to noncanonical receptive field structures. The goal of this study was to analyze synaptic inputs to smooth monostratified RGCs to determine the origins of their complex spatial receptive fields, which contain isolated regions of high sensitivity called "hotspots." Interestingly, smooth monostratified RGCs co-stratify with the well-studied parasol RGCs and are thus constrained to receiving input from bipolar and amacrine cells with processes sharing the same layer, raising the question of how their functional differences originate. Through 3D reconstructions of circuitry and synapses onto ON smooth monostratified and ON parasol RGCs from central macaque retina, we identified four distinct sampling strategies employed by smooth and parasol RGCs to extract diverse response properties from co-stratifying bipolar and amacrine cells. The two RGC types differed in the proportion of amacrine cell input, relative contributions of co-stratifying bipolar cell types, amount of synaptic input per bipolar cell, and spatial distribution of bipolar cell synapses. Our results indicate that the smooth RGC's complex receptive field structure arises through spatial asymmetries in excitatory bipolar cell input which formed several discrete clusters comparable with physiologically measured hotspots. Taken together, our results demonstrate how the striking differences between ON parasol and ON smooth monostratified RGCs arise from distinct strategies for sampling a common set of synaptic inputs.

Differences in spike generation instead of synaptic inputs determine the feature selectivity of two retinal cell types

Retinal ganglion cells (RGCs) are the spiking projection neurons of the eye that encode different features of the visual environment. The circuits providing synaptic input to different RGC types to drive feature selectivity have been studied extensively, but there has been less research aimed at understanding the intrinsic properties and how they impact feature selectivity. We introduce an RGC type in the mouse, the Bursty Suppressed-by-Contrast (bSbC) RGC, and compared it to the OFF sustained alpha (OFFsA). Differences in their contrast response functions arose from differences not in synaptic inputs but in their intrinsic properties. Spike generation was the key intrinsic property behind this functional difference; the bSbC RGC undergoes depolarization block while the OFFsA RGC maintains a high spike rate. Our results demonstrate that differences in intrinsic properties allow these two RGC types to detect and relay distinct features of an identical visual stimulus to the brain.

Regional Variation of Gap Junctional Connections in the Mammalian Inner Retina

The retinas of many species show regional specialisations that are evident in the differences in the processing of visual input from different parts of the visual field. Regional specialisation is thought to reflect an adaptation to the natural visual environment, optical constraints, and lifestyle of the species. Yet, little is known about regional differences in synaptic circuitry. Here, we were interested in the topographical distribution of connexin-36 (Cx36), the major constituent of electrical synapses in the retina. We compared the retinas of mice, rats, and cats to include species with different patterns of regional specialisations in the analysis. First, we used the density of Prox1-immunoreactive amacrine cells as a marker of any regional specialisation, with higher cell density signifying more central regions. Double-labelling experiments showed that Prox1 is expressed in AII amacrine cells in all three species. Interestingly, large Cx36 plaques were attached to about 8-10% of Prox1-positive amacrine cell somata, suggesting the strong electrical coupling of pairs or small clusters of cell bodies. When analysing the regional changes in the volumetric density of Cx36-immunoreactive plaques, we found a tight correlation with the density of Prox1-expressing amacrine cells in the ON, but not in the OFF sublamina in all three species. The results suggest that the relative contribution of electrical synapses to the ON- and OFF-pathways of the retina changes with retinal location, which may contribute to functional ON/OFF asymmetries across the visual field.

Visual Deprivation Retards the Maturation of Dendritic Fields and Receptive Fields of Mouse Retinal Ganglion Cells

It was well documented that both the size of the dendritic field and receptive field of retinal ganglion cells (RGCs) are developmentally regulated in the mammalian retina, and visual stimulation is required for the maturation of the dendritic and receptive fields of mouse RGCs. However, it is not clear whether the developmental changes of the RGC receptive field correlate with the dendritic field and whether visual stimulation regulates the maturation of the dendritic field and receptive field of RGCs in a correlated manner. The present work demonstrated that both the dendritic and receptive fields of RGCs continuously develop after eye opening. However, the correlation between the developmental changes in the receptive field size and the dendritic field varies among different RGC types. These results suggest a continuous change of synaptic converging of RGC synaptic inputs in an RGC type-dependent manner. Besides, light deprivation impairs both the development of dendritic and receptive fields.

Ocular Hypertension Drives Remodeling of AMPA Receptors in Select Populations of Retinal Ganglion Cells

AMPA-type glutamate receptors in the CNS are normally impermeable to Ca2+, but the aberrant expression of Ca2+-permeable AMPA receptors (CP-AMPARs) occurs in pathological conditions such as ischemia or epilepsy, or degenerative diseases such as ALS. Here, we show that select populations of retinal ganglion cells (RGCs) similarly express high levels of CP-AMPARs in a mouse model of glaucoma. CP-AMPAR expression increased dramatically in both On sustained alpha and Off transient alpha RGCs, and this increase was prevented by genomic editing of the GluA2 subunit. On sustained alpha RGCs with elevated CP-AMPAR levels displayed profound synaptic depression, which was reduced by selectively blocking CP-AMPARs, buffering Ca2+ with BAPTA, or with the CB1 antagonist AM251, suggesting that depression was mediated by a retrograde transmitter which might be triggered by the influx of Ca2+ through CP-AMPARs. Thus, glaucoma may alter the composition of AMPARs and depress excitatory synaptic input in select populations of RGCs.

Diverse inhibitory and excitatory mechanisms shape temporal tuning in transient OFF α ganglion cells in the rabbit retina

KEY POINTS

Neurons combine excitatory and inhibitory signals to perform computations. In the retina, interactions between excitation and inhibition enable neurons to detect specific visual features. We describe how several excitatory and inhibitory mechanisms work together to allow transient OFF α ganglion cells in the rabbit retina to respond selectively to high temporal frequencies and thus detect faster image motion. The weightings of these different mechanisms change with the contrast and spatiotemporal properties of the visual input, and thereby support temporal tuning in α cells over a range of visual conditions. The results help us understand how ganglion cells selectively integrate excitatory and inhibitory signals to extract specific information from the visual input.

ABSTRACT

The 20 to 30 types of ganglion cell in the mammalian retina represent parallel signalling pathways that convey different information to the brain. α ganglion cells are selective for high temporal frequencies in visual inputs, which makes them particularly sensitive to rapid motion. Although α ganglion cells have been studied in several species, the synaptic basis for their selective temporal tuning remains unclear. Here, we analyse excitatory synaptic inputs to transient OFF α ganglion cells (t-OFF α GCs) in the rabbit retina. We show that convergence of excitatory and inhibitory synaptic inputs within the bipolar cell terminals presynaptic to the t-OFF α GCs shifts the temporal tuning to higher temporal frequencies. GABAergic inhibition suppresses the excitatory input at low frequencies, but potentiates it at high frequencies. Crossover glycinergic inhibition and sodium channel activity in the presynaptic bipolar cells also potentiate high frequency excitatory inputs. We found differences in the spatial and temporal properties, and contrast sensitivities of these mechanisms. These differences in stimulus selectivity allow these mechanisms to generate bandpass temporal tuning of t-OFF α GCs over a range of visual conditions.

Circuit Mechanisms of a Retinal Ganglion Cell with Stimulus-Dependent Response Latency and Activation Beyond Its Dendrites

Center-surround antagonism has been used as the canonical model to describe receptive fields of retinal ganglion cells (RGCs) for decades. We describe a newly identified RGC type in the mouse, called the ON delayed (OND) RGC, with receptive field properties that deviate from center-surround organization. Responding with an unusually long latency to light stimulation, OND RGCs respond earlier as the visual stimulus increases in size. Furthermore, OND RGCs are excited by light falling far beyond their dendrites. We unravel details of the circuit mechanisms behind these phenomena, revealing new roles for inhibition in controlling both temporal and spatial receptive field properties. The non-canonical receptive field properties of the OND RGC-integration of long temporal and large spatial scales-suggest that unlike typical RGCs, it may encode a slowly varying, global property of the visual scene.

Spatial relationships between GABAergic and glutamatergic synapses on the dendrites of distinct types of mouse retinal ganglion cells across development

Neuronal output requires a concerted balance between excitatory and inhibitory (I/E) input. Like other circuits, inhibitory synaptogenesis in the retina precedes excitatory synaptogenesis. How then do neurons attain their mature balance of I/E ratios despite temporal offset in synaptogenesis? To directly compare the development of glutamatergic and GABAergic synapses onto the same cell, we biolistically transfected retinal ganglion cells (RGCs) with PSD95CFP, a marker of glutamatergic postsynaptic sites, in transgenic Thy1­YFPγ2 mice in which GABA_A receptors are fluorescently tagged. We mapped YFPγ2 and PSD95CFP puncta distributions on three RGC types at postnatal day P12, shortly before eye opening, and at P21 when robust light responses in RGCs are present. The mature I_GABA/E ratios varied among ON-Sustained (S) A-type, OFF-S A-type, and bistratified direction selective (DS) RGCs. These ratios were attained at different rates, before eye-opening for ON-S and OFF-S A-type, and after eye-opening for DS RGCs. At both ages examined, the I_GABA/E ratio was uniform across the arbors of the three RGC types. Furthermore, measurements of the distances between neighboring PSD95CFP and YFPγ2 puncta on RGC dendrites indicate that their local relationship is established early in development, and cannot be predicted by random organization. These close spatial associations between glutamatergic and GABAergic postsynaptic sites appear to represent local synaptic arrangements revealed by correlative light and EM reconstructions of a single RGC's dendrites. Thus, although RGC types have different I_GABA/E ratios and establish these ratios at separate rates, the local relationship between excitatory and inhibitory inputs appear similarly constrained across the RGC types studied.

Directional summation in non-direction selective retinal ganglion cells

Retinal ganglion cells receive inputs from multiple bipolar cells which must be integrated before a decision to fire is made. Theoretical studies have provided clues about how this integration is accomplished but have not directly determined the rules regulating summation of closely timed inputs along single or multiple dendrites. Here we have examined dendritic summation of multiple inputs along On ganglion cell dendrites in whole mount rat retina. We activated inputs at targeted locations by uncaging glutamate sequentially to generate apparent motion along On ganglion cell dendrites in whole mount retina. Summation was directional and dependent13 on input sequence. Input moving away from the soma (centrifugal) resulted in supralinear summation, while activation sequences moving toward the soma (centripetal) were linear. Enhanced summation for centrifugal activation was robust as it was also observed in cultured retinal ganglion cells. This directional summation was dependent on hyperpolarization activated cyclic nucleotide-gated (HCN) channels as blockade with ZD7288 eliminated directionality. A computational model confirms that activation of HCN channels can override a preference for centripetal summation expected from cell anatomy. This type of direction selectivity could play a role in coding movement similar to the axial selectivity seen in locust ganglion cells which detect looming stimuli. More generally, these results suggest that non-directional retinal ganglion cells can discriminate between input sequences independent of the retina network.

Visual information is coded by the output of retinal ganglion cells. Through evolution retinal ganglion cells acquired unique properties that allowed them to transmit to the brain such signals as direction of movement. The quest for the cellular mechanism of the detection of movement by retinal ganglion cells has been the holy grail of research on direction selectivity. In this study we have found a mechanism that allows individual non-direction selective On retinal ganglion cells to code sequences of excitatory inputs moving either away or toward the soma. We observed that inputs moving away from the soma resulted in enhanced, supralinear EPSP summation. Evidence from computational modeling suggests that expression of a specific set of voltage-dependent channels in dendrites introduces nonlinearities that could give a ganglion cell the ability to code looming motion. We predict that in a retinal network, such a directional tuning mechanism, together with asymmetric presynaptic inhibition, could be the building block for the development of more complex detection of visual motion.

The spatial structure of a nonlinear receptive field

Understanding a sensory system implies the ability to predict responses to a variety of inputs from a common model. In the retina, this includes predicting how the integration of signals across visual space shapes the outputs of retinal ganglion cells. Existing models of this process generalize poorly to predict responses to new stimuli. This failure arises in part from properties of the ganglion cell response that are not well captured by standard receptive-field mapping techniques: nonlinear spatial integration and fine-scale heterogeneities in spatial sampling. Here we characterize a ganglion cell's spatial receptive field using a mechanistic model based on measurements of the physiological properties and connectivity of only the primary excitatory circuitry of the retina. The resulting simplified circuit model successfully predicts ganglion-cell responses to a variety of spatial patterns and thus provides a direct correspondence between circuit connectivity and retinal output.

A continuum model of the retinal network and its response to electrical stimulation

A continuum network model of the retina is presented, consisting of an active implementation of the retinal ganglion cell tissue layer and passive implementation of deeper cell layers. The retinal ganglion cell layer receives excitatory presynaptic inputs from the bipolar layer and inhibitory presynaptic inputs from the amacrine layer. Simulations were performed to investigate the behavior of retinal tissue activation with epiretinal and suprachoroidal electrode stimulation. The results indicated the presence of both early and late onset action potentials consistent with experimental findings.

Ectopic retinal ON bipolar cell synapses in the OFF inner plexiform layer: contacts with dopaminergic amacrine cells and melanopsin ganglion cells

A key principle of retinal organization is that distinct ON and OFF channels are relayed by separate populations of bipolar cells to different sublaminae of the inner plexiform layer (IPL). ON bipolar cell axons have been thought to synapse exclusively in the inner IPL (the ON sublamina) onto dendrites of ON-type amacrine and ganglion cells. However, M1 melanopsin-expressing ganglion cells and dopaminergic amacrine (DA) cells apparently violate this dogma. Both are driven by ON bipolar cells, but their dendrites stratify in the outermost IPL, within the OFF sublamina. Here, in the mouse retina, we show that some ON cone bipolar cells make ribbon synapses in the outermost OFF sublayer, where they costratify with and contact the dendrites of M1 and DA cells. Whole-cell recording and dye filling in retinal slices indicate that type 6 ON cone bipolars provide some of this ectopic ON channel input. Imaging studies in dissociated bipolar cells show that these ectopic ribbon synapses are capable of vesicular release. There is thus an accessory ON sublayer in the outer IPL.

Receptive fields and functional architecture in the retina

Functional architecture of the striate cortex is known mostly at the tissue level--how neurons of different function distribute across its depth and surface on a scale of millimetres. But explanations for its design--why it is just so--need to be addressed at the synaptic level, a much finer scale where the basic description is still lacking. Functional architecture of the retina is known from the scale of millimetres down to nanometres, so we have sought explanations for various aspects of its design. Here we review several aspects of the retina's functional architecture and find that all seem governed by a single principle: represent the most information for the least cost in space and energy. Specifically: (i) why are OFF ganglion cells more numerous than ON cells? Because natural scenes contain more negative than positive contrasts, and the retina matches its neural resources to represent them equally well; (ii) why do ganglion cells of a given type overlap their dendrites to achieve 3-fold coverage? Because this maximizes total information represented by the array--balancing signal-to-noise improvement against increased redundancy; (iii) why do ganglion cells form multiple arrays? Because this allows most information to be sent at lower rates, decreasing the space and energy costs for sending a given amount of information. This broad principle, operating at higher levels, probably contributes to the brain's immense computational efficiency.

The spatial distribution of glutamatergic inputs to dendrites of retinal ganglion cells

The spatial pattern of excitatory glutamatergic input was visualized in a large series of ganglion cells of the rabbit retina, by using particle-mediated gene transfer of an expression plasmid for postsynaptic density 95-green fluorescent protein (PSD95-GFP). PSD95-GFP was confirmed as a marker of excitatory input by co-localization with synaptic ribbons (RIBEYE and kinesin II) and glutamate receptor subunits. Despite wide variation in the size, morphology, and functional complexity of the cells, the distribution of excitatory synaptic inputs followed a single set of rules: 1) the linear density of synaptic inputs (PSD95 sites/linear mum) varied surprisingly little and showed little specialization within the arbor; 2) the total density of excitatory inputs across individual arbors peaked in a ring-shaped region surrounding the soma, which is in accord with high-resolution maps of receptive field sensitivity in the rabbit; and 3) the areal density scaled inversely with the total area of the dendritic arbor, so that narrow dendritic arbors receive more synapses per unit area than large ones. To achieve sensitivity comparable to that of large cells, those that report upon a small region of visual space may need to receive a denser synaptic input from within that space.

Design of a neuronal array

Retinal ganglion cells of a given type overlap their dendritic fields such that every point in space is covered by three to four cells. We investigated what function is served by such extensive overlap. Recording from pairs of ON or OFF brisk-transient ganglion cells at photopic intensities, we confirmed that this overlap causes the Gaussian receptive field centers to be spaced at approximately 2 SDs (sigma). This, together with response nonlinearities and variability, was just sufficient to provide an ideal observer with uniform contrast sensitivity across the retina for both threshold and suprathreshold stimuli. We hypothesized that overlap might maximize the information represented from natural images, thereby optimizing retinal performance for many tasks. Indeed, tested with natural images (which contain statistical correlations), a model ganglion cell array maximized information represented in its population responses with approximately 2sigma spacing, i.e., the overlap observed in the retina. Yet, tested with white noise (which lacks statistical correlations), an array maximized its information by minimizing overlap. In both cases, optimal overlap balanced greater signal-to-noise ratio (from larger receptive fields) against greater redundancy (because of larger receptive field overlap). Thus, dendritic overlap improves vision by taking optimal advantage of the statistical correlations of natural scenes.

Different types of ganglion cell share a synaptic pattern

Retinal ganglion cells comprise about 10 morphological types that also differ functionally. To determine whether functional differences might arise partially from differences in excitatory input, we quantified the distributions of ribbon contacts to four mammalian ganglion cell types [brisk-transient (BT), brisk-sustained (BS), local edge (LE), directionally selective (DS)], comparing small vs. large and "sluggish" vs. "brisk." Cells in guinea pig retina were filled with fluorescent dye, immunostained for synaptic ribbons, and reconstructed with their ribbon contacts by confocal microscopy. False-positive contacts were corrected by performing the same analysis on processes that lack synapses: glial stalks and rod bipolar axons. All types shared a domed distribution of membrane that was well fit by a Gaussian function (R(2) = 0.96 +/- 0.01); they also shared a constant density of contacts on the dendritic membrane, both across each arbor and across cell types (19 +/- 1 contacts/100 microm(2) membrane). However, the distributions of membrane across the retina differed markedly in width (BT > DS approximately BS > LE) and peak density (BS > DS > LE > BT). Correspondingly, types differed in peak density of contacts (BS > DS approximately LE > BT) and total number (BS approximately BT > DS > LE). These differences between cell types in spatial extent and local concentration of membrane and synapses help to explain certain functional differences.

Developmental patterning of glutamatergic synapses onto retinal ganglion cells

Background

Neurons receive excitatory synaptic inputs that are distributed across their dendritic arbors at densities and with spatial patterns that influence their output. How specific synaptic distributions are attained during development is not well understood. The distribution of glutamatergic inputs across the dendritic arbors of mammalian retinal ganglion cells (RGCs) has long been correlated to the spatial receptive field profiles of these neurons. Thus, determining how glutamatergic inputs are patterned onto RGC dendritic arbors during development could provide insight into the cellular mechanisms that shape their functional receptive fields.

Results

We transfected developing and mature mouse RGCs with plasmids encoding fluorescent proteins that label their dendrites and glutamatergic postsynaptic sites. We found that as dendritic density (dendritic length per unit area of dendritic field) decreases with maturation, the density of synapses along the dendrites increases. These changes appear coordinated such that RGCs attain the mature average density of postsynaptic sites per unit area (areal density) by the time synaptic function emerges. Furthermore, stereotypic centro-peripheral gradients in the areal density of synapses across the arbor of RGCs are established at an early developmental stage.

Conclusion

The spatial pattern of glutamatergic inputs onto RGCs arises early in synaptogenesis despite ensuing reorganization of dendritic structure. We raise the possibility that these early patterns of synaptic distributions may arise from constraints placed on the number of contacts presynaptic neurons are able to make with the RGCs.

How robust is a neural circuit?

Design in engineering begins with the problem of robustness-by what factor should intrinsic capacity exceed normal demand? Here we consider robustness for a neural circuit that crosses the retina from cones to ganglion cells. The circuit's task is to represent the visual scene at many successive stages, each time by modulating a stream of stochastic events: photoisomerizations, then transmitter quanta, then spikes. At early stages, the event rates are high to achieve some critical signal-to-noise ratio and temporal bandwidth, which together set the information rate. Then neural circuits concentrate the information and repackage it, so that nearly the same total information can be represented by modulating far lower event rates. This is important for spiking because of its high metabolic cost. Considering various measurements at the outer and inner retina, we conclude that the "safety factors" are about 2-10, similar to other tissues.

Microcircuitry for two types of achromatic ganglion cell in primate fovea

Synaptic circuits in primate fovea have been quantified for midget/parvocellular ganglion cells. Here, based on partial reconstructions from serial electron micrographs, we quantify synaptic circuits for two other types of ganglion cell: the familiar parasol/magnocellular cell and a smaller type, termed "garland." The excitatory circuits both derive from two types of OFF diffuse cone bipolar cell, DB3 and DB2, which collected unselectively from at least 6 +/- 1 cones, including the S type. Cone contacts to DB3 dendrites were usually located between neighboring triads, whereas half of the cone contacts to DB2 were triad associated. Ribbon outputs were as follows: DB3, 69 +/- 5; DB2, 48 +/- 4. A complete parasol cell (30 microm dendritic field diameter) would collect from approximately 50 cones via approximately 120 bipolar and approximately 85 amacrine contacts; a complete garland cell (25 microm dendritic field) would collect from approximately 40 cones via approximately 75 bipolar and approximately 145 amacrine contacts. The bipolar types contributed differently: the parasol cell received most contacts (60%) from DB3, whereas the garland cell received most contacts (67%) from DB2. We hypothesize that DB3 is a transient bipolar cell and that DB2 is sustained. This would be consistent with their relative inputs to the brisk-transient (parasol) ganglion cell. The garland cell, with its high proportion of DB2 inputs plus its high proportion of amacrine synapses (70%) and dense mosaic, might correspond to the local-edge cell in nonprimate retinas, which serves finer acuity at low temporal frequencies. The convergence of S cones onto both types could contribute S-cone input for cortical areas primary visual cortex and the middle temporal area.

Axons and dendrites originate from neuroepithelial-like processes of retinal bipolar cells

The cellular mechanisms underlying axogenesis and dendritogenesis are not completely understood. The axons and dendrites of retinal bipolar cells, which contact their synaptic partners within specific laminae in the inner and outer retina, provide a good system for exploring these issues. Using transgenic mice expressing enhanced green fluorescent protein (GFP) in a subset of bipolar cells, we determined that axonal and dendritic arbors of these interneurons develop directly from apical and basal processes attached to the outer and inner limiting membranes, respectively. Selective stabilization of processes contributed to stratification of axonal and dendritic arbors within the appropriate synaptic layer. This unusual mode of axogenesis and dendritogenesis from neuroepithelial-like processes may act to preserve neighbor-neighbor relationships in synaptic wiring between the outer and inner retina.

Parallel Circuits from Cones to the On-Beta Ganglion Cell

Neural integration depends critically upon circuit architecture; yet the architecture has never been established quantitatively (numbers of cells and synapses) for any vertebrate local circuit. Here we describe circuits in the cat retina that connect cones to the on-beta ganglion cell. This cell type is important because on- and off-beta cells contribute about 50% of the optic nerve fibres and the major retinal input to the striate cortex. Three adjacent on-beta cells in the area centralis and their bipolar connections to cones were reconstructed from electron micrographs of 279 serial sections. The beta dendritic field is 34+/-2 microm in diameter and encompasses 35 cones. All of these cones connect to the beta cell via 14 - 17 bipolar cells. These bipolar cells were shown previously by cluster analysis to be of four types (b1 - b4); three of these types (b1, b2 and b3) provided 97% of the bipolar contacts to the beta cell, in the ratio 4:2:1. On average, bipolar cells nearest the centre of the beta dendritic field contribute more synapses than those towards the edge, but the peaked distribution of bipolar synapses across the dendritic field is only slightly broader than the optical pointspread function of the cat's eye, and is narrower by half than the centre of the ganglion cell receptive field. This implies that the distribution of bipolar synapses across the beta cell dendritic field contributes little to the extent or shape of the receptive field. Since all three bipolar circuits connect to the same set of cones, they must carry the same spatial and chromatic information; they might convey different temporal frequencies. The numbers of bipolar synapses (mean +/- SD=154+/-8) and amacrine synapses (59 +/- 5) converging on three adjacent beta cellsare remarkably constant (SD approximately +/-5% of the mean). Thus, as the circuits repeat locally, the fundamental design is accurately reproduced.

The receptive fields of cat retinal ganglion cells in physiological and pathological states: where we are after half a century of research

Studies on the receptive field properties of cat retinal ganglion cells over the past half-century are reviewed within the context of the role played by the receptive field in visual information processing. Emphasis is placed on the work conducted within the past 20 years, but a summary of key contributions from the 1950s to 1970s is provided. We have sought to review aspects of the ganglion cell receptive field that have not been featured prominently in previous review articles. Our review of the receptive field properties of X- and Y-cells focuses on quantitative studies and includes consideration of the function of the receptive field in visual signal processing. We discuss the non-classical as well as the classical receptive field. Attention is also given to the receptive field properties of the less well-studied cat ganglion cells-the W-cells-and the effect of pathology on cat ganglion cell properties. Although work from our laboratories is highlighted, we hope that we have given a reasonably balanced view of the current state of the field.

Bipolar cells contribute to nonlinear spatial summation in the brisk-transient (Y) ganglion cell in mammalian retina

The receptive field of the Y-ganglion cell comprises two excitatory mechanisms: one integrates linearly over a narrow field, and the other integrates nonlinearly over a wide field. The linear mechanism has been attributed to input from bipolar cells, and the nonlinear mechanism has been attributed to input from a class of amacrine cells whose nonlinear "subunits" extend across the linear receptive field and beyond. However, the central component of the nonlinear mechanism could in theory be driven by bipolar input if that input were rectified. Recording intracellularly from the Y-cell in guinea pig retina, we blocked the peripheral component of the nonlinear mechanism with tetrodotoxin and found the remaining nonlinear receptive field to be precisely co-spatial with the central component of the linear receptive field. Both linear and nonlinear mechanisms were caused by an excitatory postsynaptic potential that reversed near 0 mV. The nonlinear mechanism depended neither on acetylcholine nor on feedback involving GABA or glycine. Thus the central components of the ganglion cell's linear and nonlinear mechanisms are apparently driven by synapses from the same rectifying bipolar cell.

Bipolar cells contribute to nonlinear spatial summation in the brisk-transient (Y) ganglion cell in mammalian retina

The receptive field of the Y-ganglion cell comprises two excitatory mechanisms: one integrates linearly over a narrow field, and the other integrates nonlinearly over a wide field. The linear mechanism has been attributed to input from bipolar cells, and the nonlinear mechanism has been attributed to input from a class of amacrine cells whose nonlinear "subunits" extend across the linear receptive field and beyond. However, the central component of the nonlinear mechanism could in theory be driven by bipolar input if that input were rectified. Recording intracellularly from the Y-cell in guinea pig retina, we blocked the peripheral component of the nonlinear mechanism with tetrodotoxin and found the remaining nonlinear receptive field to be precisely co-spatial with the central component of the linear receptive field. Both linear and nonlinear mechanisms were caused by an excitatory postsynaptic potential that reversed near 0 mV. The nonlinear mechanism depended neither on acetylcholine nor on feedback involving GABA or glycine. Thus the central components of the ganglion cell's linear and nonlinear mechanisms are apparently driven by synapses from the same rectifying bipolar cell.

Visual processing: the devil is in the details

Ganglion cells convey information from the retina back to the brain. Recent experiments have examined how ganglion cell receptive fields are assembled from many incoming signals.

Confronting complexity: strategies for understanding the microcircuitry of the retina

The mammalian retina contains upward of 50 distinct functional elements, each carrying out a specific task. Such diversity is not rare in the central nervous system, but the retina is privileged because its physical location, the distinctive morphology of its neurons, the regularity of its architecture, and the accessibility of its inputs and outputs permit a unique variety of experiments. Recent strategies for confronting the retina's complexity attempt to marry genetic approaches to new kinds of anatomical and electrophysiological techniques.

Receptive field microstructure and dendritic geometry of retinal ganglion cells

We studied the fine spatial structure of the receptive fields of retinal ganglion cells and its relationship to the dendritic geometry of these cells. Cells from which recordings had been made were microinjected with Lucifer yellow, so that responses generated at precise locations within the receptive field center could be directly compared with that cell's dendritic structure. While many cells with small receptive fields had domeshaped sensitivity profiles, the majority of large receptive fields were composed of multiple regions of high sensitivity. The density of dendritic branches at any one location did not predict the regions of high sensitivity. Instead, the interactions between a ganglion cell's dendritic tree and the local mosaic of bipolar cell axons seem to define the fine structure of the receptive field center.

Parallel cone bipolar pathways to a ganglion cell use different rates and amplitudes of quantal excitation

The cone signal reaches the cat's On-beta (X) ganglion cell via several parallel circuits (bipolar cell types b1, b2, and b3). These circuits might convey different regions of the cone's temporal bandwidth. To test this, I presented a step of light that elicited a transient depolarization followed by a sustained depolarization. The contribution of bipolar cells to these response components was isolated by blocking action potentials with tetrodotoxin and by blocking inhibitory synaptic potentials with bicuculline and strychnine. Stationary fluctuation analysis of the sustained depolarization gave the rate of quantal bombardment: approximately 5100 quanta sec(-1) for small central cells and approximately 45,000 quanta sec(-1) for large peripheral cells. Normalizing these rates for the vastly different numbers of bipolar synapses (150-370 per small cell vs 2000 per large cell), quantal rate was constant across the retina, approximately 22 quanta synapse(-1) sec(-1). Nonstationary fluctuation analysis gave the mean quantal EPSP amplitude: approximately 240 microV for the transient depolarization and 30 microV for the sustained depolarization. The b1 bipolar cell is known from noise analysis of the On-alpha ganglion cell to have a near-maximal sustained release of only approximately two quanta synapse(-1) sec(-1). This implies that the other bipolar types (b2 and b3) contribute many more quanta to the sustained depolarization (>/=46 synapse(-1) sec(-1)). Type b1 probably contributes large quanta to the transient depolarization. Thus, bipolar cell types b1 and b2/b3 apparently constitute parallel circuits that convey, respectively, high and low frequencies.

Parallel cone bipolar pathways to a ganglion cell use different rates and amplitudes of quantal excitation

The cone signal reaches the cat's On-beta (X) ganglion cell via several parallel circuits (bipolar cell types b1, b2, and b3). These circuits might convey different regions of the cone's temporal bandwidth. To test this, I presented a step of light that elicited a transient depolarization followed by a sustained depolarization. The contribution of bipolar cells to these response components was isolated by blocking action potentials with tetrodotoxin and by blocking inhibitory synaptic potentials with bicuculline and strychnine. Stationary fluctuation analysis of the sustained depolarization gave the rate of quantal bombardment: approximately 5100 quanta sec(-1) for small central cells and approximately 45,000 quanta sec(-1) for large peripheral cells. Normalizing these rates for the vastly different numbers of bipolar synapses (150-370 per small cell vs 2000 per large cell), quantal rate was constant across the retina, approximately 22 quanta synapse(-1) sec(-1). Nonstationary fluctuation analysis gave the mean quantal EPSP amplitude: approximately 240 microV for the transient depolarization and 30 microV for the sustained depolarization. The b1 bipolar cell is known from noise analysis of the On-alpha ganglion cell to have a near-maximal sustained release of only approximately two quanta synapse(-1) sec(-1). This implies that the other bipolar types (b2 and b3) contribute many more quanta to the sustained depolarization (>/=46 synapse(-1) sec(-1)). Type b1 probably contributes large quanta to the transient depolarization. Thus, bipolar cell types b1 and b2/b3 apparently constitute parallel circuits that convey, respectively, high and low frequencies.

Rate of quantal excitation to a retinal ganglion cell evoked by sensory input

To determine the rate and statistics of light-evoked transmitter release from bipolar synapses, intracellular recordings were made from ON-alpha ganglion cells in the periphery of the intact, superfused, cat retina. Sodium channels were blocked with tetrodotoxin to prevent action potentials. A light bar covering the receptive field center excited the bipolar cells that contact the alpha cell and evoked a transient then a sustained depolarization. The sustained depolarization was quantified as change in mean voltage (Deltav), and the increase in voltage noise that accompanied it was quantified as change in voltage variance (Deltasigma(2)). As light intensity increased, Deltav and Deltasigma(2) both increased, but their ratio held constant. This behavior is consistent with Poisson arrival of transmitter quanta at the ganglion cell. The response component attributable to glutamate quanta from bipolar synapses was isolated by application of 6-cyano-7-nitroquinoxaline (CNQX). As CNQX concentration increased, the signal/noise ratio of this response component (Deltav(CNQX)/Deltasigma(CNQX)) held constant. This is also consistent with Poisson arrival and justified the application of fluctuation analysis. Two different methods of fluctuation analysis applied to Deltav(CNQX) and Deltasigma(CNQX) produced similar results, leading to an estimate that a just-maximal sustained response was caused by approximately 3,700 quanta s(-1). The transient response was caused by a rate that was no more than 10-fold greater. Because the ON-alpha cell at this retinal locus has approximately 2,200 bipolar synapses, one synapse released approximately 1.7 quanta s(-1) for the sustained response and no more than 17 quanta s(-1) for the transient. Consequently, within the ganglion cell's integration interval, here calculated to be approximately 16 ms, a bipolar synapse rarely releases more than one quantum. Thus for just-maximal sustained and transient depolarizations, the conductance modulated by a single bipolar cell synapse is limited to the quantal conductance ( approximately 100 pS at its peak). This helps preserve linear summation of quanta. The Deltav/Deltasigma(2) ratio remained constant even as the ganglion cell's response saturated, which suggested that even at the peak of sensory input, summation remains linear, and that saturation occurs before the bipolar synapse.

Deciphering the retina's wiring diagram

By simultaneously recording from retinal ganglion cells while stimulating a single cone, Chichilnisky and Baylor demonstrate that the strength of physiological connections within a retinal microcircuit is linearly proportional to the number of anatomically defined synapses.

Receptive-field microstructure of blue-yellow ganglion cells in primate retina

We examined the functional microcircuitry of cone inputs to blue-ON/yellow-OFF (BY) ganglion cells in the macaque retina using multielectrode recording. BY cells were identified by their ON responses to blue light and OFF responses to red or green light. Cone-isolating stimulation indicated that ON responses originated in short (S) wavelength-sensitive cones, whereas OFF responses originated in both long (L) and middle (M) wavelength-sensitive cones. Stimulation with fine spatial patterns revealed locations of individual S cones in BY cell receptive fields. Neighboring BY cells received common but unequal inputs from one or more S cones. Inputs from individual S cones differed in strength, indicating different synaptic weights, and summed approximately linearly to control BY cell firing.

A comparison of receptive-field and tracer-coupling size of amacrine and ganglion cells in the rabbit retina

Recent studies have shown that amacrine and ganglion cells in the mammalian retina are extensively coupled as revealed by the intercellular movement of the biotinylated tracers biocytin and Neurobiotin. These demonstrations of tracer coupling suggest that electrical networks formed by proximal neurons (i.e. amacrine and ganglion cells) may underlie the lateral propagation of signals across the inner retina. We studied this question by comparing the receptive-field size, dendritic-field size, and extent of tracer coupling of amacrine and ganglion cells in the dark-adapted, superfused, isolated retina eyecup of the rabbit. Our results indicate that while the center-receptive fields of proximal neurons are approximately 15% larger than their corresponding dendritic diameters, this slight difference can be explained by factors other than electrical coupling such as tissue shrinkage associated with histological processing. However, the extent of tracer coupling of amacrine and ganglion cells was, on average, about twice the size of the corresponding receptive fields. Thus, the receptive field of an individual proximal neuron matched far more closely to its dendritic diameter than to the size of the tracer-coupled network of cells to which it belonged. The exception to this rule was the AII amacrine cells for which center-receptive fields were 2-3 times the size of their dendritic diameters but matched closely to the size of the tracer-coupled arrays. Thus, with the exception of AII cells, our data indicate that tracer coupling between proximal neurons is not associated with an enlargement of their receptive fields. Our results, then, provide no evidence for electrical coupling or, at least, indicate that extensive lateral spread of visual signals does not occur in the proximal mammalian retina.

Parallel cone bipolar to on-beta ganglion cell pathways in the cat retina: spatial responses, spatial aliasing, and spatial variance

An important issue in understanding the retina is finding candidate functional roles for different cell pathways and the details of their anatomy and physiology. We consider various spatial properties of the three main cone ==> cone bipolar cell ==> on-beta ganglion cell pathways in the cat retina and possible roles for the particulars of their anatomy. The cone bipolar cells in these pathways have distinct morphologies and modest differences in their convergence, divergence, densities, and synaptic weighting; and it is unclear whether the pathways differ in their spatial properties or in some other manner. Since differences in spatial processing of cells are best studied on a systemwide level, we developed the multirate filter-based method of retinal modeling, a technique for relating the anatomy of multiple cell layers to its systemic effects. We demonstrate that (1) despite the anatomic distinctions among the three main cone bipolar cell pathways, their spatial responses are essentially identical; (2) despite the spatial averaging in the pathways, there is essentially no filtering of the nonaliasing signal components after the cone layer; (3) instead, this averaging combined with prefiltering by the eye's optics and cone gap junctions prevents spatial aliasing; and (4) the averaging and prefiltering combined allow cell responses to be similar despite significant cell-to-cell anatomic differences.

Conductances evoked by light in the ON-beta ganglion cell of cat retina

When a bar of light (215 x 5000 microns) illuminates the receptive field of an ON-beta ganglion cell of cat retina, the cell depolarizes. Intracellular recording from the cat eyecup preparation shows that this depolarization is due to an increase in conductance (2.4 +/- 0.6 nS). Different phases of this depolarization have different reversal potentials, but all of these reversal potentials are more positive than the cell's resting potential in the dark. When the light is turned on, there is an initial transient depolarization; the reversal potential measured for this transient is positive (23 +/- 11 mV). As the light is left on, the cell partially repolarizes to a sustained depolarization; the reversal potential measured for this sustained depolarization is close to zero (-1 +/- 5 mV). When the light is turned off, the cell repolarizes further; the reversal potential measured for this repolarization is negative (-18 +/- 7 mV), but still above the resting potential in the dark (-50 mV). To explain this variety of reversal potentials, at least two different synaptic conductances are required: one to ions which have a positive reversal potential and another to ions which have a negative reversal potential. Comparing the responses to broad and narrow bars suggests that these two conductances are associated with the center and surround, respectively. Finally, since an ON-beta cell in the area centralis receives about 200 synapses, these results indicate that a single synapse produces an average conductance increase of about 15 pS during a near-maximal depolarization.

Signal sampling and propagation through multiple cell layers in the retina: modeling and analysis with multirate filtering

The retina is a multilayered structure. Each layer consists of one or more classes of cell, each at its own density and with its own anatomic and physiologic properties. Signals converge from many cells in one layer onto single cells in another layer, and a signal from a single cell diverges to many cells in the next layer. In this methods paper we develop a general approach to retinal analysis and modeling that incorporates multiple cell classes, their densities, and related anatomic properties. The method is based on multirate filtering, a branch of signal processing in which signals of different sampling rates are manipulated. By drawing a correspondence between cell density and signal sampling rate, we define multirate models that incorporate different cell densities, convergence, divergence, variation in dendritic field shape, cell-to-cell variation in synaptic weights, and other anatomic features. We develop the multirate approach and apply it to the cat cone-->cone bipolar CBb1-->on-beta ganglion cell pathway as an example. We calculate the spatial frequency responses of the CBb1 and on-beta cells based on the cone spatial frequency response and find that the attenuation of high frequencies in the cones prevents aliasing that would otherwise occur in CBb1 and on-beta cells. We compare the calculations with cat psychophysics. We show that the optics of the cat eye are insufficient in themselves for the prevention of aliasing in these cells; additional attenuation by the cone-cone gap junctions and the cone aperture is necessary. By including this postreceptoral filtering, we demonstrate that the highest spatial frequency that can be passed by the retina without aliasing is determined not always only by the densities of cones, bipolar cells, and ganglion cells but also by the synaptic and the dendritic weighting between these cells.

OFF-alpha and OFF-beta ganglion cells in cat retina. I: Intracellular electrophysiology and HRP stains

Six OFF-alpha ganglion cells and a single OFF-beta ganglion cell were penetrated with intracellular microelectrodes and marked with horseradish peroxidase (HRP) in a perfused cat eyecup. Gaussian center radii (Rc) ranging from 40 to 217 microns were measured for receptive fields mapped with slits, values in agreement with previous extracellular reports. ON and OFF response components revealed nearly identical Rc's and center locations. Although Gaussian diameters (2Rc) were about 80% of dendritic field diameters overall, in this sample dendritic and receptive fields were not well correlated. Spatial tuning of ganglion cells was evidenced in peaked amplitude-vs.-width functions, fit by difference-of-Gaussians models. Such plots yielded Rc values about 40% less than position-vs amplitude plots. Rs values for surrounds ranged from 200 to 1,700 microns. Rod and cone signals were investigated with flicker. Rod flicker signals in OFF-alpha cells were larger and of shorter latency than in either horizontal or AII amacrine cells. Cone flicker signals were also short in latency, with an ON response time constant of 9 msec, and an OFF response time constant of 3 msec. The OFF-alpha rod-cone transition involved a latency increase of 20-30 msec. The spontaneous and light-evoked impulse rates of OFF-alpha responses varied linearly with extrinsic current, but the amplitude of ON hyperpolarization was little affected. After injection of staining current, the OFF-beta cell transiently depolarized at ON, suggestive of ON inhibition with reversed chloride gradient, a result not seen in OFF-alpha responses. Events (peaked, depolarizing voltage fluctuations) of high, low, and intermediate amplitudes were studied in OFF-alpha responses. High amplitude events (impulses), were OFF-correlated with the stimulus, and exhibited mean rise times (transit time from 25 to 75% of peak amplitude) from 255 to 392 microseconds. Intermediate level events (presumed synaptic origin) were also OFF correlated and had longer rise times (325 microseconds to 1.56 microseconds). Low level events (234-685 microseconds) revealed either ON, ON/OFF, or not stimulus correlation.

OFF-alpha and OFF-beta ganglion cells in cat retina: II. Neural circuitry as revealed by electron microscopy of HRP stains

An OFF-center alpha and an OFF-center beta ganglion cell in cat retina, which had been recorded from and intracellularly stained with horseradish peroxidase (HRP) were examined by serial section electron microscopy. We counted synapses and identified presynaptic neurons to the HRP-stained cells in 20 microns radial slices through the centers of their dendritic trees. Presynaptic amacrine and bipolar cells were identified on cytological criteria known from previous studies. The OFF-beta cell with a 62 microns dendritic arbor, restricted to S1 and S2 (sublamina a) of the inner plexiform layer (IPL), received 38% bipolar and 62% amacrine cell synapses. The bipolar input was from both cb1 and cb2 cone bipolar types. Input from three distinct amacrine cell types occurred upon the dendrites, namely from: (1) AII amacrine lobular appendages, (2) large pale amacrine profiles (possibly A2 or A3 cells), and (3) small, dark amacrine types (possibly A8 cells). Large pale amacrine profiles (possibly A13) were found on the cell body and apical dendrite in sublamina b of the IPL. In addition, several amacrine profiles synapsed directly on the sides and base of the cell body in the ganglion cell layer. We estimate that the complete dendritic tree of this beta cell received about 1,000 synapses contributed by 12-14 bipolar cells, 7-10 AII amacrines and 28-41 other amacrine cells. The OFF-alpha cell had a dendritic tree size of 680 x 920 microns. A 250 microns length of two major dendrites stratifying narrowly in S2 of the IPL was reconstructed. Amacrine cells provided most of the synaptic input (80%). This input came from: (1) AII amacrine lobular appendages, (2) amacrines exhibiting large, pale synaptic profiles (possibly A2 or A3 cells), (3) pale amacrines with large mitochondria and a few neurotubules (unknown type), and (4) densely neurotubule-filled amacrine profiles (possibly A19 cells). A large pale amacrine cell type (possibly A13) provided synaptic input to the cell body as a serial synaptic intermediary with rod bipolar cells. Cone bipolar synapses were from only one type of cone bipolar, the cb2 type and formed 20% of the total synaptic input. We estimate that a minimum of 142 bipolar cells, 256 AII amacrine cells and 1,011 other amacrine cells, altogether providing 6,000-10,000 synapses, converged on the dendritic tree of this OFF-alpha cell.