Heterogeneity of retinogeniculate axon arbors

The retinogeniculate synapse transmits information from retinal ganglion cells (RGC) in the eye to thalamocortical relay neurons in the visual thalamus, the dorsal lateral geniculate nucleus (dLGN). Studies in mice have identified genetic markers for distinct classes of RGCs encoding different features of the visual space, facilitating the dissection of RGC subtype‐specific physiology and anatomy. In this study, we examine the morphological properties of axon arbors of the BD‐RGC class of ON‐OFF direction selective cells that, by definition, exhibit a stereotypic dendritic arbor and termination pattern in the retina. We find that axon arbors from the same class of RGCs exhibit variations in their structure based on their target region of the dLGN. Our findings suggest that target regions may influence the morphologic and synaptic properties of their afferent inputs.     

Retinal ganglion cell dendrites undergo a visual activity-dependent redistribution after eye opening

Recent studies showed that light stimulation is required for the maturational segregation of retinal ganglion cell (RGC) synaptic connectivity with ON and OFF bipolar cells in mammalian retina. However, it is not clear to what extent light stimulation regulates the maturation of RGC dendritic ramification and synaptic connections. The present work quantitatively analyzed the dendritic ramification patterns of different morphological subtypes of RGCs of developing mouse retinas and demonstrated that RGCs in all four major morphological subtypes underwent profound dendritic redistributions from the center to specific stratum of the IPL after eye opening. Light deprivation preferentially blocked the developmental RGC dendritic redistribution from the center to sublamina a of the IPL. Interestingly, this developmental redistribution of RGC dendrites could not be explained by a simple developmental elimination of "excess" dendrites and, therefore, suggests a possible mechanism that requires both selective dendritic growth and elimination guided by visual activity.     

Age-related alterations in neurons of the mouse retina

The behavioral consequences of age-related alterations in neural function are well documented, but less is known about their cellular bases. To characterize such changes, we analyzed 14 molecularly identified subsets of mouse retinal projection neurons (retinal ganglion cells or RGCs) and interneurons (amacrine, bipolar, and horizontal cells). The retina thinned but expanded with age, maintaining its volume. There was minimal decline in the number of RGCs, interneurons, or photoreceptors, but the diameter of RGC dendritic arbors decreased with age. Together, the increased retinal area and the decreased dendritic area may lead to gaps in RGC coverage of the visual field. Axonal arbors of RGCs in the superior colliculus also atrophied with age, suggesting that the relay of visual information to central targets may decline over time. On the other hand, the laminar restriction of RGC dendrites and the interneuronal processes that synapse on them were not detectably disturbed, and RGC subtypes exhibited distinct electrophysiological responses to complex visual stimuli. Other neuronal types aged in different ways: amacrine cell arbors did not remodel detectably, whereas horizontal cell processes sprouted into the photoreceptor layer. Bipolar cells showed arbor-specific alterations: their dendrites sprouted but their axons remained stable. In summary, retinal neurons exhibited numerous age-related quantitative alterations (decreased areas of dendritic and axonal arbors and decreased density of cells and synapses), whereas their qualitative features (molecular identity, laminar specificity, and feature detection) were largely preserved. Together, these data reveal selective age-related alterations in neural circuitry, some of which could underlie declines in visual acuity.     

A general principle governs vision‐dependent dendritic patterning of retinal ganglion cells

Dendritic arbors of retinal ganglion cells (RGCs) collect information over a certain area of the visual scene. The coverage territory and the arbor density of dendrites determine what fraction of the visual field is sampled by a single cell and at what resolution. However, it is not clear whether visual stimulation is required for the establishment of branching patterns of RGCs, and whether a general principle directs the dendritic patterning of diverse RGCs. By analyzing the geometric structures of RGC dendrites, we found that dendritic arbors of RGCs underwent a substantial spatial rearrangement after eye‐opening. Light deprivation blocked both the dendritic growth and the branch patterning, suggesting that visual stimulation is required for the acquisition of specific branching patterns of RGCs. We further showed that vision‐dependent dendritic growth and arbor refinement occurred mainly in the middle portion of the dendritic tree. This nonproportional growth and selective refinement suggest that the late‐stage dendritic development of RGCs is not a passive stretching with the growth of eyes, but rather an active process of selective growth/elimination of dendritic arbors of RGCs driven by visual activity. Finally, our data showed that there was a power law relationship between the coverage territory and dendritic arbor density of RGCs on a cell‐by‐cell basis. RGCs were systematically less dense when they cover larger territories regardless of their cell type, retinal location, or developmental stage. These results suggest that a general structural design principle directs the vision‐dependent patterning of RGC dendrites. J. Comp. Neurol. 522:3403–3422, 2014. © 2014 Wiley Periodicals, Inc.     

Regulation of neonatal development of retinal ganglion cell dendrites by neurotrophin‐3 overexpression

The morphology of dendrites constrains and reflects the nature of synaptic inputs to neurons. The visual system has served as a useful model to show how visual function is determined by the arborization patterns of neuronal processes. In retina, light ON and light OFF responding ganglion cells selectively elaborate their dendritic arbors in distinct sublamina, where they receive, respectively, inputs from ON and OFF bipolar cells. During neonatal maturation, the bilaminarly distributed dendritic arbors of ON‐OFF retinal ganglion cells (RGCs) are refined to more narrowly localized monolaminar structures characteristic of ON or OFF RGCs. Recently, brain‐derived neurotrophic factor (BDNF) has been shown to regulate this laminar refinement, and to enhance the development of dendritic branches selectively of ON RGCs. Although other related neurotrophins are known to regulate neuronal process formation in the central nervous system, little is known about their action in maturing retina. Here, we report that overexpression of neurotrophin‐3 (NT‐3) in the eye accelerates RGC laminar refinement before eye opening. Furthermore, NT‐3 overexpression increases dendritic branch number but reduces dendritic elongation preferentially in ON‐OFF RGCs, a process that also occurs before eye opening. NT‐3 overexpression does affect dendritic maturation in ON RGCs, but to a much less degree. Taken together, our results suggest that NT‐3 and BDNF exhibit overlapping effects in laminar refinement but distinct RGC‐cell‐type specific effects in shaping dendritic arborization during postnatal development. J. Comp. Neurol. 514:449–458, 2009. © 2009 Wiley‐Liss, Inc.     

Overlapping morphological and functional properties between M4 and M5 intrinsically photosensitive retinal ganglion cells

Multiple retinal ganglion cell (RGC) types in the mouse retina mediate pattern vision by responding to specific features of the visual scene. The M4 and M5 melanopsin-expressing, intrinsically photosensitive retinal ganglion cell (ipRGC) subtypes are two RGC types that are thought to play major roles in pattern vision. The M4 ipRGCs overlap in population with ON-alpha RGCs, while M5 ipRGCs were recently reported to exhibit opponent responses to different wavelengths of light (color opponency). Despite their seemingly distinct roles in visual processing, previous reports have suggested that these two populations may exhibit overlap in their morphological and functional properties, which calls into question whether these are in fact distinct RGC types. Here, we show that M4 and M5 ipRGCs are distinct morphological classes of ipRGCs, but they cannot be exclusively differentiated based on color opponency and dendritic morphology as previously reported. Instead, we find that M4 and M5 ipRGCs can only be distinguished based on soma size and the number of dendritic branch points in combination with SMI-32 immunoreactivity. These results have important implications for clearly defining RGC types and their roles in visual behavior.     

Pathway-specific maturation, visual deprivation, and development of retinal pathway.

One of the fundamental features of the visual system is the segregation of neural circuits that process increments and decrements of luminance into ON and OFF pathways. In mature retina, the dendrites of retinal ganglion cells (RGCs) in the inner plexiform layer (IPL) of retina are separated into ON or OFF sublamina-specific stratification. At an early developmental stage, however, the dendrites of most RGCs are ramified throughout the IPL. The maturation of RGC ON/OFF dendritic stratification requires neural activities mediated by afferent inputs from bipolar and amacrine cells. The synchronized spontaneous burst activities in early postnatal developing retina regulate RGC dendritic filopodial movements and the maintenance or elimination of dendritic processes. After eye opening, visual experience further remodels and consolidates the retinal neural circuit into mature forms. Several neurotransmitter systems, including glutamatergic, acetylcholinergic, GABAergic, and glycinergic systems, might act together to modulate the RGC dendritic refinement. In addition, both the bipolar cells and cholinergic amacrine cells may provide laminar cues for the maturation of RGC dendritic stratification.     

Undersized dendritic arborizations in retinal ganglion cells of the rd1 mutant mouse: a paradigm of early onset photoreceptor degeneration

Retinitis pigmentosa (RP) is a family of inherited diseases causing progressive photoreceptor death. Retinal ganglion cells (RGCs) form the biological substrate for various therapeutic approaches designed to restore vision in RP individuals. Assessment of survival and preservation of RGCs in animal paradigms mimicking the human disease is of key importance for appropriate implementation of vision repair strategies. Here we studied the survival of RGCs in the rd1 mutant mouse, a known model of early onset, autosomic recessive RP, at various stages of photoreceptor degeneration. Furthermore, we analyzed the morphology of various types of RGCs using the newly generated transgenic mouse rd1/Thy1-GFP, in which the rd1 mutation is associated with green fluorescent protein (GFP) expression in a small population of different RGCs. We found excellent survival of cells at up to 1 year of age, a time at which the inner retina is known to have severely reorganized and partially degenerated. However, 50% of the cells analyzed within all RGC types exhibit an undersized dendritic tree, spanning about half of the normal area. Undersized cells are found both in adult and in very young (1-month-old) mice. This suggests that their aberrant phenotype is due to incomplete dendritic development, possibly as a consequence of altered visual input at the time of dendritic arbor refinement. These data show the importance of the timing of photoreceptor death in RGC dendritic development.     

Recording from defined populations of retinal ganglion cells using a high-density CMOS-integrated microelectrode array with real-time switchable electrode selection

In order to understand how retinal circuits encode visual scenes, the neural activity of defined populations of retinal ganglion cells (RGCs) has to be investigated. Here we report on a method for stimulating, detecting, and subsequently targeting defined populations of RGCs. The possibility to select a distinct population of RGCs for extracellular recording enables the design of experiments that can increase our understanding of how these neurons extract precise spatio-temporal features from the visual scene, and how the brain interprets retinal signals. We used light stimulation to elicit a response from physiologically distinct types of RGCs and then utilized the dynamic-configurability capabilities of a microelectronics-based high-density microelectrode array (MEA) to record their synchronous action potentials. The layout characteristics of the MEA made it possible to stimulate and record from multiple, highly overlapping RGCs simultaneously without light-induced artifacts. The high-density of electrodes and the high signal-to-noise ratio of the MEA circuitry allowed for recording of the activity of each RGC on 14±7 electrodes. The spatial features of the electrical activity of each RGC greatly facilitated spike sorting. We were thus able to localize, identify and record from defined RGCs within a region of mouse retina. In addition, we stimulated and recorded from genetically modified RGCs to demonstrate the applicability of optogenetic methods, which introduces an additional feature to target a defined cell type. The developed methodologies can likewise be applied to other neuronal preparations including brain slices or cultured neurons.     

Morphological identification and systematic classification of mammalian retinal ganglion cells. I. Rabbit retinal ganglion cells

Retinal ganglion cells (RGCs) convey visual signals to 50 regions of the brain. For reasons of interest and convenience, they constitute an excellent system for the study of brain structure and function. There is general agreement that, absent a complete “parts list,” understanding how the nervous system processes information will remain an elusive goal. Recent studies indicate that there are 30–50 types of ganglion cell in mouse retina, whereas only a few years ago it was still written that mice and the more visually oriented lagomorphs had less than 20 types of RGC. More than 30 years ago, I estimated that rabbits have about 40 types of RGC. The present study indicates that this number is much too low. I have employed the old but powerful method of Golgi-impregnation to rabbit retina, studying the range of component neurons in this already well-studied retinal system. Close quantitative and qualitative analyses of 1,142 RGCs in 26 retinas take into account cell body and dendritic field size, level(s) of dendritic stratification in the retina's inner plexiform layer, and details of dendritic branching. Ninety-one morphologies are recognized. Of these, at least 32 can be correlated with physiologically studied RGCs, dye-injected for morphological analysis. It is unlikely that rabbits have 91 types of RGC, but is argued here that this number lies between 60 and 70. The present study provides a “yardstick” for measuring the output of future molecular studies that may be more definitive in fixing the number of RGC types in rabbit retina.     

DSCAM differentially modulates pre- and postsynaptic structural and functional central connectivity during visual system wiring

Background

Proper patterning of dendritic and axonal arbors is a critical step in the formation of functional neuronal circuits. Developing circuits rely on an array of molecular cues to shape arbor morphology, but the underlying mechanisms guiding the structural formation and interconnectivity of pre- and postsynaptic arbors in real time remain unclear. Here we explore how Down syndrome cell adhesion molecule (DSCAM) differentially shapes the dendritic morphology of central neurons and their presynaptic retinal ganglion cell (RGC) axons in the developing vertebrate visual system.

Methods

The cell-autonomous role of DSCAM, in tectal neurons and in RGCs, was examined using targeted single-cell knockdown and overexpression approaches in developing Xenopus laevis tadpoles. Axonal arbors of RGCs and dendritic arbors of tectal neurons were visualized using real-time in vivo confocal microscopy imaging over the course of 3 days.

Results

In the Xenopus visual system, DSCAM immunoreactivity is present in RGCs, cells in the optic tectum and the tectal neuropil at the time retinotectal synaptic connections are made. Downregulating DSCAM in tectal neurons significantly increased dendritic growth and branching rates while inducing dendrites to take on tortuous paths. Overexpression of DSCAM, in contrast, reduced dendritic branching and growth rate. Functional deficits mediated by tectal DSCAM knockdown were examined using visually guided behavioral assays in swimming tadpoles, revealing irregular behavioral responses to visual stimulus. Functional deficits in visual behavior also corresponded with changes in VGLUT/VGAT expression, markers of excitatory and inhibitory transmission, in the tectum. Conversely, single-cell DSCAM knockdown in the retina revealed that RGC axon arborization at the target is influenced by DSCAM, where axons grew at a slower rate and remained relatively simple. In the retina, dendritic arbors of RGCs were not affected by the reduction of DSCAM expression.

Conclusions

Together, our observations implicate DSCAM in the control of both pre- and postsynaptic structural and functional connectivity in the developing retinotectal circuit, where it primarily acts as a neuronal brake to limit and guide postsynaptic dendrite growth of tectal neurons while it also facilitates arborization of presynaptic RGC axons cell autonomously.

     

Early neural activity and dendritic growth in turtle retinal ganglion cells

Early neural activity, both prenatal spontaneous bursts and early visual experience, is believed to be important for dendritic proliferation and for the maturation of neural circuitry in the developing retina. In this study, we have investigated the possible role of early neural activity in shaping developing turtle retinal ganglion cell (RGC) dendritic arbors. RGCs were back-labelled from the optic nerve with horseradish peroxidase (HRP). Changes in dendritic growth patterns were examined across development and following chronic blockade or modification of spontaneous activity and/or visual experience. Dendrites reach peak proliferation at embryonic stage 25 (S25, one week before hatching), followed by pruning in large field RGCs around the time of hatching. When spontaneous activity is chronically blocked in vivo from early embryonic stages (S22) with curare, a cholinergic nicotinic antagonist, RGC dendritic growth is inhibited. On the other hand, enhancement of spontaneous activity by dark-rearing (Sernagor & Grzywacz (1996)Curr. Biol., 6, 1503-1508) promotes dendritic proliferation in large-field RGCs, an effect that is counteracted by exposure to curare from hatching. We also recorded spontaneous activity from individual RGCs labelled with lucifer yellow (LY). We found a tendency of RGCs with large dendritic fields to be spontaneously more active than small-field cells. From all these observations, we conclude that immature spontaneous activity promotes dendritic growth in developing RGCs.     

Migration of phagocytotic cells and development of the murine intraretinal microglial network: an in vivo study using fluorescent dyes

This work was undertaken to study whether retinal ganglion cell (RGC) death, which occurs during postnatal development of the mouse retina could aid in assessing the topological and chronological pattern of microglial cell migration. The study was conducted from postnatal day 0 (P0) to adulthood. The fluorescent dyes Fluorogold (FG) or (4-[4-didecylaminostyryl]-N-methylpyridinium iodide (4Di-10ASP) used in this study, were transported retrogradely to the RGC soma when either dye was injected into the superior colliculus (SC) at P0. Some of these labeled RGCs die due to natural apoptosis during this stage of development and are phagocytosed by microglial cells, which move to the site of RGC death, to become labeled with the same dye. The retinas were examined to quantify the microglial cells from P5 to adulthood. In addition, the reaction of microglia to optic nerve crush was studied in adult animals. Both dyes labeled RGCs in the contralateral retina and a few RGCs in the retina ipsilateral to the injected SC. The density of labeled RGCs decreased by 22% between P5 and P7. During this phase, microglial cells become visible as they ingested the fluorescent detritus of the dying RGCs. Microglial cells were evenly distributed across the entire retinal surface and migrated to the outer plexiform layer. Migrating microglia consecutively altered their morphology from the amoeboid to the ramified form. In terms of intracellular storage of the dyes, resident microglial cells retained the fluorescent dye 4Di-10ASP over a period of 12 months. In contrast, FG was completely transferred from the RGCs and microglial cells to intramural cells (pericytes) of the retinal capillaries after 10 months. This resulted in delineation of the entire intraretinal vascular network. Finally, resident retinal microglial cells were also activated by injury to the adult optic nerve and phagocytosed degenerating neurons. Retinal microglial cells can be monitored with vital fluorescent dyes while they migrate across the retina and establish their intra-retinal network. It is possible to label microglia with lipophilic dyes and they remain labeled for a long time. In addition, intramural pericytes can be labeled by slow release of FG from RGCs and microglial cells. The findings suggest that ingested fluorescent dyes having different properties can be used to study different populations of retinal cells in vivo.     

Anatomy and spatial organization of Müller glia in mouse retina

Müller glia, the most abundant glia of vertebrate retina, have an elaborate morphology characterized by a vertical stalk that spans the retina and branches in each retinal layer. Müller glia play diverse, critical roles in retinal homeostasis, which are presumably enabled by their complex anatomy. However, much remains unknown, particularly in mouse, about the anatomical arrangement of Müller cells and their arbors, and how these features arise in development. Here we use membrane‐targeted fluorescent proteins to reveal the fine structure of mouse Müller arbors. We find sublayer‐specific arbor specializations within the inner plexiform layer (IPL) that occur consistently at defined laminar locations. We then characterize Müller glia spatial patterning, revealing how individual cells collaborate to form a pan‐retinal network. Müller cells, unlike neurons, are spread across the retina with homogenous density, and their arbor sizes change little with eccentricity. Using Brainbow methods to label neighboring cells in different colors, we find that Müller glia tile retinal space with minimal overlap. The shape of their arbors is irregular but nonrandom, suggesting that local interactions between neighboring cells determine their territories. Finally, we identify a developmental window at postnatal Days 6 to 9 when Müller arbors first colonize the synaptic layers beginning in stereotyped inner plexiform layer sublaminae. Together, our study defines the anatomical arrangement of mouse Müller glia and their network in the radial and tangential planes of the retina, in development and adulthood. The local precision of Müller glia organization suggests that their morphology is sculpted by specific cell to cell interactions with neurons and each other.     

Stereotyped axonal arbors of retinal ganglion cell subsets in the mouse superior colliculus

Mouse retinal ganglion cells (RGCs) have been classified into around 20 subtypes based on the shape, size, and laminar position of their dendritic arbors. In most cases tested, RGC subtypes classified in this manner also have distinct functional signatures. Here we asked whether RGC subtypes defined by dendritic morphology have stereotyped axonal arbors in their main central target, the superior colliculus (SC). We used transgenic and viral methods to sparsely label RGCs and characterized both dendritic and axonal arbors of individual RGCs. Axon arbors varied in size, shape, and laminar position. For each of 12 subtypes defined dendritically, however, axonal arbors in the contralateral SC showed considerable stereotypy. We found no systematic relationship between the laminar position of an RGC's dendrites within the inner plexiform layer and that of its axon within the retinorecipient zone of the SC, suggesting that distinct developmental mechanisms specify dendritic and axonal laminar positions. We did, however, note a significant correlation between the dendritic field sizes of RGCs and the laminar position of their axon arbors: RGCs with larger dendritic areas, and hence larger receptive fields, projected to deeper strata within the SC. Finally, combining these new results with previous physiological analyses, we find that RGC subtypes that share similar functional properties, such as directional selectivity, project to similar depths within the SC.     

Efficient estimation of retinal ganglion cell number: a stereological approach

Retinal ganglion cells (RGCs) are the only output neurons of the retina, and their degeneration after damage to the optic nerve or in glaucoma is a well established system for studying apoptosis in the central nervous system. Frequently used procedures for assessing RGC number in retinal flat mounts suffer from two problems: RGC densities are not uniform across retinal flat mounts, and density measures may therefore not reflect total number, and flat mounts do not allow efficient use of tissue. To overcome these problems we developed a stereological method for efficiently assessing RGC number in cryostat sections of the retina. We empirically demonstrate that only approximately 1:20 sections need be assessed to accurately estimate the total number of RGCs in the rat retina, providing ample tissue for additional studies in the same retina and saving considerably on more exhaustive sampling strategies. Using this method, we estimate that there are 86,282+/-4759 RGCs in the normal Brown Norway rat retina. These counts match well with estimates of axon counts in optic nerve. In a pilot study of experimental glaucoma, we determined a reduction of RGCs to 53,862+/-4272 (p<0.05). The current technique should prove advantageous to assess neuroprotective strategies in these experimental models.     

Asymmetric distribution of retinal ganglion cells in goldfish

The distribution of retinal ganglion cells (RGCs) in goldfish was determined by removing an eye and applying cobaltous-lysine to the optic nerve for 24 hr. This procedure allowed the cobalt label to be in continuous contact with the cut ends of the optic axons and thereby backfilled many RGCs. RGC density was determined across three different sizes of retinae by using fish with different eye sizes. Confirming earlier work, we found that RGC density diminished as retinal area increased. However, irrespective of the retinal size, the density of RGCs was elevated along the temporal boundary between the dorsal and the ventral retina. A conservative estimate indicated that the RGC density in the temporal retina was at least 1.8-2.5 times higher than the mean RGC density of the entire retina. Thus, the goldfish retina does not appear to have a homogeneous distribution of RGCs as was previously considered. Small and large retinae differed with respect to the percentage of cells in the RGC layer that was RGCs. In small retinae, even when the noncobalt-filled cells (glia and displaced amacrine cells) were added to the cobalt-filled RGCs, the density of all cell types was elevated in the temporal retina relative to the remainder of the retina. Furthermore, in small retinae, the percentage of cells in the RGC layer that was RGCs (75%) was constant across the radial and circumferential aspects of the retina. In marked contrast, in medium-large retinae, a homogeneous distribution of cells across the entire retina resulted when the noncobalt-filled cells were added to the cobalt-filled cells. However, the percentage of cells that was cobalt-filled RGCs was significantly greater in the temporal retina (50%) than in the remainder of the retina (35%). In large retinae, as in small retinae, the percentage of cells that was RGCs did not vary as a function of distance from the optic disc. These data suggest that, in the course of retinal maturation, cell density in the temporal retina is elevated initially and then declines subsequently to the level of the surrounding retina. Over time, more displaced to the level of the surrounding retina. Over time, more displaced amacrine cells may be added to the tissue surrounding the temporal retina. Alternatively, more RGCs outside the temporal retina may become displaced amacrine cells. Such events could account for the growth-associated, disproportionate decrease in the percentage of cells that is RGCs in the tissue surrounding the temporal retina.(ABSTRACT TRUNCATED AT 400 WORDS)     

Selective synaptic connections in the retinal pathway for night vision

The mammalian retina encodes visual information in dim light using rod photoreceptors and a specialized circuit: rods→rod bipolar cells→AII amacrine cell. The AII amacrine cell uses sign-conserving electrical synapses to modulate ON cone bipolar cell terminals and sign-inverting chemical (glycinergic) synapses to modulate OFF cone cell bipolar terminals; these ON and OFF cone bipolar terminals then drive the output neurons, retinal ganglion cells (RGCs), following light increments and decrements, respectively. The AII amacrine cell also makes direct glycinergic synapses with certain RGCs, but it is not well established how many types receive this direct AII input. Here, we investigated functional AII amacrine→RGC synaptic connections in the retina of the guinea pig (Cavia porcellus) by recording inhibitory currents from RGCs in the presence of ionotropic glutamate receptor (iGluR) antagonists. This condition isolates a specific pathway through the AII amacrine cell that does not require iGluRs: cone→ON cone bipolar cell→AII amacrine cell→RGC. These recordings show that AII amacrine cells make direct synapses with OFF Alpha, OFF Delta and a smaller OFF transient RGC type that co-stratifies with OFF Alpha cells. However, AII amacrine cells avoid making synapses with numerous RGC types that co-stratify with the connected RGCs. Selective AII connections ensure that a privileged minority of RGC types receives direct input from the night-vision pathway, independent from OFF bipolar cell activity. Furthermore, these results illustrate the specificity of retinal connections, which cannot be predicted solely by co-stratification of dendrites and axons within the inner plexiform layer.