Molecular Fingerprinting of On-Off Direction-Selective Retinal Ganglion Cells Across Species and Relevance to Primate Visual Circuits.

The ability to detect moving objects is an ethologically salient function. Direction-selective neurons have been identified in the retina, thalamus, and cortex of many species, but their homology has remained unclear. For instance, it is unknown whether direction-selective retinal ganglion cells (DSGCs) exist in primates and, if so, whether they are the equivalent to mouse and rabbit DSGCs. Here, we used a molecular/circuit approach in both sexes to address these issues. In mice, we identify the transcription factor Satb2 (special AT-rich sequence-binding protein 2) as a selective marker for three RGC types: On-Off DSGCs encoding motion in either the anterior or posterior direction, a newly identified type of Off-DSGC, and an Off-sustained RGC type. In rabbits, we find that expression of Satb2 is conserved in On-Off DSGCs; however, it has evolved to include On-Off DSGCs encoding upward and downward motion in addition to anterior and posterior motion. Next, we show that macaque RGCs express Satb2 most likely in a single type. We used rabies virus-based circuit-mapping tools to reveal the identity of macaque Satb2-RGCs and discovered that their dendritic arbors are relatively large and monostratified. Together, these data indicate Satb2-expressing On-Off DSGCs are likely not present in the primate retina. Moreover, if DSGCs are present in the primate retina, it is unlikely that they express Satb2.SIGNIFICANCE STATEMENT The ability to detect object motion is a fundamental feature of almost all visual systems. Here, we identify a novel marker for retinal ganglion cells encoding directional motion that is evolutionarily conserved in mice and rabbits, but not in primates. We show in macaque monkeys that retinal ganglion cells (RGCs) that express this marker comprise a single type and are morphologically distinct from mouse and rabbit direction-selective RGCs. Our findings indicate that On-Off direction-selective retinal neurons may have evolutionarily diverged in primates and more generally provide novel insight into the identity and organization of primate parallel visual pathways.

Identification of a pathway from the retina to koniocellular layer K1 in the lateral geniculate nucleus of marmoset.

Three well characterized pathways in primate vision (midget-parvocellular, parasol-magnocellular, bistratified-koniocellular) have been traced from the first synapse in the retina, through the visual thalamus (lateral geniculate nucleus, LGN), to the visual cortex. Here we identify a pathway from the first synapse in the retina to koniocellular layer K1 in marmoset monkeys (Callithrix jacchus). Particle-mediated gene transfer of an expression plasmid for the postsynaptic density 95-green fluorescent protein (PSD95-GFP) was used to label excitatory synapses on retinal ganglion cells and combined with immunofluorescence to identify the presynaptic bipolar cells. We found that axon terminals of one type of diffuse bipolar cell (DB6) provide dominant synaptic input to the dendrites of narrow thorny ganglion cells. Retrograde tracer injections into the LGN and photofilling of retinal ganglion cells showed that narrow thorny cells were preferentially labeled when koniocellular layer K1 was targeted. Layer K1 contains cells with high sensitivity for rapid movement, and layer K1 sends projections to association visual areas as well as to primary visual cortex. We hypothesize that the DB6-narrow thorny-koniocellular pathway contributes to residual visual functions ("blindsight") that survive injury to primary visual cortex in adult or early life.

Kainate receptors mediate synaptic input to transient and sustained OFF visual pathways in primate retina.

Visual signals are segregated into parallel pathways at the first synapse in the retina between cones and bipolar cells. Within the OFF pathways of mammals, the selective expression of AMPA or kainate-type glutamate receptors in the dendrites of different OFF-bipolar cell types is thought to contribute to formation of distinct temporal channels. AMPA receptors, with rapid recovery from desensitization, are proposed to transmit high temporal frequency signals, whereas kainate receptors (KARs) are presumed to encode lower temporal frequencies. Here we studied the glutamate receptors expressed by OFF-bipolar cells in slice preparations of macaque monkey retina, where the low (midget/parvocellular) and high-frequency (parasol/magnocellular) temporal channels are well characterized. We found that all OFF-bipolar types receive input primarily through KARs and that KAR antagonists block light-evoked input to both OFF-midget and OFF-parasol ganglion cells. KAR subunits were differentially expressed in OFF-bipolar types; the diffuse bipolar (DB) cells, DB2 and DB3b, expressed GluK1 and showed transient responses to glutamate and the KAR agonist, ATPA. In contrast, flat midget bipolar, DB1, and DB3a cells lacked GluK1 and showed relatively sustained responses. Finally, we found that the KAR accessory protein, Neto1, is expressed at the base of cone pedicles but is not colocalized with the GluK1 subunit. In summary, the results indicate that transient signaling in the OFF pathway of macaques is not dependent on AMPA receptors and that heterogeneity of KARs and accessory proteins may contribute to the formation of parallel temporal channels.

Organisation of koniocellular-projecting ganglion cells and diffuse bipolar cells in the primate fovea.

The roles of the midget and parasol pathways as the anatomical foundation for high-acuity vision at the fovea are well established. There is also evidence for the presence of other (non-midget, non-parasol) ganglion cell types in the foveal retina, but it is not established whether these cells receive input from cone photoreceptors in the central few degrees of the visual field, i.e. the region most important for conscious visual perception. To address this question, we targeted injections of retrograde tracer to the koniocellular layers in the posterior aspect of the lateral geniculate nucleus, where the central visual field is represented, in marmoset monkeys (Callithrix jacchus). Labeled ganglion cells were photofilled to reveal their dendritic morphology. Potential inputs to foveal koniocellular cells from diffuse bipolar cells were investigated in vertical sections through the fovea of marmoset and macaque (Macaca fascicularis) monkey retinas using immunohistochemistry. Forty koniocellular-projecting ganglion cells were analysed. We used an established model of marmoset foveal topography to show that all these koniocellular-projecting cells receive cone inputs from the central-most 6°, with about half the cells receiving input from below 2° eccentricity, in the rod-free central bouquet of cones at the foveola. In addition, all diffuse bipolar types investigated were present in the fovea at stratification depths similar to those of their counterparts in the peripheral retina. We conclude that the diverse visual representations established for koniocellular pathways in the peripheral retina are also a feature of the fovea, suggesting that koniocellular pathways contribute to foveal vision.

Amacrine and bipolar inputs to midget and parasol ganglion cells in marmoset retina.

Retinal ganglion cells receive excitatory synapses from bipolar cells and inhibitory synapses from amacrine cells. Previous studies in primate suggest that the strength of inhibitory amacrine input is greater to cells in peripheral retina than to foveal (central) cells. A comprehensive study of a large number of ganglion cells at different eccentricities, however, is still lacking. Here, we compared the amacrine and bipolar input to midget and parasol ganglion cells in central and peripheral retina of marmosets (Callithrix jacchus). Ganglion cells were labeled by retrograde filling from the lateral geniculate nucleus or by intracellular injection. Presumed amacrine input was identified with antibodies against gephyrin; presumed bipolar input was identified with antibodies against the GluR4 subunit of the AMPA receptor. In vertical sections, about 40% of gephyrin immunoreactive (IR) puncta were colocalized with GABAA receptor subunits, whereas immunoreactivity for gephyrin and GluR4 was found at distinct sets of puncta. The density of gephyrin IR puncta associated with ganglion cell dendrites was comparable for midget and parasol cells at all eccentricities studied (up to 2 mm or about 16 degrees of visual angle for midget cells and up to 10 mm or >80 degrees of visual angle for parasol cells). In central retina, the densities of gephyrin IR and GluR4 IR puncta associated with the dendrites of midget and parasol cells are comparable, but the average density of GluR4 IR puncta decreased slightly in peripheral parasol cells. These anatomical results indicate that the ratio of amacrine to bipolar input does not account for the distinct functional properties of parasol and midget cells or for functional differences between cells of the same type in central and peripheral retina.

Calcium-permeable AMPA receptors on AII amacrine cells mediate sustained signaling in the On-pathway of the primate retina.

Midget and parasol ganglion cells (GCs) represent the major output channels from the primate eye to the brain. On-type midget and parasol GCs exhibit a higher background spike rate and thus can respond more linearly to contrast changes than their Off-type counterparts. Here, we show that a calcium-permeable AMPA receptor (CP-AMPAR) antagonist blocks background spiking and sustained light-evoked firing in On-type GCs while preserving transient light responses. These effects are selective for On-GCs and are occluded by a gap-junction blocker suggesting involvement of AII amacrine cells (AII-ACs). Direct recordings from AII-ACs, cobalt uptake experiments, and analyses of transcriptomic data confirm that CP-AMPARs are expressed by primate AII-ACs. Overall, our data demonstrate that under some background light levels, CP-AMPARs at the rod bipolar to AII-AC synapse drive sustained signaling in On-type GCs and thus contribute to the more linear contrast signaling of the primate On- versus Off-pathway.

Glycinergic Inhibition Targets Specific Off Cone Bipolar Cells in Primate Retina.

Adapting between scotopic and photopic illumination involves switching the routing of retinal signals between rod and cone-dominated circuits. In the daytime, cone signals pass through parallel On and Off cone bipolar cells (CBCs), that are sensitive to increments and decrements in luminance, respectively. At night, rod signals are routed into these cone-pathways via a key glycinergic interneuron, the AII amacrine cell (AII-AC). AII-ACs also provide On-pathway-driven crossover inhibition to Off-CBCs under photopic conditions. In primates, it is not known whether all Off-bipolar cell types receive functional inputs from AII-ACs. Here, we show that select Off-CBC types receive significantly higher levels of On-pathway-driven glycinergic input than others. The rise and decay kinetics of the glycinergic events are consistent with involvement of the α1 glycine receptor (GlyR) subunit, a result supported by a higher level of GLRA1 transcript in these cells. The Off-bipolar types that receive glycinergic input have sustained physiological properties and include the flat midget bipolar (FMB) cells, which provide excitatory input to the Off-midget ganglion cells (GCs; parvocellular pathway). Our results suggest that only a subset of Off-bipolar cells have the requisite receptors to respond to AII-AC input. Taken together with results in mouse retina, our findings suggest a conserved motif whereby signal output from AII-ACs is preferentially routed into sustained Off-bipolar signaling pathways.

Survey of retinal ganglion cell morphology in marmoset.

In primate retina, the midget, parasol, and small bistratified cell populations form the large majority of ganglion cells. In addition, there is a variety of low-density wide-field ganglion cell types that are less well characterized. Here we studied retinal ganglion cells in the common marmoset, Callithrix jacchus, using particle-mediated gene transfer. Ganglion cells were transfected with an expression plasmid for the postsynaptic density 95-green fluorescent protein. The retinas were processed with established immunohistochemical markers for bipolar and/or amacrine cells to determine ganglion cell dendritic stratification. In total over 500 ganglion cells were classified based on their dendritic field size, morphology, and stratification in the inner plexiform layer. Over 17 types were distinguished, including midget, parasol, broad thorny, small bistratified, large bistratified, recursive bistratified, recursive monostratified, narrow thorny, smooth monostratified, large sparse, giant sparse (melanopsin) ganglion cells, and a group that may contain several as yet uncharacterized types. Assuming each characterized type forms a hexagonal mosaic, the midget and parasol cells account for over 80% of all ganglion cells in the central retina but only ∼50% of cells in the peripheral (>2 mm) retina. We conclude that the fovea is dominated by midget and parasol cells, but outside the fovea the ganglion cell diversity in marmoset is likely as great as that reported for nonprimate retinas. Taken together, the ganglion cell types in marmoset retina resemble those described previously in macaque retina with respect to morphology, stratification, and change in proportion across the retina.

Directional excitatory input to direction-selective ganglion cells in the rabbit retina.

Directional responses in retinal ganglion cells are generated in large part by direction-selective release of γ-aminobutyric acid from starburst amacrine cells onto direction-selective ganglion cells (DSGCs). The excitatory inputs to DSGCs are also widely reported to be direction-selective, however, recent evidence suggests that glutamate release from bipolar cells is not directional, and directional excitation seen in patch-clamp analyses may be an artifact resulting from incomplete voltage control. Here, we test this voltage-clamp-artifact hypothesis in recordings from 62 ON-OFF DSGCs in the rabbit retina. The strength of the directional excitatory signal varies considerably across the sample of cells, but is not correlated with the strength of directional inhibition, as required for a voltage-clamp artifact. These results implicate additional mechanisms in generating directional excitatory inputs to DSGCs.

Connectivity between the OFF bipolar type DB3a and six types of ganglion cell in the marmoset retina.

Parallel visual pathways originate at the first synapse in the retina, where cones make connections with cone bipolar cells that in turn contact ganglion cells. There are more ganglion cell types than bipolar types, suggesting that there must be divergence from bipolar to ganglion cells. Here we analyze the contacts between an OFF bipolar type (DB3a) and six ganglion cell types in the retina of the marmoset monkey (Callithrix jacchus). Ganglion cells were transfected via particle-mediated gene transfer of an expression plasmid for the postsynaptic density 95-green fluorescent protein (PSD95-GFP), and DB3a cells were labeled via immunohistochemistry. Ganglion cell types that fully or partially costratified with DB3a cells included OFF parasol, OFF midget, broad thorny, recursive bistratified, small bistratified, and large bistratified cells. On average, the number of DB3a contacts to parasol cells (18 contacts per axon terminal) is higher than that to other ganglion cell types (between four and seven contacts). We estimate that the DB3a output to OFF parasol cells accounts for at least 30% of the total DB3a output. Furthermore, we found that OFF parasol cells receive approximately 20% of their total bipolar input from DB3a cells, suggesting that other diffuse bipolar types also provide input to OFF parasol cells. We conclude that DB3a cells preferentially contact OFF parasol cells but also provide input to other ganglion cell types.

Analysis of bipolar and amacrine populations in marmoset retina.

About 15 parallel ganglion cell pathways transmit visual signals to the brain, but the interneuron (bipolar and amacrine) populations providing input to ganglion cells remain poorly understood in primate retina. We carried out a quantitative analysis of the inner nuclear layer in the retina of the marmoset (Callithrix jacchus). Vertical Vibratome sections along the horizontal meridian were processed with immunohistochemical markers. Image stacks were taken with a confocal microscope, and densities of cell populations were determined. The density of flat midget bipolar cells fell from 15,746 cells/mm(2) at 1 mm (8 deg) to 7,827 cells/mm(2) at 3 mm (25 deg). The rod bipolar cell density fell from 8,640 cells/mm(2) at 1 mm to 4,278 cells/mm(2) at 3 mm, but the ratio of the two bipolar cell types did not change with eccentricity. The amacrine cell density ranged from 30,000 cells/mm(2) at 8 deg to less than 15,000 cells/mm(2) at 25 deg, but throughout the retina, the ratio of glycinergic to γ-aminobutyric acid (GABA)ergic to amacrine cells remained relatively constant. The fractions of rod bipolar, cone bipolar, amacrine, Müller, and horizontal cells of all cells in the inner nuclear layer were comparable in central and peripheral retina. Marmosets had lower proportions of midget bipolar and rod bipolar in comparison with macaque. These differences were correlated with differences in rod and cone densities between the two species and did not reflect fundamental differences in the wiring between the two species.

Synaptic inputs to two types of koniocellular pathway ganglion cells in marmoset retina.

The retinal connectivity of the diverse group of cells contributing to koniocellular visual pathways (widefield ganglion cells) is largely unexplored. Here we examined the synaptic inputs onto two koniocellular-projecting ganglion cell types named large sparse and broad thorny cells. Ganglion cells were labeled by retrograde tracer injections targeted to koniocellular layer K3 in the lateral geniculate nucleus in marmosets (Callithrix jacchus) and subsequently photofilled. Retinal preparations were processed with antibodies against the C-terminal binding protein 2, the AMPA receptor subunit GluR4, and against CD15 to identify bipolar (excitatory) and/or antibodies against gephyrin to identify amacrine (inhibitory) input. Large sparse cells are narrowly stratified close to the ganglion cell layer. Broad thorny ganglion cells are broadly stratified in the center of the inner plexiform layer. Bipolar input to large sparse cells derives from DB6 and maybe other ON bipolar types, whereas that to broad thorny cells derives from ON and OFF bipolar cell types. The total number of putative synapses on broad thorny cells is higher than the number on large sparse cells but the density of inputs (between 2 and 5 synapses per 100 µm(2) dendritic area) is similar for the two cell types, indicating that the larger number of synapses on broad thorny cells is attributable to the larger membrane surface area of this cell type. Synaptic input density is comparable to previous values for midget-parvocellular and parasol-magnocellular pathway cells. This suggests functional differences between koniocellular, parvocellular, and magnocellular pathways do not arise from variation in synaptic input densities.

Synaptic inputs onto small bistratified (blue-ON/yellow-OFF) ganglion cells in marmoset retina.

The inner plexiform layer of the retina contains functional subdivisions, which segregate ON and OFF type light responses. Here, we studied quantitatively the ON and OFF synaptic input to small bistratified (blue-ON/yellow-OFF) ganglion cells in marmosets (Callithrix jacchus). Small bistratified cells display an extensive inner dendritic tier that receives blue-ON input from short-wavelength-sensitive (S) cones via blue cone bipolar cells. The outer dendritic tier is sparse and is thought to receive yellow-OFF input from medium (M)- and long (L)-wavelength-sensitive cones via OFF diffuse bipolar cells. In total, 14 small bistratified cells from different eccentricities were analyzed. The cells were retrogradely labeled from the koniocellular layers of the lateral geniculate nucleus and subsequently photofilled. Retinal preparations were processed with antibodies against the C-terminal binding protein 2, the AMPA receptor subunit GluR4, and/or gephyrin to identify bipolar and/or amacrine input. The results show that the synaptic input is evenly distributed across the dendritic tree, with a density similar to that reported previously for other ganglion cell types. The population of cells showed a consistent pattern, where bipolar input to the inner tier is about fourfold greater than bipolar input to the outer tier. This structural asymmetry of bipolar input may help to balance the weight of cone signals from the sparse S cone array against inputs from the much denser M/L cone array.

Distribution of bipolar input to midget and parasol ganglion cells in marmoset retina.

Different types of retinal ganglion cell show differences in their response properties. Here we investigated the question of whether these differences are related to the distribution of the synaptic input to the dendritic tree. We measured the distribution and density of synaptic input to the dendrites of midget and parasol ganglion cells in the retina of a New World monkey, the marmoset, Callithrix jacchus. Ganglion cells were retrogradely labeled by dye injection into parvocellular or magnocellular regions of the lateral geniculate nucleus and subsequently photo-filled. Presumed bipolar cell synapses were identified immunocytochemically using antibodies against the ribbon protein CtBP2 or the GluR4 subunit of the AMPA receptor. For all cells, colocalized immunoreactive puncta were distributed across the entire dendritic tree. The density of the presumed bipolar input to midget ganglion cells was comparable for both synaptic markers, suggesting that the AMPA receptor GluR4 subunit is expressed at all synapses between midget bipolar and midget ganglion cells. Midget ganglion cells had an average of nine colocalized immunoreactive puncta per 100 microm2 dendritic surface, and parasol cells had an average of seven colocalized immunoreactive puncta per 100 microm2 dendritic surface. The densities were comparable in different regions of the dendritic tree and were not influenced by the location of the cells with respect to the fovea. Our findings suggest that the differences in the response characteristics of midget and parasol cells are not due to differences in the density of synaptic input to their dendritic tree.