S-cone-specific circuitry in the outer plexiform layer of a cone-dominant mammal.

S-cone-specific circuitry in the outer plexiform layer of a cone-dominant mammal - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Proc Natl Acad Sci U S A . 2025 Dec 1;122(49):e2504954122. doi: 10.1073/pnas.2504954122 S-cone-specific circuitry in the outer plexiform layer of a cone-dominant mammal Yizhen Zhang a, 1, 2 , Shan Chen a, 1 , Haohua Qian b , Wei Li a, 3, 4 , John M Ball a, 4 Author information Article notes Copyright and License information a Retinal Neurophysiology Section, National Eye Institute, National Institutes of Health, Bethesda, MD 20892 b Visual Function Core, National Eye Institute, National Institutes of Health, Bethesda, MD 20892 4 To whom correspondence may be addressed. Email: liweivision@sjtu.edu.cn or john.ball2@nih.gov . Edited by Rachel Wong, University of Washington, Seattle, WA; received March 3, 2025; accepted September 22, 2025 1 Y.Z. and S.C. contributed equally to this work. 2 Present address: Molecular Genetics Section, National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, MD 20892. 3 Present address: Shanghai Academy of Natural Sciences Institute for Neuroscience and Vision Research, Shanghai Jiao Tong University School of Medicine, Shanghai 201399, China. Received 2025 Mar 3; Accepted 2025 Sep 22; Issue date 2025 Dec 9. Copyright © 2025 the Author(s). Published by PNAS. This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND) . PMC Copyright notice PMCID: PMC12674004  NIHMSID: NIHMS2114588 PMID: 41325528 Significance Retinal circuitry that detects and encodes blue light in the visual field holds importance for color perception, myopia, and mood disorders that depend upon light exposure. Few animal models are ideally situated for the study of human S-cone vision; for example, mice have a rod-dominant retina and a large population of mixed green-blue sensitive cones that are not present in humans. The ground squirrel, in contrast, possesses a cone-dominant retina more akin to the ancient mammalian color circuit from which the human system evolved. The anatomical and physiological data provided here establishes critical features of color discrimination circuits in the ground squirrel retina. Keywords: color vision, retina, ground squirrel, S-cone, connectomics Abstract In the vertebrate retina, short wavelength-sensitive S-cones and their downstream interneurons play unique roles in both image forming and non-image-forming vision. Due to their conserved relative rarity in the retina and the high density of rods found in many model species, S-cone circuitry is challenging to examine in detail. Here, we combine electrophysiology, high-quality 3D electron microscopy reconstruction, and electroretinography to characterize in fine detail the S-cone selective circuitry in the outer plexiform layer (OPL) of the cone-dominant retina of the thirteen-lined ground squirrel (13-LGS). The 13-LGS retina has no S-OFF bipolar cell, and both horizontal cell types show distinct S-cone preferences. Signal processing in the 13-LGS OPL appears to have evolved for separate channels that independently modulate synaptic sensitivity of S- and M-cone networks. Color vision in most mammals is mediated by two cone photoreceptor types that appear well-adapted for life in a largely green environment under a blue sky: Long or middle wavelength-sensitive cones (L- or M-cones) are most sensitive to light that humans perceive as green, while short wavelength-sensitive cones (S-cones) instead preferentially absorb light perceived as blue ( 1 ). S-cones and their downstream circuitry are perhaps the best-studied pathway for color vision in the retina ( 1 , 2 ); however, significant gaps remain in our understanding of this system. For example, although it is understood that color discrimination results from local comparisons of S- and L/M- cones, our understanding of the mechanistic basis for discrimination is impeded by the existence of intricate interactions between blue and green encoding neurons across retinal synaptic layers ( 3 , 4 ). Notably, congenital stationary night blindness type 1 (CSNB1) patients demonstrate normal perceptual color vision despite having deficits in S-cone ERGs in the peripheral retina, implying differential effects on central vs. peripheral S-cone circuitry ( 5 ). Furthermore, experimental modulation of the effective chromatic aberration in the eye by differentially blurring short-wavelength light influences eye growth ( 6 )—this observation is illuminating in the context of the current myopia epidemic, theorized to arise from the stark difference between the natural conditions shaping visual evolution and the artificial visual stimuli that have saturated modern electronic life ( 7 ). It has also become clear that retinal neurons carrying S-cone signals make preferential synaptic contacts with intrinsically photosensitive retinal ganglion cells (ipRGCs) ( 8 , 9 ), directly linking blue light-sensitivity to circadian entrainment and mood disorders ( 10 ). Therefore, a thorough understanding of S-cone circuitry is crucial for understanding the ways in which blue light influences both image-forming and non-image-forming vision. Uncertainties remain regarding the role of S-cones in color perception vs. other chromatically guided visual behaviors, especially concerning the retinal circuitry involved and the extent to which they overlap. In humans and nonhuman primates, “midget” retinal circuitry, known for its hallmark specificity and private-line connections, subserves high-resolution vision in the primate fovea ( 4 , 11 – 14 ).This system appears to be unique to primates, although differences in the S-cone midget pathway seemingly exist between old- and new-world monkeys ( 3 , 15 , 16 ). In contrast, in most mammals and in the peripheral human retina, S-cones and their downstream circuitry also form a key component of a more ancient “primordial” color circuit ( 17 ) that may subserve a number of chromatically guided behaviors such as prey capture and threat avoidance ( 18 , 19 ). The S-cone-driven signals in this circuit may code chromatic cues in the visual world that aid survival, in contrast to the color-coded segmentation of the visual scene we experience as humans ( 4 , 20 – 22 ). Cones, being individually unable to distinguish differences in wavelength from differences in intensity ( 23 ), require downstream retinal circuits that combine and compare signals from these cone types in order to generate color discrimination ( 24 ). This begins in the outer plexiform layer (OPL), where ON and OFF cone bipolar cells (CBCs), named for their opposite polarity responses to glutamate released by cones, extend dendrites that form synaptic contacts with ribbon synapses within cone terminals ( Fig. 1 A ) ( 25 ). In primates, this primordial color circuit features an ON S-cone bipolar cell (SCBC) that receives exclusive input from S-cones and synapses primarily onto a small bistratified retinal ganglion cell (sbRGC) ( 26 ). While the sbRGC has not been identified in other species, the SCBC is well conserved in mammalian retinas ( 23 ). The presence of a complementary S-cone selective OFF bipolar cell remains controversial, although the midget system varyingly employs an S-OFF midget bipolar cell according to species ( 3 , 27 ). Horizontal cells (HCs) contact cone terminals within their dendritic field, pooling cone signals and feeding back onto photoreceptor terminals by shaping their synaptic output ( 28 ). Where the SCBC appears to be the preferred channel for S-cone signals to reach the inner retina, evolution seemingly has allowed HCs more leeway, and the contribution of HCs to color discrimination remains unsettled: Most species lack an S-cone-specific HC, with the exception of the primate Hii cell, whose dense preferential contacts with S-cone terminals shape chromatic surround inhibition, contributing to S-ON, L/M-OFF polarity responses in sbRGCs ( 29 , 30 ). HC subtypes in many species also feature a long enigmatic axon that projects laterally within the OPL to target photoreceptor terminals of varying type specificity ( 31 ); the specific function of these axons remains elusive. Fig. 1. Open in a new tab Dendritic reconstructions in the 13-LGS OPL. ( A ). Cartoon section of 13-LGS OPL depicting cones (above; green) with terminals containing synaptic ribbons (yellow). CBCs (below; blue-green), and HCs ( middle ; magenta) in the OPL. Dashed lines indicate the borders of the OPL. ( B ). Reconstructed dendritic processes in the SBEM volume. Processes are colored as in panel A . The dashed outer yellow line encompasses the approximate region containing ribbons (see F ). ( C ). Whole-mount view of the two terminals within the white dashed line in B . The bottom terminal is the S-cone marked “i” in this figure and others. ( D ). Single EM slice through S-cone terminal i. SCBC is marked with an arrow, cone axon with an arrowhead. Colors as in panels A and B . The dashed region is shown in detailed view in the Lower panel: Two ribbons (arrowheads) are flanked by three HC processes, the center of which belongs to an H2 cell (arrow). ( E ). Whole-mount view of cone terminals in the SBEM (S-cones marked i and ii). ( F ). Reconstructed SCBCs. S-cone ribbons are shown in red. Note the blind ending (arrowhead) of the SCBC contacting S-cone i. ( G ). Detailed view of CBC tips contacting S-cone i. Right : CBC tips with the SCBC hidden. ( H ). Detailed view of CBC tips contacting S-cone ii. Here, we take advantage of the dichromatic cone-dominant retina of the thirteen-lined ground squirrel (13-LGS) ( 32 ) and present a thorough accounting of its S-cone circuitry in the OPL using multiple convergent lines of anatomical and physiological evidence including high-resolution serial block-face electron microscopy (SBEM), electrophysiology, electroretinography, and immunocytochemistry. Despite prior anatomical ( 33 – 39 ) and physiological characterization of the ground squirrel retina ( 40 – 43 ), the S-cone specificity of the 13-LGS SCBC has not yet been confirmed at the functional or EM level, and the absence of evidence for an S-OFF bipolar cell leaves the issue unresolved. The specificity of the two HC subtypes in the 13-LGS has only been probed using light microscopy ( 39 ). We find the only S-cone-specific bipolar cell to be the canonical ON SCBC, and find no evidence of an OFF SCBC as a core component of S-cone circuitry in the 13-LGS retina, indicating that the S-cone amacrine cell (SCA) is likely to be the main substrate for S-OFF signaling ( 44 , 45 ). Further, we show S-cone selectivity of the H2 cell anatomically and functionally. Interestingly, we also find that despite the nonselectivity of the H1 dendritic tree, their axons give rise to a dense network of collateral processes contacting exclusively M-cones. Combined, these findings underscore the particularly chromatically privileged nature of the 13-LGS S-cone circuit in the OPL, a promising model for uncovering the primordial mammalian S-cone circuit. Results ON S-Cone Bipolar Cells Are the Only S-Cone-Specific Bipolar Cells in 13-LGS Retina. To identify the postsynaptic neurons in the 13-LGS OPL that contact S-cones, we reconstructed dendritic processes in a SBEM retinal volume from the medial dorsal retina focusing upon photoreceptor synaptic terminals in the OPL and containing the border between the outer and inner nuclear layers (ONL and INL, respectively; Fig. 1 A ). This volume contained 39 complete photoreceptor terminals, all of which were cone pedicles ( Fig. 1 B and C ); we found no rods, which would have been identified by their small, sparsely innervated synaptic terminals containing fewer synaptic ribbons ( 38 ). Using the criteria that SCBCs are expected to have 1 to 2 thick dendritic trunks that exclusively and deeply innervate S-cone terminals to occupy invaginating ON positions ( Fig. 1 D ) ( 23 , 39 ), we identified two cone terminals as belonging to S-cones ( Fig. 1 D and E ), constituting 5.1% of the cones in the volume, consistent with reported estimates ( 46 – 48 ). This proportion falls within our observations in the dorsal 13-LGS retina using immunocytochemistry ( SI Appendix , Fig. S1 ). S-cone terminals were somewhat smaller than M-cone terminals, similar to other species ( 49 ) ( Fig. 1 E and SI Appendix , Fig. S2 ). Putative S-cones also had significantly more and smaller ribbons than did M-cones (43 ± 4.2 vs. 24.7 ± 3.3; SI Appendix , Fig. S2 A – C ), a curious observation given that S-cones typically have fewer ribbons ( 3 ). In a 13-LGS retina immunolabeled for S-opsin and CTBP2, we also found higher S-cone ribbon counts in both dorsal and ventral retina ( SI Appendix , Fig. S2 H ). This difference was not as dramatic as seen in our SBEM, potentially due to the resolution limits of light microscopy. Average total ribbon volume (in 3D reconstructions) and total ribbon length (using light microscopy) were both somewhat smaller in S-cones than in M-cones (~15%, SI Appendix , Fig. S2 D and K ), although the former did not reach significance. Notably, the only S-cone-specific CBCs encountered in this volume match descriptions of the ON SCBC ( Fig. 1 F ). All other CBCs contacting S-cones also projected to nearby M-cones with larger and more numerous dendritic processes ( Fig. 1 G and H ). While the lack of information about axon terminals in the IPL in this volume precludes definitive identification, cells matching the descriptions of OFF CB1a and CB1b cells [due to their lobule-like dendritic tips ( 41 , 43 )] and CB3 cells [due to their small dendritic fields and tightly clustered dendritic tips ( 34 , 43 )] avoided putative S-cone terminals, reinforcing our judgment of these cells as S-cones ( SI Appendix , Fig. S3 A and B ). Additionally, we found that dendrites resembling ON CB7b cells avoided S-cones ( 34 ), although the expansive dendrites belonging to a putative CB7a cell sparsely contacted cones including a single S-cone ( SI Appendix , Fig. S3 C ). In summary, SBEM reconstructions from 13-LGS retina identified multiple S-cone-specific CBC wiring motifs, but we found no evidence for the presence of an OFF polarity SCBC. To confirm the S-cone selectivity of SCBCs, we also recorded light responses in bipolar cells in 13-LGS retinal slices ( Fig. 2 ) using sinusoidal antiphase blue and green light as S-cone isolating stimuli ( 50 , 51 ). Fig. 2 depicts four representative examples with clear anatomical ( Fig. 2 A – D ) and physiological ( Fig. 2 E – H ) characteristics. SCBCs were clearly identified by their expected morphology ( Fig. 2 A and B ), and their ON polarity responses to blue light flicker could not be silenced by antiphase green flicker ( Fig. 2 E ). In contrast, two examples are presented of non-S-cone-selective CBCs ( Fig. 2 C and D ) whose response to a flickering blue LED was readily silenced by calibrated antiphase green flicker ( Fig. 2 F and G ). Due to their response characteristics and morphology, we classify these neurons as an ON CB7a ( Fig. 2 C and F ) and an OFF CB3b ( Fig. 2 D and G ). Over the course of approximately 80 such recordings, 7 cells were verified as SCBCs using S-cone isolating stimuli, and one additional SCBC was identified anatomically but did not have light responses during recordings. Notably, we did not record from any bipolar cell matching the description of an OFF SCBC. Fig. 2. Open in a new tab S-cone bipolar cell recordings in 13-LGS retina. ( A ). Neurobiotin injection of a SCBC contacting a labeled S-cone (recording in panel E ). Note the axonal lamination in the inner IPL. Dashed white lines indicate the approximate boundaries of the OPL and IPL. (All scale bars, 10 µm.) ( B ). A second SCBC. ( C ). An ON CB7a cell. Note the large dendritic arborization and axonal lamination in the inner IPL. ( D ). An OFF CB3b cell. Note the bistratification of the axon terminal in the outer IPL. ( E ). Responses of the SCBCs in A and B to S-cone isolating stimuli. Flicker stimuli depicted below panel G . ( F ). Responses of the CB7a cell in C . ( G ). Responses of the CB3b cell in D . We also searched for S-cone-selective OFF signals in the 13-LGS using electroretinography (ERG). Using prolonged light flashes, the ERG waveform can be separated into a-, b-, and d-wave components ( 52 ), the latter of which is believed to primarily reflect the activity OFF bipolar cells ( 53 , 54 ). To isolate an S-cone-selective OFF CBC component of the ERG, as for single-cell recordings, we calibrated a flickering green LED (31.25 Hz) to determine the luminance ratio that minimized flicker responses to antiphase blue LED light ( Fig. 3 A ), reflecting equiluminant activation of M-cones. A green–blue luminance ratio of 2.8:1 reliably minimized the amplitude of the ERG flicker response ( Fig. 3 B ). Here, we report human photopic luminous efficiency values per the ERG equipment used; see the Materials and Methods for a discussion of 13-LGS spectral sensitivity. ERG responses to brief blue LED presentations on this calibrated green LED background (interrupted during blue light presentation) featured reduced a- and b-waves, due presumably to M-cone cancellation and ON SCBC activation, respectively, and notably, an abolished d-wave ( Fig. 3 C ). Increasing absolute intensity while preserving this 2.8:1 green:blue luminance ratio produced higher b-wave amplitudes but no change in d-wave amplitudes ( Fig. 3 D ). Put together, these data further support the absence of S-cone-specific OFF CBCs in the 13-LGS retina. Fig. 3. Open in a new tab No S-cone selective OFF bipolar responses in 13-LGS ERG. ( A ). S-cone isolating stimulus ERG calibration. For fixed flickering blue light (31.25 Hz), antiphase flickering green light of increasing intensities was used to minimize the response amplitude. Luminance values refer to the human photopic luminous efficiency curve ( Materials and Methods ). ( B ). Amplitude of the 31.25 Hz Fourier response component as a function of green light intensity. Silencing was achieved at a luminance ratio of 2.8:1. Error bars indicate S.E.M. (n = 3 animals). ( C ). Example responses to 1-s blue or green LED stimuli as indicated. The calibrated S-cone isolating stimulus ( Bottom ) produced no d-wave. ( D ). b- and d-wave response amplitudes to 1-s silent substitution stimuli for the calibrated blue and green LED luminance values shown (n = 3). S-Cone Selectivity of 13-LGS Horizontal Cells. In our SBEM volume, HCs were readily identified by their dendritic tips with lateral elements that flanked cone ribbons ( Fig. 1 C and D ). Two reconstructed HC processes exclusively contacted S-cones ( Fig. 4 A ); we refer to these two dendrites as H2 cells, which have been previously suspected to be specific for S-cones, while ultrastructural evidence has been lacking ( 39 ). These dendrites took the form of long thick processes extending beyond the limits of the volume in both directions, precluding definitive determination of whether both branches arise from different H2 cells or from the same cell (although dendrites from a given H2 appeared to avoid crossings, see below). Each H2 branch only extended processes toward the outer retina to make multiple contacts with each S-cone ( Fig. 4 A – D ), ignoring multiple nearby M-cones. These terminals thoroughly innervated each S-cone contacted, occupying locations near many (although not all) ribbons. In one S-cone, each H2 cell innervated approximately half of the terminal ( Fig. 4 C ), whereas a single H2 occupied sites throughout the other S-cone terminal ( Fig. 4 D ). Nearby H1 cells, clearly identified due to their distinct morphology ( 38 , 55 ), contacted all cone terminals within their dendritic field including S-cones ( Fig. 4 E ). We observed that of all the S-cone ribbon triads in the volume, no H2–H2 pairs were found; only H1–H2 or H1–H1 pairs were found flanking each ribbon and H1–H2 pairs were more prevalent (58 and 70%, respectively). Of the H1–H1 pairs, each individual H1 terminal arose from a different H1 cell, consistent with previous reports ( 28 ). Thus, we also cannot rule out the possibility that the lack of H2–H2 triads results from each H2 branch arising from the same neuron. Fig. 4. Open in a new tab Reconstructed S-cone-specific HC circuitry in the 13-LGS OPL. ( A ). Reconstructed S-cone selective H2 cell branches, both of which pass beyond the volume boundary. ( B ). EM slice through S-cone i highlighting HC dendrites ( Fig. 1 D ). Yellow: H1; Blue: H2; Red: Ribbons. ( C ). Detailed view of HC contacts with S-cone i. Left : H1 (yellow), H2 (blue), and SCBC (gray); Center: H2 and SCBC; Right : H2 only. ( D ). Detailed view of HC contacts with S-cone ii. ( E ). Whole mount view of HCs contacting S-cones. Yellow/orange: H1 cells; Blue/cyan: H2 cells. Insets show detailed views of S-cones i and ii (bracketed) including SCBCs (gray). We also evaluated S-cone-specific connectivity using antibody staining in 13-LGS retinal explants. As shown previously ( 39 , 42 ), we used antibodies to GluA4 receptor subunits (AMPA receptors) and GluK1 (kainate receptors) to label cone terminals and specifically identify S-cone terminals, which strongly stain for GluA4 but weakly or not at all for GluK1 ( Fig. 5 A and B ). We also found that a particular lot of antibodies to calbindin serendipitously differentially labeled HCs in 13-LGS retina, strongly labeling H1-like cells in dorsal retina and H2-like cells in ventral retina. Such putative H2 cells exclusively contacted cone terminals matching the above description of S-cones ( Fig. 5 A ), while putative H1 cells contacted all cones ( Fig. 5 B ). We also presented S-cone isolating stimuli to HCs recorded from horizontally sectioned retinal samples ( 56 ). Neurobiotin-injected HCs fell into two categories: A few cells had large dendritic fields with sparse terminals in the OPL, presumably H2 cells ( Fig. 5 C ). Most cells had smaller dendritic fields with dense dendritic terminals and a long axon, likely H1 cells ( Fig. 5 D ). In both cases, neurobiotin passed into other HCs of the same type (H1–H1 or H2–H2), presumably through gap junction coupling ( 28 ). In these recordings, HCs also exhibited two distinct responses to S-cone isolating light stimuli: Presumptive H2 cells displayed an OFF polarity response to blue light but no response to green LED light ( Fig. 5 E ). Apparent H1 cells displayed responses to blue light flicker that was readily silenced by antiphase green light flicker ( Fig. 5 F ). Due to the high difficulty of these recordings, we are unable to provide examples of recordings and dye fills from the same cell; however, due to their OFF response polarity, blue light response specificity, and absence of OFF SCBCs recorded in vertical retinal slices, it is unlikely that recorded cells identified as H2 cells are mislabeled ON or OFF SCBCs. Of approximately 39 such recordings, 5 cells were identified as H2 cells using S-cone isolating stimuli but were not recovered for imaging; 3 additional H2 cells without light responses were dye-filled and verified anatomically. Put together, the anatomical and physiological evidence strongly supports the finding that the 13-LGS retina has two HC types: an H1 cell that receives achromatic input from cones and an H2 cell that receives exclusively S-cone input. Fig. 5. Open in a new tab Differential S-cone selectivity by 13-LGS HCs. ( A ). Immunolabeled H2 cells selectively contacted S-cones. GluA4 antibodies label all cone terminals, while GluK1 antibodies only weakly label S-cones (Arrowheads). This lot of calbindin antibodies sparsely labeled H2 cells strongly in the ventral 13-LGS retina. ( B ). Nonselective contacts by H1 cells (strongly labeled with calbindin antibody in the ventral 13-LGS retina) with all cone terminals. Putative S-cones are marked with arrowheads as in A . ( C ). Neurobiotin injected H2 cell during patch clamp recordings. Note the weak transfer of neurobiotin to a neighboring H2 cell, presumably through H2–H2 gap junctions. ( D ). Neurobiotin-injected axon-bearing H1 cell with neurobiotin transfer to neighboring H1 cells through H1–H1 gap junctions. ( E ). Silent substitution stimulus-elicited S-cone selective voltage clamp recordings in a putative H2 cell. See Fig. 2 for details. ( F ). Voltage clamp recordings from two putative H1 cells whose responses to S-cone isolating stimuli were silenced by antiphase green light. Our SBEM volume allowed us to observe the cone subtype selectivity of H1 cell axons, which has not been explored in the 13-LGS retina ( Fig. 6 ). In mice, HC axon terminals contact rod spherules ( 28 ), but it remains unclear what photoreceptor terminals are contacted by 13-LGS H1 axons, although it has been reported that they contact cones ( 38 ). The features of H1 axons are difficult to observe due to the density of H1 cells in the retina, the inability to easily distinguish axonal contacts from dendritic contacts anatomically using antibody labeling ( Fig. 5 B ), and the difficulty of studying their electrical properties using patch clamp recording ( 28 , 55 ). Fig. 6 A shows an exceptional example of an H1 with an intact axon obtained during recordings. Such processes projected hundreds of micrometers in seemingly random directions as reported previously ( 38 , 55 ), periodically extending fine collateral processes that presumably contacted photoreceptor terminals ( Fig. 6 B ). In the 13-LGS SBEM volume, we also reconstructed several processes matching this description, putative H1 axonal segments, which passed through the block ( Fig. 6 C ) and projected multiple such collaterals that ended in normal lateral elements within cone terminals ( Fig. 6 D – F ). The morphology of these axonal segments resembles that of the identified H2 segments in the SBEM volume with two exceptions: First, they commonly contacted M-cones; Second, whereas H2 segments extended multiple processes toward each S-cone terminal and made contacts with several ribbons, putative H1 axons typically extended a single thin collateral toward each M-cone that contacted 2 to 3 ribbons near the peripheral edge of its terminal ( Fig. 6 D ). These axonal processes were also distinguished from proximal H1 cell dendrites, which extended from the cell body, branched multiple times, and formed several contact sites within a cone terminal ( Fig. 4 E ). Interestingly, many collaterals projecting from these axon segments had varicosities that appeared to partly wrap or engulf the dendrites of other neurons en route to a cone terminal ( Fig. 6 E ). Because we did not reconstruct all such axonal processes due to their high density (see below), it is unknown what portion of these contacts formed such varicosities; however, the dendrites wrapped by these processes were usually bipolar cells. H1 axon segments, which were easily identified by their thick ~1 μm-diameter processes projecting laterally through the OPL ( Fig. 6 F ), were surprisingly numerous and dense in the SBEM volume, precluding their complete manual volumetric reconstruction. Instead, we performed skeletonized reconstructions of H1 axons in the volume, yielding 85 total axonal segments, none of which contacted S-cone terminals ( Fig. 6 G ). The M-cones in the volume, in contrast, were densely innervated by H1 axon collaterals (contacted by 6.9 ± 2.1 collaterals per cone; mean ± s.d.), which gave rise to one or more lateral elements, in total flanking more than one-third of M-cone ribbons on average (39.5 ± 15.5%, mean ± s.d.; min 16.0%, max 78.9%; Fig. 6 H ). We never observed an instance where a single axon collateral provided both flanking elements for a single M-cone ribbon; however, it was common to find both sides of a ribbon to be occupied by such axon collaterals. The number of axon-flanking ribbons was uncorrelated with total ribbon count ( Fig. 6 H ), leading to no obvious conclusion regarding the contributions of H1 axon signaling to overall cone synaptic signaling due to proportions alone. We corroborated the finding of S-cone avoidance by H1 axons using lipophilic DiI crystals inserted into 13-LGS retinas ( SI Appendix , Fig. S4 ): Labeled H1 cell bodies contacted all cone terminals within their dendritic arbor ( SI Appendix , Fig. S4 A and B ), whereas labeled H1 axons avoided S-cones, instead extending collaterals toward the periphery of M-cone terminals ( SI Appendix , Fig. S4 B – F ). This avoidance of S-cones by H1 axons is an as-yet unreported feature of wiring in the 13-LGS OPL. Fig. 6. Open in a new tab S-cone terminal avoidance by 13-LGS H1 horizontal cell axons. ( A ). Recorded H1 cell with an intact axon. Bracketed area shown in B . ( B ). Higher magnification of the axon terminal in A . ( C ). Example H1 axon segments from the SBEM volume. S-cone terminals outlined in red; Camera icons (d and e) depict views shown in panels D and E . ( D ). Axon collateral with a tip flanking an M-cone peripheral ribbon. ( E ). Axon collateral varicosity engulfing CBC dendrites en route to an M-cone terminal (foreground; large arrowhead) and another example in the background (small arrowhead). Multiple CBCs were engulfed by this cell; a single dendrite is shown for each collateral for clarity. ( F ). Cross-section of the SBEM volume highlighting H1 axons in green. An H2 branch (blue, *) highlighted for comparison. Ribbons flanked by H1 axon collateral tips shown in red. ( G ). Skeletons of all H1 axons in the SBEM volume. Collateral tips are highlighted in depth-coded shades of green. Circles indicate S-cone (red) and M-cone (black) terminals. ( H ). Counts of ribbons flanked by H1 axon collaterals vs. total ribbons in each M-cone. Solid and dashed lines indicate the linear slope regression (n.s.) and 95% CI, respectively. Discussion Peculiarities of S-Cones in the 13-LGS Retina. In our SBEM volume taken from the medial dorsal 13-LGS retina, we found only two S-cones, a ratio directly in line with previous reports in both ground squirrel and primate ( 46 – 48 ) and what we observed from S-opsin antibody staining in that region ( SI Appendix , Fig. S1 ). Interestingly, unlike either the enriched ventronasal zone of “true” S-cones present in mice ( 19 ) or the high density of S-cones previously reported in the dorsonasal extent of the California ground squirrel ( 48 ), we found a general uniformity of S-cone proportions across the 13-LGS retina, with only a subtle gradient from 2.4% in the ventronasal 13-LGS retina to 6.7% in its dorsotemporal extent ( SI Appendix , Fig. S1 ). Although we did not distinguish cones from rods in this count owing to the scarcity of the latter in 13-LGS, we see very few rods in the ventral 13-LGS retina even where they are most prominent [<5% of all photoreceptors; also in contrast with California ground squirrel ( 48 )], suggesting that even within the margin of error of this estimate, the reduced S-cone proportion in ventral retina is likely real. Unlike other species reported, largely primates ( 3 , 49 , 57 ), we found that S-cones in the 13-LGS retina have more numerous ribbons than did M-cones; although ribbons in S-cones were somewhat smaller, these two factors did not quite counteract one another ( SI Appendix , Fig. S2 ), leading to a slightly smaller average total ribbon content per cone terminal in S-cones (~15%), which implies a slightly smaller readily releasable pool of vesicles at the active zone. Note that we did not see an obvious difference in ribbon height from the active zone, a factor expected to determine the size of the ribbon-attached pool of vesicles, and in turn, the rate of replenishment of the readily releasable pool ( 58 ). A complete accounting of S-cone synaptic signaling properties in ground squirrel remains to be obtained; however, it has been previously shown in paired cone-CBC recordings that postsynaptic currents have the same kinetics in response to depolarization in either cone type and that the presynaptic calcium current does not differ between the two ( 43 ). Given the somewhat smaller S-cone terminal size that we see here, it is possible that release size is conserved between cone types; however, the situation is complicated by the potential role that presynaptic calcium may play in ribbon size as well as the differential roles of global calcium vs. localized calcium microdomains domains ( 58 ). However, given the considerable differences in postsynaptic partners, it is plausible that the pressures that shape ribbon architecture may also differ between the two cone types, allowing for more fragmentation of ribbons in S-cones. The clearest difference is that while each M-cone is contacted by more than a dozen CBCs, each S-cone primarily services 1-2 SCBC dendritic tips that occupy the majority of the invaginating ribbon sites in the terminal ( Fig. 1 and SI Appendix , Fig. S3 D and E ). Additionally, OFF CBCs in 13-LGS occupy relatively stereotyped locations relative to ribbon positions in the cone terminal ( 41 , 59 ); given that CB1 and CB3 cells avoid S-cones entirely, the organizing principles for S-cone terminal ribbon triad architecture are likely different. Additionally, given the apparent variability of cell type combinations that we see near an M-cone ribbon, we hypothesize that each ribbon triad could constitute a partially separate signaling entity whose properties will be determined by the response properties of the postsynaptic CBC receptors as well as the combination of HC processes present at that site (see also below). Ribbon organization in the S-cone terminal, in contrast, is likely to be defined by more uniform organizational pressures given the reduced variety of SCBC, H1, and H2 processes at most ribbon sites. Anatomy of 13-LGS S-Cone Selective Bipolar Cell Circuitry in the Outer Retina. Here we provide a definitive demonstration of S-cone isolating light responses in single SCBCs in the ground squirrel. This cell was first identified by West ( 37 ), who noted its resemblance to a midget BC. This cell was later proposed to be the blue bipolar cell ( 35 ) due to its resemblance to that cell in macaque ( 60 ). The cell was shown to make functional contacts with S-cones using single-cell recordings ( 43 ), and later, use of a custom HCN4 antibody showed that SCBCs in the ground squirrel indeed make selective contacts with S-cone terminals ( 39 ). In the retina of mammals other than ground squirrel, SCBCs have been identified in the mouse ( 17 ), rat ( 61 ), rabbit ( 62 ), cat ( 63 ), marmoset ( 64 ), and macaque ( 65 ). We reconstructed three separate SCBC dendrites, but due to the limits of the SBEM volume and their apparent trajectories, it seems likely that two of these processes arise from the same cell—previous reports indicate that SCBCs in the ground squirrel tend to prefer 1-to-1 connections with S-cones ( 35 , 43 ), although it has been shown following laser ablation of photoreceptors in 13-LGS that SCBCs will send multiple dendritic branches seeking S-cones and fail to refine their dendritic tree afterward ( 66 ). The functional consequences of convergence and divergence in S-cones to SCBCs remains a topic of future study. In addition to the SCBC, routing through other CBCs constitutes another pathway for S-cone signals to contribute to vision ( Fig. 1 ). In our SBEM volume, S-cones were avoided by OFF CB1a/b cells [previously termed b7 cells ( 34 )], which are easily identified by their central club-like dendritic tip ( 41 , 59 ). Additionally, reconstructed BCs resembling CB3a/b dendrites also avoided S-cone terminals ( Fig. 1 ). These findings are consistent with a previous report in which CB1 and CB3 cells did not respond to S-cone depolarization in paired recordings ( 59 ). ON CB7b reconstructions likewise avoided S-cones; however, a single reconstructed partial CB7a dendritic tree with a large dendritic tree making sparse cone contacts did contact a single S-cone with a small dendritic tip that did not deeply innervate the terminal ( SI Appendix , Fig. S3 ). Evidence for S-cone preferences by other CBC types in this block is more equivocal largely due to inability to definitively determine cell type from our small sample volume and given that CBC subtypes are expected to have dendrites that tile over a spatial extent much larger than our SBEM volume ( 34 ). Types CB2 and CB5 are expected to make contacts with S-cones ( 43 ), and to our knowledge the selectivity of CB6 cells is currently unreported. We saw a mixture of unidentified ON CBC types that made invaginating contacts with S-cones; this population includes OFF CB2 cells due to their unusual invaginating dendritic tips ( 41 ). The dendritic tips of this mixed population of CB2/5/6 cells often did not reach S-cone ribbons in the characteristic ON bipolar cell termination site ( Fig. 1 and SI Appendix , Fig. S3 ) even when occupying those locations in neighboring M-cone terminals, raising interesting questions about the interaction between receptor location, sensitivity, and kinetics for signal integration in those CBC types. Aside from the presence of S-OFF midget bipolar cells in human and macaque ( 3 , 11 , 12 , 16 ), but not marmoset ( 3 , 15 ), the presence of an OFF polarity equivalent to the SCBC (not to be confused with the S-OFF midget bipolar cell) has been observed only in rabbit ( 67 ) and has not been observed in other species including macaque ( 57 ). West ( 37 ) proposed that a bipolar cell he termed B7, featuring a thin dendrite in the OPL and outer IPL axonal terminations, might be the OFF SCBC. Later termed the Type X cell by ( 34 ), this cell has been suggested to be an interplexiform cell. We reconstructed a lone dendritic tree resembling this description that contacted M-cones, perhaps belonging to an interplexiform cell ( 34 ). All other BCs we reconstructed contacted both S- and M-cones or else avoided S-cones, leaving no candidate for an S-OFF cell. Here, S-cone isolating stimuli elicited no evident sign of S-cone-specific OFF bipolar response during ERGs ( Fig. 3 ). In the 13-LGS, the primary source of the S-OFF signal has been proposed to be carried by a sign-inverting SCA ( 44 , 45 ). The presence of multiple contacts from other CBC types onto S-cones in our SBEM volume provides alternate pathways for S-cone signal processing. While we do not have the data to definitively establish these patterns as uniform across the 13-LGS retina, the relative uniformity of S-cone proportions ( SI Appendix , Fig. S1 ) and ribbon counts ( SI Appendix , Fig. S2 ) across the retina lends support to the idea that these motifs represent core features of wiring in the 13-LGS OPL. Isolation of 13-LGS S-Cone Signals in the Outer Retina Via Horizontal Cells. HC wiring specificity in the OPL appears to be somewhat variable across species, confounding a definitive explanation of their role in signal processing, although it is generally assumed that HCs pool signals from cones within their dendritic field and provide negative feedback to individual cone terminals, providing means for local and global signaling under varying circumstances ( 28 ). The 13-LGS H1 cell contacts all cones within its dendritic field, indicating that glutamate release at each cone receives negative feedback based on an integrated signal in the H1 cell derived from all contacted cone terminals, predominantly M-cones, and this configuration is typically considered a luminance-dependent modulation of synaptic throughput ( 68 , 69 ). Our finding that the 13-LGS H2 cell only contacts S-cones indicates that it cannot participate in color opponency, unlike primate Hii cells ( 30 , 69 ). Therefore, this circuit probably increases contrast detection of short wavelength light by enhancing synaptic sensitivity at S-cone terminals through lateral inhibition ( 28 , 70 , 71 ). Given the evolutionary age of the S-cone circuit, it is appealing to consider that such isolation may serve the computation of chromatic cues used to guide the direction of eye growth, a process that is dysregulated in myopia ( 72 ). Such specificity of S-cone signal normalization in the OPL is also likely to play a significant role on non-image-forming ipRGC circuits that receive inputs from SCBCs ( 8 , 10 , 12 ). A Preponderance of Horizontal Cell Axons in the 13-LGS Retina. However, it is interesting to note that in the 13-LGS, while H1 dendrites contact all cone terminals within their dendritic field, and H2 cells seem to contact only S-cones, H1 axon collaterals constitute an extensive network exclusively contacting M-cone terminals ( Fig. 6 and SI Appendix , Fig. S4 ). Collaterals from H1 axons contributed significantly to synaptic circuitry in the OPL: We reconstructed 85 axon skeletons, and M-cones received contacts from an average of 6.9 axon collaterals each, which contributed lateral elements to 39.5% of M-cone ribbons. While the limited scope of the SBEM volume precludes the definitive identification of the cells giving rise to these processes, we find it unlikely either that they represent the outer limits of HC dendritic trees or that they are misidentified H2 cell processes. Although many H1 axon collaterals in the SBEM volume have varicosities below the cone terminal that “ensheathe” bipolar cell dendrites, we do not see any clear signs in EM images, such as mitochondria, vesicles, or synaptic densities, that would indicate that these varicosities possess synaptic machinery, as in mouse and primate ( 73 , 74 ). Seemingly similar “wrapping” by HC axons has been seen in rabbits ( 75 ). The role of HC axons in the retina is poorly understood. Each B-type HC (H1 cells in 13-LGS) contributes only a single axon that makes sparse contacts with cones [putative multiaxonal HCs in tree shrew notwithstanding ( 55 )]; while the rod contacts by HCs were not examined in this study, we presume that H1 axon terminals mediate their lateral inhibition as in other species, a configuration that would be most relevant in the peripheral ventral 13-LGS retina, where rod density reaches its maximum ( 48 ). It is commonly assumed that B-type HC axon terminals are electrically isolated from their dendrites [( 76 , 77 ); although see end-note #8 of Nelson et al. ( 76 ) for a discussion of how this situation may change for HC axons lacking an extensively ramified terminal arborization]; in such a case, M-cones may be seen to be connected by an extensive M-cone selective lateral inhibitory network via processes that are nearly indistinguishable from the H2 processes specifically connecting the S-cone network. Given the extensive gap junction coupling among horizontal cells ( 28 , 78 ), it is plausible that such an extensive network of H1 axons simply expands the effective receptive field for lateral inhibition in the OPL. Axon terminals in axon-bearing HCs in other mammals are gap junction coupled ( 70 , 78 , 79 ); however, the nature of gap junction coupling in TLGS H1 axons, either at the terminal or at the collaterals shown here, remains unknown. It may be that the axon exists essentially electrically isolated from the soma and that short-distance axonal coupling between nearby M-cones along the length of the H1 axon provides a form of local lateral inhibition. Such connections among axons would provide an M-cone-specific channel of lateral feedback, the properties of which being determined by the strength of connectivity and local uniformity of the visual stimulus. The relatively large diameter of the H1 axons shown here (~1 μm) lends credence to the notion that axonal signaling may be predominantly local in nature. It is interesting to note that 40% of M-cone ribbons on average are flanked by reconstructed H1 axon collaterals, compared to 60% of S-cone ribbons that are flanked by H2 processes (recall all S-cone ribbons are also flanked by H1 processes because no H2–H2 triads were found). It may be possible that under different coupling circumstances, as may be modulated by diurnal dopamine concentrations ( 80 ), that H1 dendritic arbors modulate local inhibition to both M- and S-cones while H1 axon collaterals and H2 cells mediate separate inhibition to M- vs. S-cones, respectively, in long-distance or local manner depending on the nature of signal spread in those cells. Single-cell recordings of H1 cells using intracellular calcium indicator dyes will be instrumental for understanding the spread of signaling in these cells and the chromatic tuning of their contribution to feedback in the OPL Spatial, Temporal, and Chromatic Processing in the OPL of a Cone-Dominant Mammal. Put together, signal processing at the first synapse in the OPL of the cone-dominant 13-LGS depends on a complex set of anatomical wiring rules. The S-cone-SCBC synapse is the dominant feature of the S-cone terminal, constituting mostly 1:1 connections with postsynaptic SCBCs with the exception of a minority of 2:1 or 1:2 connections. In addition, S-cones are contacted by a few sparse dendritic tips from ON CBC types as well as OFF CB2 cells—the only OFF CBC in 13-LGS that seems to contact S-cones. Notably, S-cone contacts with putative CB2 and CB5 cells are intriguing given the documented specialization of each cell toward fast signal processing, with CB2 cells expressing AMPA receptors at dendritic tips near release sites despite being an OFF CBC ( 40 , 41 ) and CB5b cells possessing sodium channels that allow spiking ( 81 ). Plausibly, strong responses of the CB2 to S-cone stimulation under circumstances that mask M-cone signaling could constitute a contextually dependent S-OFF pathway. However, dendritic tips from all non-SCBC bipolar cells that contact S-cones in this volume are much smaller and obviously less numerous than those contacting neighboring M-cones ( Fig. 1 and SI Appendix , Fig. S3 ), and responses to S-cone stimulation in CB2 cells have half the amplitude of that seen from M-cone stimulation ( 43 ), leaving the potential significant contribution of S-cone signaling to these pathways uncertain. Follow-up studies measuring the features of CB2 and CB5 responses under conditions of S-cone isolating stimuli will be valuable to assess these possibilities. Differences in receptor kinetics and response nonlinearities may add temporal signal processing to the spatial and chromatic computational capabilities expected from this circuit. S-cones commonly have slower light responses than do longer wavelength-sensitive cones, and S-cones response kinetics are more uniform across background light levels in primate cones ( 82 ). While we did not systematically test the light response kinetics of S- vs. M-cones, suction electrode recordings of 13-LGS S- and M-cones show light responses without markedly different kinetics ( 83 ); however, response kinetics can depend significantly upon experimental conditions ( 84 ). Previously it has been shown that S- and M-cones in California ground squirrels have similar flicker rate sensitivities and the authors proposed that the slower short-wavelength detection capabilities they observed must come from downstream signal processing ( 85 ). Additionally, our ERG flicker responses to S-cone isolating stimuli showed no obvious sign of slowing S-cone responses up to 50 Hz, although residual S-cone signals are likely small during these experiments due to their rarity ( SI Appendix , Fig. S6 ). Because 13-LGS cones are expected to be extremely electrically compact, one might assume uniformity of glutamate release at each ribbon site if voltage-gated calcium channels and other release machinery are otherwise the same; however, HC feedback provides a likely mechanism for differential modulation of release at each ribbon within the terminal ( 28 , 71 ). The arrangement of HC tips thus presents a potential mechanism for selective processing according to the signaling properties of those cells. Here, we found that H1 axon collateral tips show an apparent bias for the periphery of the M-cone terminal ( Fig. 6 ), near tips of putative CB3 cells but far from putative CB1 tips. Although the functional consequences of this arrangement are complicated by the diffusion distance from the cone ribbons as well as the response properties of the expressed glutamate receptors ( 41 ), it seems likely that CB3 responses will more often be modulated according to the properties of H1 axons and that CB1 responses will depend more upon the properties of H1 dendrites. Due to the proximity to ribbons, invaginating OFF CB2 and ON CBC tips are expected to be more locally driven than to pool global glutamate signals within the terminal ( 40 , 41 ); without definitive identification of all CBC types within the SBEM volume, we saw no clear patterns emerge in which specific invaginating CBC types preferred contacts with ribbons modulated by dendritic vs. axonal H1 elements. In S-cones, all triads included H1 elements because no H2–H2 triads were observed; assuming this to be generally true, all S-cone release would be modulated by H1 dendrites and a subset of ribbons would also be modulated by H2 cells. In S-cones, we also saw no obvious patterns concerning CBCs that preferred H1–H2 triads or H1–H1 triads. In both cases, future data collection and larger sample areas will help clarify this picture; it may be that triad pattern preferences are determined at the level of the retinal ganglion cells that collect input from several bipolar cells within their receptive fields. Materials and Methods Animal Care and Use. All experiments and animal care were conducted in accordance with protocols approved by the Animal Care and Use Committee of the NIH. Electrophysiology. Normal retinal slices from ground squirrels were prepared as previously described ( 44 ). Recordings were performed using a potassium gluconate-based internal solution including Neurobiotin for anatomical verification, and Na-HCO 3 -buffered Ames’ medium external solution. Low melting-point agarose-embedded horizontal retinal slices were prepared for horizontal cell recordings using a Leica VT1000 vibratome (Leica Biosystems). Recordings were performed with an Axopatch 200B amplifier (Molecular Devices) controlled by Igor Pro 6 ( https://www.wavemetrics.com/ ) using custom routines (AClamp). Silent substitution stimuli ( 44 ) were presented using antiphase sinusoidally modulated (1 Hz) blue (465 nm) and green (574 nm) LEDs mounted to an auxiliary video port of the microscope. Sections were taken preferentially from medial dorsal retina to access regions of the retina with lower cell densities and avoid peripheral regions at higher risk of tissue damage. Throughout the course of the experiments resulting in data presented here, recordings were performed in retinas taken from approximately 30 to 35 animals. Immunocytochemistry. Following electrophysiological recording, retinal slices were fixed and processed using standard immunohistochemical techniques as described previously ( 42 ). Cells were injected with neurobiotin and processed for anatomical identification. Retinal whole-mount staining for S-cone density quantification was performed with RPE left intact, using 10% H 2 O 2 to bleach the pigment. S-cone and photoreceptor densities were obtained from selected locations using a semiautomated method combining ImageJ ( 86 ) and CellPose ( 87 ). For ribbon measurements, another whole-mount retina was similarly prepared using antibodies to CTBP2, but without the RPE or bleaching using H 2 O 2 ; 405 nm-excited autofluorescence of photoreceptor inner and outer segments was used to match photoreceptor and synaptic terminal mosaics for S-cone identification and high-resolution imaging of S-cone terminals. Analysis was performed in ImageJ. H1 axon labeling using DiI crystals was performed omitting Triton-X during processing and using extracellularly targeted antibodies to S-opsin as well as PNA to label terminals, after which point DiI crystals were inserted and samples observed for several days until ready for imaging. Electroretinography. Electroretinograms (ERGs) were recorded on light-adapted anesthetized 13-LGS using an Espion E2 system (Diagnosys, Lowell, MA). Silent substitution stimuli were produced using antiphase sinusoidal blue (461 nm) and green (517 nm) LEDs at 31.25 Hz. Luminance values in the main text refer to the human photopic luminance function as calibrated by the Espion E2 system; spectral measurements at this luminance values were used to confirm matched activation of M-cones using calculations from published ocular transmittance and cone activation curves ( SI Appendix , Fig. S5 ). Data analysis was performed in MATLAB. Serial Block-Face EM Reconstructions. Volume EM data were acquired from the central dorsal retina of an approximately 1-y-old nonhibernating adult male 13-LGS by Renovo Neural, Inc. using a 3View imaging system (Gatan, Inc.). Volume size was approximately 25 × 20 × 50 μm 3 , with imaging resolution of 7 nm and slice thickness of 50 nm. Image alignment was performed using the Etomo utility of IMOD ( https://bio3d.colorado.edu/imod/ ), and manual volumetric reconstructions were performed in IMOD and Reconstruct ( https://synapseweb.clm.utexas.edu/ ). Skeleton reconstructions of horizontal cell axons were performed in Knossos ( https://knossos.app/ ). Further visualization and analysis were performed using Blender 3D ( https://www.blender.org/ ). Statistics. Statistical tests were performed using GraphPad Prism (v.9.5.1; GraphPad Software, LLC). Significance indicators were used as shown: n.s.: not significant; * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001. Supplementary Material Appendix 01 (PDF) pnas.2504954122.sapp.pdf (19.9MB, pdf) Acknowledgments This research was supported by the Intramural Research Program of the NIH. The contributions of the NIH author(s) were made as part of their official duties as NIH federal employees, are in compliance with agency policy requirements, and are considered Works of the United States Government. However, the findings and conclusions presented in this paper are those of the author(s) and do not necessarily reflect the views of the NIH or the U.S. Department of Health and Human Services. We would like to thank Steve DeVries (Northwestern University) and Juan Angueyra (University of Maryland) for constructive criticism and suggestions on the manuscript. Author contributions S.C., H.Q., W.L., and J.M.B. designed research; Y.Z., S.C., H.Q., W.L., and J.M.B. performed research; Y.Z., S.C., H.Q., W.L., and J.M.B. analyzed data; W.L. conception, supervision, editing; and Y.Z. and J.M.B. wrote the paper. Competing interests The authors declare no competing interest. Footnotes This article is a PNAS Direct Submission. Contributor Information Wei Li, Email: liweivision@sjtu.edu.cn. John M. Ball, Email: john.ball2@nih.gov. Data, Materials, and Software Availability Reconstructions, imaging, recording data have been deposited in Zenodo ( 10.5281/zenodo.17187000 ) ( 88 ). All other data are included in the manuscript and/or SI Appendix . Supporting Information References 1. Hunt D. M., Peichl L., S cones: Evolution, retinal distribution, development, and spectral sensitivity. Vis. Neurosci. 31, 115–138 (2014). [ DOI ] [ PubMed ] [ Google Scholar ] 2. Calkins D. J., Seeing with S cones. Prog. Retin. Eye Res. 20, 255–287 (2001). [ DOI ] [ PubMed ] [ Google Scholar ] 3. Kim Y. J., et al. , Comparative connectomics reveals noncanonical wiring for color vision in human foveal retina. Proc. Natl. Acad. Sci. U.S.A. 120, e2300545120 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 4. Neitz J., Neitz M., Evolution of the circuitry for conscious color vision in primates. Eye 31, 286–300 (2017). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Terasaki H., et al. , Blue-on-yellow perimetry in the complete type of congenital stationary night blindness. Invest. Ophthalmol. Vis. Sci. 40, 2761–2764 (1999). [ PubMed ] [ Google Scholar ] 6. Gawne T. J., She Z., Norton T. T., Chromatically simulated myopic blur counteracts a myopiagenic environment. Exp. Eye Res. 222, 109187 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Dolgin E., The myopia boom. Nature 519, 276–278 (2015). [ DOI ] [ PubMed ] [ Google Scholar ] 8. Dacey D. M., et al. , Melanopsin-expressing ganglion cells in primate retina signal colour and irradiance and project to the LGN. Nature 433, 749–754 (2005). [ DOI ] [ PubMed ] [ Google Scholar ] 9. Patterson S. S., Kuchenbecker J. A., Anderson J. R., Neitz M., Neitz J., A color vision circuit for non-image-forming vision in the primate retina. Curr. Biol. 30, 1269–1274.e2 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. Do M. T. H., Melanopsin and the intrinsically photosensitive retinal ganglion cells: Biophysics to behavior. Neuron 104, 205–226 (2019). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 11. Klug K., Herr S., Ngo I. T., Sterling P., Schein S., Macaque retina contains an S-Cone OFF midget pathway. J. Neurosci. 23, 9881–9887 (2003). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Patterson S. S., Kuchenbecker J. A., Anderson J. R., Neitz M., Neitz J., A color vision circuit for non-image-forming vision in the primate retina. Curr. Biol. 30, 1269–1274.e2 (2020). 10.1016/j.cub.2020.01.040. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Patterson S. S., Neitz M., Neitz J., Reconciling color vision models with midget ganglion cell receptive fields. Front. Neurosci. 13, 865 (2019). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 14. Zhang C., et al. , Circuit reorganization shapes the developing human foveal midget connectome toward single-cone resolution. Neuron 108, 905–918.e3 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 15. Lee S. C. S., Telkes I., Grünert U., S-cones do not contribute to the OFF-midget pathway in the retina of the marmoset. Callithrix jacchus. Eur. J. Neurosci. 22, 437–447 (2005). [ DOI ] [ PubMed ] [ Google Scholar ] 16. Wool L. E., Packer O. S., Zaidi Q., Dacey D. M., Connectomic identification and three-dimensional color tuning of S-OFF midget ganglion cells in the primate retina. J. Neurosci. 39, 7893–7909 (2019). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 17. Haverkamp S., et al. , The primordial, blue-cone color system of the mouse retina. J. Neurosci. 25, 5438–5445 (2005). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Baden T., Ancestral photoreceptor diversity as the basis of visual behaviour. Nat. Ecol. Evol. 8, 374–386 (2024). [ DOI ] [ PubMed ] [ Google Scholar ] 19. Nadal-Nicolás F. M., et al. , True S-cones are concentrated in the ventral mouse retina and wired for color detection in the upper visual field. eLife 9, e56840 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 20. Franke K., et al. , Asymmetric distribution of color-opponent response types across mouse visual cortex supports superior color vision in the sky. eLife 12, RP89996 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Patterson S. S., et al. , An S-cone circuit for edge detection in the primate retina. Sci. Rep. 9, 11913 (2019). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Yoshimatsu T., Schröder C., Nevala N. E., Berens P., Baden T., Fovea-like photoreceptor specializations underlie single UV cone driven prey-capture behavior in zebrafish. Neuron 107, 320–337.e6 (2020). 10.1016/j.neuron.2020.04.021. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 23. Angueyra J., Li W., “1.31 - Subcortical color pathways in mammals” in The Senses: A Comprehensive Reference, Fritzsch B., Ed. (Elsevier, 2020), pp. 508–523. [ Google Scholar ] 24. Baden T., Osorio D., The retinal basis of vertebrate color vision. Annu. Rev. Vis. Sci. 5, 177–200 (2019). [ DOI ] [ PubMed ] [ Google Scholar ] 25. Haverkamp S., Grünert U., Wässle H., The cone pedicle, a complex synapse in the retina. Neuron 27, 85–95 (2000). [ DOI ] [ PubMed ] [ Google Scholar ] 26. Dacey D. M., Lee B. B., The “blue-on” opponent pathway in primate retina originates from a distinct bistratified ganglion cell type. Nature 367, 731–735 (1994). [ DOI ] [ PubMed ] [ Google Scholar ] 27. Kim Y. J., Packer O., Dacey D. M., A circuit motif for color in the human foveal retina. Proc. Natl. Acad. Sci. U.S.A. 121, e2405138121 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 28. Chapot C. A., Euler T., Schubert T., How do horizontal cells ‘talk’ to cone photoreceptors? Different levels of complexity at the cone–horizontal cell synapse. J. Physiol. 595, 5495–5506 (2017). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 29. Crook J. D., et al. , Parallel ON and OFF cone bipolar inputs establish spatially coextensive receptive field structure of blue-yellow ganglion cells in primate retina. J. Neurosci. 29, 8372–8387 (2009). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 30. Packer O. S., Verweij J., Li P. H., Schnapf J. L., Dacey D. M., Blue-yellow opponency in primate S cone photoreceptors. J. Neurosci. 30, 568–572 (2010). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 31. Peichl L., Sandmann D., Boycott B. B., “Comparative anatomy and function of mammalian horizontal cells” in Development and Organization of the Retina: From Molecules to Function, Chalupa L. M., Finlay B. L., Eds. (Springer, US, 1998), pp. 147–172. [ Google Scholar ] 32. Li W., Ground squirrel – A cool model for a bright vision. Semin. Cell Dev. Biol. 106, 127–134 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. Cuenca N., et al. , The neurons of the ground squirrel retina as revealed by immunostains for calcium binding proteins and neurotransmitters. J. Neurocytol. 31, 649–666 (2002). [ DOI ] [ PubMed ] [ Google Scholar ] 34. Light A. C., et al. , Organizational motifs for ground squirrel cone bipolar cells. J. Comp. Neurol. 520, 2864–2887 (2012). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 35. Linberg K. A., Suemune S., Fisher S. K., Retinal neurons of the California ground squirrel, Spermophilus beecheyi: A Golgi study. J. Comp. Neurol. 365, 173–216 (1996). [ DOI ] [ PubMed ] [ Google Scholar ] 36. Puller C., Ondreka K., Haverkamp S., Bipolar cells of the ground squirrel retina. J. Comp. Neurol. 519, 759–774 (2011). [ DOI ] [ PubMed ] [ Google Scholar ] 37. West R. W., Light and electron microscopy of the ground squirrel retina: Functional considerations. J. Comp. Neurol. 168, 355–377 (1976). [ DOI ] [ PubMed ] [ Google Scholar ] 38. West R. W., Dowling J. E., Anatomical evidence for cone and rod-like receptors in the gray squirrel, ground squirrel, and prairie dog retinas. J. Comp. Neurol. 159, 439–459 (1975). [ DOI ] [ PubMed ] [ Google Scholar ] 39. Puller C., Haverkamp S., Bipolar cell pathways for color vision in non-primate dichromats. Vis. Neurosci. 28, 51–60 (2011). [ DOI ] [ PubMed ] [ Google Scholar ] 40. DeVries S. H., Bipolar cells use kainate and AMPA receptors to filter visual information into separate channels. Neuron 28, 847–856 (2000). [ DOI ] [ PubMed ] [ Google Scholar ] 41. Grabner C. P., et al. , Mechanisms of simultaneous linear and nonlinear computations at the mammalian cone photoreceptor synapse. Nat. Commun. 14, 3486 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 42. Li W., DeVries S. H., Separate blue and green cone networks in the mammalian retina. Nat. Neurosci. 7, 751–756 (2004). [ DOI ] [ PubMed ] [ Google Scholar ] 43. Li W., DeVries S. H., Bipolar cell pathways for color and luminance vision in a dichromatic mammalian retina. Nat. Neurosci. 9, 669–675 (2006). [ DOI ] [ PubMed ] [ Google Scholar ] 44. Chen S., Li W., A color-coding amacrine cell may provide a blue-Off signal in a mammalian retina. Nat. Neurosci. 15, 954–956 (2012). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 45. Sher A., DeVries S. H., A non-canonical pathway for mammalian blue-green color vision. Nat. Neurosci. 15, 952–953 (2012). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 46. Ahnelt P. K., Characterization of the color related receptor mosaic in the ground squirrel retina. Vision Res. 25, 1557–1567 (1985). [ DOI ] [ PubMed ] [ Google Scholar ] 47. Curcio C. A., et al. , Distribution and morphology of human cone photoreceptors stained with anti-blue opsin. J. Comp. Neurol. 312, 610–624 (1991). [ DOI ] [ PubMed ] [ Google Scholar ] 48. Kryger Z., Galli-Resta L., Jacobs G. H., Reese B. E., The topography of rod and cone photoreceptors in the retina of the ground squirrel. Vis. Neurosci. 15, 685–691 (1998). [ DOI ] [ PubMed ] [ Google Scholar ] 49. Kolb H., Goede P., Roberts S., McDermott R., Gouras P., Uniqueness of the S-cone pedicle in the human retina and consequences for color processing. J. Comp. Neurol. 386, 443–460 (1997). [ DOI ] [ PubMed ] [ Google Scholar ] 50. Donner K. O., W. a. H. Rushton, Retinal stimulation by light substitution. J. Physiol. 149, 288–302 (1959). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 51. Estévez O., Spekreijse H., The “silent substitution” method in visual research. Vision Res. 22, 681–691 (1982). [ DOI ] [ PubMed ] [ Google Scholar ] 52. Twig G., et al. , A novel turtle retinal preparation for simultaneously measuring light-induced electrical activity and changes in metabolite levels. Biol. Bull. 203, 198–200 (2002). [ DOI ] [ PubMed ] [ Google Scholar ] 53. Sieving P. A., Murayama K., Naarendorp F., Push–pull model of the primate photopic electroretinogram: A role for hyperpolarizing neurons in shaping the b-wave. Vis. Neurosci. 11, 519–532 (1994). [ DOI ] [ PubMed ] [ Google Scholar ] 54. Xu X., Karwoski C., Current source density analysis of the electroretinographic d wave of frog retina. J. Neurophysiol. 73, 2459–2469 (1995). [ DOI ] [ PubMed ] [ Google Scholar ] 55. Mariani A. P., Multiaxonal horizontal cells in the retina of the tree shrew. Tupaia glis. J. Comp. Neurol. 233, 553–563 (1985). [ DOI ] [ PubMed ] [ Google Scholar ] 56. Feigenspan A., Babai N. Z., Preparation of horizontal slices of adult mouse retina for electrophysiological studies. J. Vis. Exp. 119, e55173 (2017), 10.3791/55173. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 57. Herr S., Klug K., Sterling P., Schein S., Inner s-cone bipolar cells provide all of the central elements for S cones in macaque retina. J. Comp. Neurol. 457, 185–201 (2003). [ DOI ] [ PubMed ] [ Google Scholar ] 58. Pangrsic T., Singer J. H., Koschak A., Voltage-gated calcium channels: Key Players in sensory coding in the retina and the inner ear. Physiol. Rev. 98, 2063–2096 (2018). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 59. DeVries S. H., Li W., Saszik S., Parallel processing in two transmitter microenvironments at the cone photoreceptor synapse. Neuron 50, 735–748 (2006). [ DOI ] [ PubMed ] [ Google Scholar ] 60. Mariani A. P., Bipolar cells in monkey retina selective for the cones likely to be blue-sensitive. Nature 308, 184–186 (1984). [ DOI ] [ PubMed ] [ Google Scholar ] 61. Euler T., Wässle H., Immunocytochemical identification of cone bipolar cells in the rat retina. J. Comp. Neurol. 361, 461–478 (1995). [ DOI ] [ PubMed ] [ Google Scholar ] 62. Famiglietti E. V., Functional architecture of cone bipolar cells in mammalian retina. Vision Res. 21, 1559–1563 (1981). [ DOI ] [ PubMed ] [ Google Scholar ] 63. Cohen E., Sterling P., Boycott B. B., Convergence and divergence of cones onto bipolar cells in the central area of cat retina. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 330, 323–328 (1997). [ DOI ] [ PubMed ] [ Google Scholar ] 64. Ghosh K. K., Martin P. R., Grünert U., Morphological analysis of the blue cone pathway in the retina of a New World monkey, the marmoset Callithrix jacchus. J. Comp. Neurol. 379, 211–225 (1997). [ PubMed ] [ Google Scholar ] 65. Kouyama N., Marshak D. W., Bipolar cells specific for blue cones in the macaque retina. J. Neurosci. 12, 1233–1252 (1992). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 66. Beier C., Palanker D., Sher A., Stereotyped synaptic connectivity is restored during circuit repair in the adult mammalian retina. Curr. Biol. 28, 1818–1824.e2 (2018). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 67. Liu P.-C., Chiao C.-C., Morphologic identification of the OFF-type blue cone bipolar cell in the rabbit retina. Invest. Ophthalmol. Vis. Sci. 48, 3388–3395 (2007). [ DOI ] [ PubMed ] [ Google Scholar ] 68. Boycott B. B., Hopkins J. M., Sperling H. G., Cone connections of the horizontal cells of the rhesus monkey’s retina. Proc. R. Soc. Lond. B Biol. Sci. 229, 345–379 (1987). [ DOI ] [ PubMed ] [ Google Scholar ] 69. Dacey D. M., Lee B. B., Stafford D. K., Pokorny J., Smith V. C., Horizontal cells of the primate retina: Cone specificity without spectral opponency. Science 271, 656–659 (1996). [ DOI ] [ PubMed ] [ Google Scholar ] 70. Chapot C. A., et al. , Local signals in mouse horizontal cell dendrites. Curr. Biol. 27, 3603–3615.e5 (2017). [ DOI ] [ PubMed ] [ Google Scholar ] 71. Kamermans M., Spekreijse H., The feedback pathway from horizontal cells to cones: A mini review with a look ahead. Vision Res. 39, 2449–2468 (1999). [ DOI ] [ PubMed ] [ Google Scholar ] 72. Gawne T. J., Grytz R., Norton T. T., How chromatic cues can guide human eye growth to achieve good focus. J. Vis. 21, 11 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 73. Behrens C., et al. , Retinal horizontal cells use different synaptic sites for global feedforward and local feedback signaling. Curr. Biol. 32, 545–558.e5 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 74. Puller C., Haverkamp S., Neitz M., Neitz J., Synaptic elements for GABAergic feed-forward signaling between HII horizontal cells and blue cone bipolar cells are enriched beneath primate S-cones. PLoS One 9, e88963 (2014). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 75. Famiglietti E. V., A new type of wide-field horizontal cell, presumably linked to blue cones, in rabbit retina. Brain Res. 535, 174–179 (1990). [ DOI ] [ PubMed ] [ Google Scholar ] 76. Nelson R., Lützow A. v., Kolb H., Gouras P., Horizontal cells in cat retina with independent dendritic systems. Science 189, 137–139 (1975). [ DOI ] [ PubMed ] [ Google Scholar ] 77. Suzuki H., Pinto L. H., Response properties of horizontal cells in the isolated retina of wild- type and pearl mutant mice. J. Neurosci. 6, 1122–1128 (1986). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 78. Hombach S., et al. , Functional expression of connexin57 in horizontal cells of the mouse retina. Eur. J. Neurosci. 19, 2633–2640 (2004). [ DOI ] [ PubMed ] [ Google Scholar ] 79. Puller C., Müller L. P. S. de, Janssen-Bienhold U., Haverkamp S., ZO-1 and the spatial organization of gap junctions and glutamate receptors in the outer plexiform layer of the mammalian retina. J. Neurosci. 29, 6266–6275 (2009). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 80. Janssen-Bienhold U., et al. , Connexin57 is expressed in dendro-dendritic and axo-axonal gap junctions of mouse horizontal cells and its distribution is modulated by light. J. Comp. Neurol. 513, 363–374 (2009). [ DOI ] [ PubMed ] [ Google Scholar ] 81. Saszik S., DeVries S. H., A Mammalian retinal bipolar cell uses both graded changes in membrane voltage and all-or-nothing Na+ spikes to encode light. J. Neurosci. 32, 297–307 (2012). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 82. Baudin J., Angueyra J. M., Sinha R., Rieke F., S-cone photoreceptors in the primate retina are functionally distinct from L and M cones. eLife 8, e39166 (2019). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 83. Ferguson R., Miyagishima K. J., Nadal-Nicolas F. M., Li W., Characterizing the rod pathway in cone-dominant thirteen-lined ground squirrels. Front. Ophthalmol. 64, 1649 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 84. Azevedo A. W., Rieke F., Experimental protocols alter phototransduction: The implications for retinal processing at visual threshold. J. Neurosci. 31, 3670–3682 (2011). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 85. Crognale M., Jacobs G. H., Temporal properties of the short-wavelength cone mechanism: Comparison of receptor and postreceptor signals in the ground squirrel. Vision Res. 28, 1077–1082 (1988). [ DOI ] [ PubMed ] [ Google Scholar ] 86. Arshadi C., Günther U., Eddison M., Harrington K. I. S., Ferreira T. A., SNT: A unifying toolbox for quantification of neuronal anatomy. Nat. Methods 18, 374–377 (2021). [ DOI ] [ PubMed ] [ Google Scholar ] 87. Stringer C., Wang T., Michaelos M., Pachitariu M., Cellpose: A generalist algorithm for cellular segmentation. Nat. Methods 18, 100–106 (2021). [ DOI ] [ PubMed ] [ Google Scholar ] 88. Zhang Y., Chen S., Qian H., Li W., Ball J., Retinal S-cone specific anatomical and physiological data in the thirteen-lined ground squirrel [Data set]. Zenodo. 10.5281/zenodo.17187000. Deposited 23 September 2025. [ DOI ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Appendix 01 (PDF) pnas.2504954122.sapp.pdf (19.9MB, pdf) Data Availability Statement Reconstructions, imaging, recording data have been deposited in Zenodo ( 10.5281/zenodo.17187000 ) ( 88 ). All other data are included in the manuscript and/or SI Appendix . Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences ACTIONS View on publisher site PDF (11.1 MB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• The Generative Cone Theory: A Structure-First Account of Instructional Sequencing and Its Falsifiable Predictions#477813
• The primate claustrum as a hub for precision weighting of prediction errors across the cortical hierarchy.#449682
• Overcoming current limitations of transcranial ultrasound imaging for preclinical and clinical applications#156059
• Early catarrhine evolution in the light of semicircular canal and vestibule shape.#1158990
• Comparative Morphometric and Structural Analysis of the Ovine Brain: Integrating Traditional Anatomical Methods with Artificial Intelligence-Driven 3D Modeling and Identification.#232426