Abstract
Summary
Visual information processing is sculpted by a diverse group of inhibitory interneurons in the retina called amacrine cells. Yet, for most of the >60 amacrine cell types, molecular identities and specialized functional attributes remain elusive. Here, we developed an intersectional genetic strategy to target a group of wide-field amacrine cells (WACs) in mouse retina that co-express the transcription factor Bhlhe22 and the Kappa Opioid Receptor (KOR; B/K WACs). B/K WACs feature straight, unbranched dendrites spanning over 0.5 mm (∼15° visual angle) and produce non-spiking responses to either light increments or decrements. Two-photon dendritic population imaging reveals Ca2+ signals tuned to the physical orientations of B/K WAC dendrites, signifying a robust structure-function alignment. B/K WACs establish divergent connections with multiple retinal neurons, including unexpected connections with non-orientation-tuned ganglion cells and bipolar cells. Our work sets the stage for future comprehensive investigations of the most enigmatic group of retinal neurons: WACs.
Introduction
Inhibitory interneurons modulate information flow in the brain dependent on their diverse morphologies and synaptic mechanisms. Indeed, the remarkable diversity of inhibitory interneurons spans various brain regions, including the cortex, hippocampus, spinal cord, and retina (Booker & Vida, 2018; He et al., 2016; Sweeney et al., 2018; Yan et al., 2020). A comprehensive understanding of the structure and function of inhibitory interneurons is crucial to decipher the functional dynamics and organizational principles of neural circuits.
In the retina, visual information is first encoded by photoreceptors, which transmit signals to the output neurons, the retinal ganglion cells (RGCs), via multiple, parallel excitatory pathways: photoreceptors → bipolar cells → RGCs (Baden et al., 2016; Bae et al., 2018; Euler et al., 2014; Goetz et al., 2022; Kerschensteiner, 2022; Tran et al., 2019). Signaling in these pathways is modulated and shaped by the major class of inhibitory interneurons, known as amacrine cells, which are the most diverse yet least understood retinal cell class (MacNeil & Masland, 1998; Masland, 2012; Yan et al., 2020). Amacrine cells provide inhibitory synaptic input to bipolar cell terminals, RGC dendrites and other amacrine cells (Demb & Singer, 2015; Diamond, 2017).
Classically, amacrine cell function is understood in the context of lateral inhibition that generates RGC receptive field surrounds (Farrow et al., 2013; Flores-Herr et al., 2001; Werblin, 1972). Amacrine cell functional diversity, however, extends well beyond this classical operation, as demonstrated by specialized computations in various cell types. For example, GABAergic synapses made by starburst amacrine cells are activated preferentially by specific directions of motion (Euler et al., 2002; Jain et al., 2020; Koren et al., 2017), which drives direction-selective (DS) firing in DS-RGCs (Briggman et al., 2011; Lee et al., 2010; Wei, 2018); VGluT3 amacrine cells are sensitive to local motion and provide excitatory input that generates object-motion selectivity in the downstream W3 RGCs (Kim et al., 2015; Krishnaswamy et al., 2015; Lee et al., 2014); S–cone amacrine cells are tuned to blue-ON signals, and their inhibitory output putatively drives blue-OFF responses in downstream RGCs (Chen & Li, 2012; Sher & DeVries, 2012). Therefore, understanding how amacrine cells process visual features is imperative for deciphering retinal computation.
Systematic understanding of the diversity of amacrine cells and their functional roles relies on the ability to identify specific cell types and to interrogate systematically their roles in retinal circuits (Jo et al., 2018; Zhu et al., 2014). Based on dendritic field diameters, amacrine cells can be divided into narrow-field (∼30 − 150 µm), medium-field (∼150 − 500 µm), and wide-field (>500 µm) subclasses (Masland, 2012). Among these, wide-field amacrine cells (WACs) are the least understood subclass. Stochastic labeling of WACs revealed dendrites that extend more than 0.5 mm from the soma, with a diameter of more than 1 mm (30° of visual angle) in mice, making them the largest cells in the retina (Badea & Nathans, 2004; Lin & Masland, 2006; Perez De Sevilla Muller et al., 2007; Zhu et al., 2014). Systematic explorations of their functions and synaptic connectivity, however, have been hindered due to the lack of genetic approaches for targeted functional recording and manipulation. In this study, we introduce an intersectional genetic approach to target a subset of non-spiking WACs featuring straight, unbranched dendrites. This approach enabled us to investigate both the functional properties and synaptic partnerships of these WACs, laying the foundation for future comprehensive studies of the roles these cells play in retinal information processing.
Results
Intersectional genetic approach for studying WACs
Our intersectional genetic approach to identify specific WACs combines three transgenic alleles: Cre and Flpo recombinases driven by two distinct genetic loci, and a reporter of dual (Cre- and Flpo-mediated) recombination (Cai, Kardon, et al., 2016; Jo et al., 2018) (Fig. 1A). In our search for markers of WACs, the transcription factor Bhlhe22 (also known as Bhlhb5) stood out. Bhlhe22 has an essential role in the development of a spectrum of GABAergic amacrine cells as well as type 2 OFF cone bipolar cells (Feng et al., 2006; Huang et al., 2014). Indeed, an existing transcriptomic database suggested that approximately half of GABAergic amacrine cells express mRNAs encoding Bhlhe22 at a relatively high level (Fig. 1B). Given that WACs are primarily GABAergic (Wassle & Boycott, 1991; Zhu et al., 2014), whereas narrow-field cells are primarily glycinergic (Wassle et al., 2009), it is likely that unidentified Bhlhe22-expressing (Bhlhe22+) cells include uncharacterized WAC types. Indeed, the Bhlhe22-Flpo transgenic line, crossed with a Flpo-dependent reporter, labeled type 2 bipolar cells as well as amacrine cell processes that stratified at four levels of the inner plexiform layer (IPL) (Cai, Kardon, et al., 2016); and our pilot studies suggested that labeled amacrine cells include WACs.
To increase the genetic specificity in targeting WACs, we crossed the Bhlhe22-Flpo line with several Cre lines and discovered a compelling intersection with the Kappa Opioid Receptor (KOR)-Cre line, which labels amacrine cell processes that stratify primarily in two layers, including a subset of those labeled in the Bhlhbe22-Flpo line (Cai, Huang, et al., 2016). Indeed, the Bhlhe22-Flpo x KOR-Cre intersection, combined with a dual ReaChR-mCitrine reporter line (Hooks et al., 2015) (Fig. 1A), labeled a group of amacrine cells with processes that stratified primarily at the borders of the IPL (Fig. 1C) and very rarely labeled bipolar cells or RGCs. Furthermore, nearly all amacrine cells expressing the reporter (98%) also were labeled by Bhlhe22 immunofluorescence (144/147 cells; n = 4 retinas) (Fig. 1C-E). For simplicity, we refer to the cells labeled by this intersectional genetic strategy as B/K amacrine cells.
Most (>80%) B/K cells had somas in the inner nuclear layer (INL), although some localized to either the ganglion cell layer (GCL) or the inner plexiform layer (IPL; Fig. 1F). The fluorescence profile of B/K processes, analyzed in stacks of confocal images, peaked in either ON or OFF layers, as defined by the known locations of ON or OFF bipolar cell axon terminals (Borghuis et al., 2013; Franke et al., 2017) (Fig. 1G). Confocal images showed long, straight processes concentrated near the fluorescence peaks in ON or OFF regions of the IPL with sparse labeling in between, near the ON/OFF boundary (Fig. 1H).
B/K amacrine cells have long, unbranched processes and do not fire action potentials
To investigate the morphology and functional properties of single B/K amacrine cells, a cell in either the GCL or INL was targeted for whole-cell recording (Park et al., 2015; Park et al., 2020; Park et al., 2018). The cell was filled with Lucifer Yellow through the pipette, and, after fixation, the dendrites were imaged by confocal microscopy (Fig. 2A, B). Dendrite stratification was normalized to the stratification of the ON and OFF ChAT (choline acetyltransferase) bands, which correspond to two sets of cholinergic starburst amacrine cell dendrites (Fig. 2C) (Manookin et al., 2008; Park et al., 2015; Park et al., 2020; Park et al., 2018). Low-resolution images of B/K WACs were traced to determine the number of extended dendrites, after initial branching within ∼50 µm of the soma (Fig. 2A, B). In many cases, the entire dendrite could not be captured because the Lucifer Yellow signal faded or because of cuts in the tissue. Nevertheless, dendrites were commonly >0.5 mm in length, consistent with the general WAC morphology.
Dendrites of ON cells primarily stratified on the proximal side of the ON ChAT band, and those of OFF cells primarily stratified on the distal side of the OFF ChAT band; although four cells in the sample (1 ON, 3 OFF) stratified between the bands (Fig. 2D). For the entire sample (n = 55 cells), dendrite number versus stratification level showed two prominent clusters: OFF cells with ∼4-6 dendrites and ON cells with ∼7-13 dendrites (Fig. 2D). Stratification was measured within ∼70-100 µm of the soma, but the dendrite retained its initial stratification level in distal dendrites, ∼500 µm from the soma (Fig. 2E). Overall, B/K amacrine cells apparently comprise several WAC types with different stratification levels.
The stratification of amacrine cell dendrites relative to the terminals of ON and OFF bipolar cells in most cases predicts either ON or OFF responses to contrast modulation (Euler et al., 2014)(for exceptions, see (Jo et al., 2023; Park et al., 2020)). Indeed, B/K WACs with dendrites in the OFF layer depolarized during the negative/OFF phase of a contrast-reversing spot stimulus, whereas B/K WACs with dendrites in the ON layer depolarized during the positive/ON phase (Fig. 3A). For both ON and OFF B/K WACs, responses to increasing spot diameters showed increased modulation amplitude over a distance of ∼1 mm (Fig. 3B), consistent with the diameter of their dendritic trees.
We did observe some heterogeneity in the B/K cells: the OFF B/K cells with somas in the GCL had relatively hyperpolarized resting potentials compared to the other groups (Fig. 3C); and OFF B/K WACs in the INL and GCL had similar patterns of response to contrast modulation, with relatively large depolarizing response to preferred contrast, whereas ON B/K cells had relatively large hyperpolarizing responses to non-preferred contrast (Fig. 3D). Notably, none of the recorded cells fired conventional action potentials (spikes). This differs from the response of another class of spiking WACs, which have distinct axonal and dendritic compartments (e.g., nNOS-1 and dopaminergic amacrine cells) (Badea & Nathans, 2004; Greschner et al., 2014; Kim et al., 2022; Lin & Masland, 2006; Murphy-Baum & Taylor, 2015; Newkirk et al., 2013; Park et al., 2020; Park et al., 2018; Stafford & Dacey, 1997; Taylor, 1996).
B/K WACs exhibit orientation-selective dendritic Ca2+ signals
The non-spiking nature of the B/K WACs suggests that they encode information through their dendrites. Further, their unique, highly oriented dendritic architecture evokes parallels with the elongated dendritic fields of a variety of orientation-selective (OS) and direction-selective (DS) neurons (Antinucci et al., 2016; Bloomfield, 1994; Kim et al., 2008; Kim et al., 2014; Nath & Schwartz, 2017; Weiler et al., 2022). To test whether B/K WACs encode orientation or direction information within their dendrites, we expressed the Ca²⁺ sensor GCaMP7s in B/K WACs using a recently developed dual reporter line (Ai195) and performed population dendritic two-photon imaging at multiple depths throughout the IPL while presenting gratings drifting in eight directions of motion. In both the ON and OFF layers, B/K dendrites exhibited substantial responses for two opposite directions of motion that shared the same orientation (Figure 4A). For individual regions of interest (ROIs), we computed both orientation- and direction-selective indices (OSI, DSI; see Methods). For the vast majority of ROIs, the OSI value was significantly larger than the DSI value (Figure 4B). ROIs with large OSI values were found at multiple levels of the IPL (Figure 4C).
Next, we evaluated the relationship between the preferred orientation tuning and the physical orientation of a dendrite. First, to visualize dendrites with common response properties from their overlapping neighbors, we utilized sparse non-negative matrix factorization (sNMF) for spatial footprint extraction from the time-lapse movies (Pnevmatikakis et al., 2016; Wang & Zhang, 2013). This analysis revealed groups of straight, unbranched processes that resemble dendritic segments of filled B/K cells (Figure 2A). Each group of processes exhibited a preferred functional orientation preference that closely aligned with their physical orientations (Figure 4B). We next compared the physical orientation to the functional tuning of orientation-selective ROIs (i.e., those with OSI > 0.4) and found a strong structure-function correlation (Figure 4E).
Collectively, we showed that B/K WAC dendrites encoded stimulus orientations that parallel their physical orientations. In theory, the orientation-selective Ca²⁺ responses could further drive orientation-tuned GABA release onto postsynaptic neurons.
B/K WACs make synapses with non-orientation tuned ganglion cells
B/K WAC dendrites stratify primarily at levels of the IPL that overlap the dendrites of commonly studied wide-field RGCs that do not encode orientation. Indeed, the OFF Delta cell (also known as OFF-Sustained Alpha RGC) (Krieger et al., 2017; Margolis & Detwiler, 2007; Tagawa et al., 1999) stratifies between the OFF ChAT band and the INL, overlapping the primary stratification of B/K WAC dendrites in the OFF layer (Fig. 5A, B). The ON Alpha cell (also known as ON-Sustained Alpha RGC or the M4 melanopsin RGC) (Estevez et al., 2012; Krieger et al., 2017) stratifies on the vitreal side of the ON ChAT band, overlapping the primary stratification of B/K WAC dendrites in the ON layer (Fig. 5E, F). We tested whether B/K WACs make synapses with these RGC types using optogenetics.
After blocking transmission from photoreceptors (Park et al., 2015; Park et al., 2020; Park et al., 2018; Pottackal et al., 2020; Pottackal, Singer, & Demb, 2021; Pottackal, Walsh, et al., 2021), light pulses that activated ReaChR expressed in B/K WACs evoked inhibitory postsynaptic currents (IPSCs) in RGCs (see Methods). IPSCs in OFF Delta RGCs were substantial (133 ± 46 pA; n = 11) and blocked largely by the GABA-A receptor antagonist gabazine (50 µM SR95531) (Fig. 5C); addition of glycine receptor antagonist strychnine (1 µM) eliminated the IPSCs completely (Fig. 5C).
Additionally, the evoked IPSCs increased monotonically with the intensity of light stimulation and ReaChR activation (Fig. 5D). Similar results were observed in recordings from ON Alpha RGCs: IPSCs (231 ± 32 pA; n = 19) were blocked largely by gabazine and completely blocked by addition of strychnine (Fig. 5E-H).
In comparison to the IPSCs recorded in OFF Delta and ON Alpha RGCs, much smaller responses were observed in the OFF-Transient Alpha cell (28 ± 11 pA; n = 8), which stratifies between the ChAT bands and overlaps B/K WAC processes more sparsely (Suppl. Fig. 1A-C), and in DS-RGCs (18 ± 8 pA; n = 4) that co-stratify with the ChAT bands (Suppl. Fig. 1D). Thus, B/K WACs provide particularly robust synaptic input to OFF Delta and ON Alpha RGCs.
B/K WACs make synapses with non-orientation tuned bipolar cells in the OFF layer and implement sign-inverted feedforward inhibition (SIFI)
To probe possible connections between B/K WAC dendrites and bipolar cells, we performed connectomic analyses. We started with an existing dataset, SBEM (scanning block-face electron microscopy) volume k0725 (Ding et al., 2016), in which all of the inhibitory synapses onto an OFF Delta RGC had been annotated and some presynaptic amacrine cells, including WACs, had been traced (Grimes et al., 2022) (Fig. 6A-C). The OFF Delta RGC in k0725 received 1298 inhibitory (non-ribbon) synapses, with ∼23% of these (278) coming from the narrow-field, glycinergic AII amacrine cell (Demb & Singer, 2012). Of the remaining synapses, 150 amacrine cells were reconstructed at least partially (Grimes et al., 2022), and these provided a total of 300 (out of 1020 non-AII) synapses to the OFF Delta RGC. This group comprised three classes: bistratified amacrine cells with dendrites in at the borders of the INL and GCL (Sublaminae S1 and S5) (Fig. 6B); amacrine cells with asymmetric dendritic trees (Fig. 6B); and single WAC neurites that ran unbranched through S1 of the IPL, in the same plane as the dendrites of the OFF Delta RGCs (Fig. 6C).
We quantified properties of the putative B/K WAC neurites that made synapses with OFF Delta RGCs by evaluating synapse density and organization of nearby output synapses from straight, unbranched neurites. There were apparently two groups of WAC neurite with either sparse or dense total number of synapses (combined inputs and outputs) (Fig. 6C). At least some of these straight neurites might belong to B/K WACs and we therefore sought to address this hypothesis to gain information about the synaptic connectivity of these cells at the ultrastructural level.
Additionally, we found that ∼25% of the inhibitory synapses to the OFF Delta RGC arose from an amacrine cell element that also provided an inhibitory synapse to a presynaptic type 2 cone bipolar cell that itself was presynaptic to the OFF Delta RGC; thus, there appeared to be simultaneous pre- and postsynaptic inhibition arising at a substantial number of sites on the OFF Delta RGC dendrite. We term this synaptic motif SIFI, for Sign-Inverting Feedforward Inhibition (Fig. 7C). SIFI is found also in AII synaptic output (140/278; ∼50% of synapses), which is expected given that AIIs provide inhibitory input both to OFF Delta RGCs and to type 2 OFF cone bipolar cells presynaptic to these RGCs (Graydon et al., 2018; Tsukamoto & Omi, 2017). Notably, synapses from other amacrine types also exhibit SIFI (205/1020; ∼20%) with the OFF Delta RGC. By contrast, the SIFI motif was observed infrequently at a reconstructed OFF Alpha RGC, where only ∼8% of AII amacrine cell synapses (20/248) and ∼4% of non-AII amacrine cells (36/916) exhibited SIFI (data not shown). Thus, SIFI seems to be implemented preferentially at synapses onto the OFF Delta RGC as compared to the OFF Alpha RGC and includes synapses made by putative B/K WACs.
Because amacrine cell types cannot be identified from the SBEM volume based on morphologies alone, we performed ultrastructural analysis of a new volume (Suppl. Fig. 2A) in which B/K WACs expressed a genetically-encoded marker for use with SBEM. We targeted the enhanced peroxidase dimeric APEX2 (Joesch et al., 2016; Zhang et al., 2019) to the mitochondrial matrix of B/K WACs by dual Cre/Flp-mediated recombination and visualized genetically-identified neurites by the DAB reaction (Fig. 6D-G). Assessment of APEX2 expression revealed DAB+ mitochondria in varicosities that are presynaptic to presumed RGC dendrites (i.e., processes that lack synaptic vesicles) and OFF cone bipolar cell terminals in S1 (i.e., axon terminals that contained ribbon synapses and vesicles) (Fig. 6G). Furthermore, some of these B/K WAC processes exhibited SIFI-connectivity (Fig. 6G), matching the pattern of straight WAC neurites reconstructed from the k0725 SBEM volume. The labeled B/K WAC dendrites exhibited a relatively high density of synapses, matching the subset of WAC neurites with high synapse density reconstructed from k0725 (Fig. 6E; Suppl. Fig. 2B, C) and include outputs to cone bipolar cells (Fig. 6E, F). The combined analysis of optogenetic and connectomic data suggest that B/K WACs, which carry orientation-tuned signals (Fig. 4), make output synapses to postsynaptic cells that are not themselves orientation selective.
Discussion
We developed an approach for systematic study of an enigmatic group of retinal interneurons: non-spiking wide-field amacrine cells (WACs) with straight, unbranched dendrites that are either ON or OFF cells, consistent with their stratification level in the IPL (Figs. 1-3). Ca2+ imaging showed that B/K WAC dendrites exhibited functional orientation tuning, consistent with their physical orientation (Fig. 4). Optogenetic studies showed that B/K WACs made output synapses onto non-orientation tuned RGC types, driving strong GABAergic inhibitory currents in ON Alpha and OFF Delta RGCs (Fig. 5). Connectomic analyses showed that B/K WACs make synapses with bipolar cell axon terminals, which likewise lack orientation tuning (Fig. 6) (Matsumoto et al., 2021). The nature of encoding visual images with orientation-tuned amacrine cell dendrites suggests a novel organization for WACs. Rather than encoding local subregions via a tiled mosaic of narrow-field cells, WACs apparently overlap substantially with their neighbors and cover orientation space through the collective processing across cells with different dendrite orientations (Fig. 7A). Networks of dendrites that are individually orientation-tuned could deliver a net tuned or non-tuned inhibition to postsynaptic neurons, depending on the pattern of convergence (Fig. 7B). The connections shown here suggest that orientation-tuned inhibition plays an unexpected role in controlling the responses of non-orientation-tuned postsynaptic neurons in the retina (Figs. 5, 6).
WACs form a diverse group of interneuron types in mammalian retina
B/K WACs share some properties with previously described cells visualized with methods for sparse labeling. Two studies identified a total of six WAC types that lacked separate dendritic and axonal compartments and can be compared to the B/K WACs: Cluster 1 and bifid cells (Badea & Nathans, 2004); and WA1, WA2-2, WA3-2 and WA4-3 cells (Lin & Masland, 2006). In comparing the soma location and dendritic tree properties, only bifid cells have properties consistent with those of B/K WACs: relatively asymmetric dendritic trees and mainly in the OFF layer with somas in the INL (Badea & Nathans, 2004). Collectively, these results suggest a diversity of WAC types in the mouse retina that extends beyond the B/K WACs described here.
Additional evidence for WAC diversity comes from both anatomical and physiological studies in mouse and other species. For example, SBEM reconstruction of mouse retina identified two groups of straight, unbranched dendrites that made synapses with one of two types of ON bipolar cell terminals (either Type 7 or Type 5A) (Hanson et al., 2023). Further, calcium signals in Type 5A bipolar cells showed orientation tuning, particularly for vertical orientations (Hanson et al., 2023). The working model for this circuit presumes that straight WAC dendrites produce orientation-tuned signals and convey this information to the Type 5A bipolar cell terminals via gap junctions; and these orientation-tuned signals were apparently further conveyed to postsynaptic DS-RGCs via glutamate synapses. The presumed orientation-tuned signals in WAC dendrites are supported by the direct evidence of OS tuning shown here (Fig. 4). However, the B/K WACs apparently comprise different cell types than those connecting Type 5A bipolar cells, which would presumably stratify proximal to the ON ChAT band and between the ChAT bands (Hanson et al., 2023).
WACs with similar morphology to B/K cells were described as ‘wiry’ in the primate retina (human and monkey) and stratify in at least three layers: ON, OFF and near the ON/OFF boundary (Kolb et al., 1992; Manookin et al., 2015; Mariani, 1990). Similar WACs were identified in rabbit retina as well as in other species (MacNeil et al., 1999). The collected findings suggest that WACs with long, straight dendrites exist across species and across multiple layers of the retina within a species. To the extent that WACs with long, straight dendrites commonly convey orientation-tuned GABA release, this property must be fundamental across mammalian retina. Notably, OS signals are also conveyed by at least one type of spiking, axon-bearing WAC in rabbit (Murphy-Baum & Taylor, 2015), and they are also present within a narrow-field non-spiking glycinergic amacrine cell in mouse (Mani et al., 2023).
Synaptic properties of B/K WACS
B/K WACs appear to be GABAergic based on the optogenetic response in postsynaptic RGCs, which was primarily blocked by a GABA-A receptor antagonist. But, some B/K WACs may have mixed GABA/glycine release (Lu et al., 2008; Sawant et al., 2021) (Fig. 1B), given that residual IPSCs were blocked by the glycine receptor antagonist strychnine. This was not the case in studies of other amacrine cell types performed in other Cre lines (VIP, CRH, ChAT nNOS); in these lines, presumed GABAergic synapses, activated by stimulation of ChR2, were blocked completely by a GABA-A receptor antagonist (Park et al., 2015; Park et al., 2020; Park et al., 2018; Yonehara et al., 2011). An alternate explanation for strychnine-sensitive responses evoked by the optogenetic stimulation of B/K cells is polysynaptic activity, e.g., B/K WACs could disinhibit glycinergic cells or could excite glycinergic cells via gap junctions (Jo et al., 2023). Further studies would be required to investigate these possibilities.
The reconstruction of B/K dendrites showed a prominent pattern of SIFI, which would presumably produce a powerful form of inhibition that hyperpolarizes the RGC dendrite while simultaneously suppressing glutamate release from the presynaptic bipolar cell terminal. The SIFI synaptic motif of B/K WACs resembles the powerful glycinergic inhibition mediated by the AII amacrine cell (Demb & Singer, 2012; Ke et al., 2014), and in both cases it seems to be a specific property of input to the OFF Delta RGC but not to the OFF Alpha RGC.
Possible roles of B/K WACs in retinal information processing
Amacrine cells with strong feature tuning to direction, motion or color are presumed to generate a similar degree of tuning in postsynaptic RGCs (Chen & Li, 2012; Sher & DeVries, 2012). For example, starburst amacrine cell pre-synapses activated by leftward motion would shape the responses of postsynaptic DS RGCs that report rightward motion (Briggman et al., 2011; Wei, 2018). Similar circuits mediate OS tuning in some RGC types in both mouse and zebrafish, where an RGC tuned to one orientation (e.g., horizontal) would receive inhibition tuned to the orthogonal orientation (e.g., vertical) (Antinucci et al., 2016; Nath & Schwartz, 2016; Venkataramani & Taylor, 2016). Electrical coupling to OS-tuned WAC dendrites also can drive OS tuning in postsynaptic bipolar cells and RGCs (Hanson et al., 2023; Nath & Schwartz, 2017).
We have shown here an additional pattern whereby OS-tuned amacrine cells could influence RGC and OFF bipolar cells that are not themselves OS-tuned. Presumably, the postsynaptic RGC or bipolar cell collects inhibitory synapses from B/K WACs of multiple orientations, preventing a strong net OS tuning in the population of connected dendrites and ensuring a lack of OS-tuning in the postsynaptic response (Fig. 7A, B). There may be an advantage to encoding visual images with OS-tuned interneurons (Fig. 7A), for example, since oriented features form core components of natural images (Olshausen & Field, 1996). The oriented dendrites of a B/K WAC could also promote the correlation of signaling across distant points on the retina (McIlwain, 1964; Neuenschwander & Singer, 1996).
There are other cases where tuned neurotransmitter release from presynaptic sites is canceled in the postsynaptic neuron. For example, acetylcholine released from starburst amacrine cells onto RGCs lacks strong DS tuning, because cholinergic transmission is paracrine and converges across multiple angles of direction tuning (Lee et al., 2010; Pottackal et al., 2020; Sethuramanujam et al., 2021). A similar model explains how OS-tuned glycinergic release from narrow-field VGluT3 amacrine cells could produce non-OS-tuned glycinergic inhibition of ON DS-RGCs (Mani et al., 2023).
Limitations of the study and next steps in understanding WAC function
Our genetic approach to labeling B/K WACs has obvious advantages for targeting cells systematically. Still, though, it remains to be understood how many cell types and what fraction of an individual type is identified in any given mouse line. For example, do the B/K WACs that stratify between the ChAT bands represent a rare and sparse cell type, or are we observing sparse labeling of a dense cell type? Our inquiries therefore would benefit from additional functional tests that distinguish B/K WAC types based on Ca2+ imaging or by molecular markers that label soma mosaics.
In the future, we also seek methods for inactivating B/K WAC types. Inactivation studies could evaluate potential roles for B/K WACs in regulating overall excitability of RGCs and shaping OS tuning in specific RGC types. Further, there is a possible role that WACs could play in development of other retinal cell types via temporary connections during postnatal development (Gamlin et al., 2020). By controlling B/K WAC activity more specifically, one could also learn how these cells shape responses that influence activity in the brain and behavior (Cruz-Martin et al., 2014; Hillier et al., 2017; Schroder et al., 2020).
Methods
Experimental model and subject details
All animal procedures were approved by the Institutional Animal Care and Use Committees at Yale University, University of Maryland or University of Pittsburgh and were in compliance with National Institutes of Health guidelines. For Bbhlhe22-Flpo knock-in/knock-out mice, Flpo recombinase was targeted to the endogenous Bhlhe22 (Bhlhb5) locus, where it replaced the Bhlhe22 gene (Cai, Kardon, et al., 2016). For KOR-Cre knock-in/knock-out mice, Cre recombinase was targeted to the endogenous KOR locus (Oprk1) (Cai, Huang, et al., 2016) (Jackson Laboratory #035045). Dual Cre/Flpo-dependent reporter mice included the following: dual-dependent tdTomato line (Ai65; Jackson Laboratory #021875), dual-dependent ReaChR-mCitrine fusion protein line (Rosa26 CAG-LoxStopLox-FrtStopFrt-ReaChR-mCitrine; Jackson Laboratory #024846), dual-dependent GCaMP7s line (Ai195; Jackson Laboratory #034112); and dual-dependent mitochondrial matrix-targeted dimeric APEX2 line (ROSA26DR-Matrix- dAPEX2; Jackson Laboratory #032764).
Mice were bred to have appropriate combinations of Cre, Flpo and reporter alleles. In the cases of KOR-Cre and Bhlhe22-Flpo alleles, all mice studied were heterozygous for each transgenic allele and therefore had one intact copy of each gene. Pilot studies that crossed either KOR-Cre with a Cre-dependent reporter line (Ai32) or Bhlhe22-Flpo mice with a Flp-dependent reporter line (Cai, Kardon, et al., 2016) suggested that both lines contained WACs with straight, unbranched dendrites, which motivated the intersectional approach described above. All mice were maintained on a 12/12-hour light/dark cycle and were studied at 1.5-6 months age. We did not expect or detect differences between male and female mice, and results have been combined across sex.
Electrophysiology and Cell Filling
Prior to retinal dissection, a mouse was dark-adapted for ∼1 h. After death, both eyes were enucleated and maintained in Ames medium (MilliporeSigma) supplemented with 22.6 mM NaHCO3 (MilliporeSigma) and suffused with 95% O2/5% CO2. The dissection was performed under infrared illumination using stereomicroscope-mounted night-vision goggles (OWL Night Vision Scopes, third generation; B. E. Meyers). The retina was removed from the eyecup, the vitreous humor was removed, and one or more relaxing cuts were made in order to flatten the retina. The retina was mounted on a filter membrane (HAWP01300, MilliporeSigma) with five 0.5-mm diameter holes punched out and maintained in a dissection dish at room temperature for up to several hours. For recording, the retina was placed into a custom recording chamber and secured with a tissue harp. The recording chamber was perfused with Ames medium flowing at 4-6 mL min-1 and maintained at 31 – 33 deg C.
Whole-cell patch-clamp recordings were obtained using patch pipettes pulled from borosilicate glass capillaries (1B120F-4, World Precision Instruments). Pipette tip resistances were ∼4-10 MΩ for amacrine cell recordings in the ganglion cell layer; ∼7-10 MΩ for amacrine cell recordings in the inner nuclear layer; and 2-10 MΩ for ganglion cell recordings. Pipettes were filled with internal solution containing (in mM): 120 K-methanesulfonate, 10 HEPES, 0.1 EGTA, 5 NaCl, 4 ATP-Mg, 0.4 GTP-Na2, and 10 phosphocreatine-Tris2, pH 7.3, 280 mOsm. The internal solution also contained 0.05% (w/v) Lucifer Yellow to label cells for subsequent immunohistochemical experiments.
The components of the internal solution were obtained from MilliporeSigma. Membrane potential was amplified (MultiClamp 700B, Molecular Devices), digitized at 10 kHz (Digidata 1440A, Molecular Devices) and recorded (pClamp 10.0, Molecular Devices). Series resistance ∼20 – 40 MΩ for amacrine cell recordings in the GCL; ∼20 – 100 MΩ for amacrine cell recordings in the INL; and ∼8-25 MΩ for ganglion cell recordings. Series resistance was typically compensated by ∼50%. Recordings were corrected for a −9 mV liquid junction potential.
Two-photon imaging
Two-photon fluorescent measurements were made using a custom-built microscope (Olympus BX-51) equipped with an Olympus 60×, 0.9 NA, LUMPlanFl/IR water immersion objective (Olympus), two fluorescence detection channels for GCaMP7s (HQ 535/50, Chroma) and SR101 (DM 565, 650/SP, Semrock), and an ultrafast pulsed laser (Chameleon Ultra II; Coherent) tuned to 910 nm. ScanImage software version 3.8.1 was used for image acquisition. Time-lapsed images (128 × 128 pixels; 26.8 × 26.8 µm frame) at each IPL depth were acquired at 7.8 frames per second with imaging laser power less than 30 mW. For volumetric imaging in the IPL, 12 time-lapsed z-stacks with 4-µm spacing were captured to cover the B/K WAC dendrites in the IPL. The z plane on the distal side (i.e., towards the INL) of the dendrites in the ON layer was defined as plane 0 for aligning across z stacks.
Visual stimulation
Visual stimuli were generated using custom software written in either Stim-Demo (C language) (Borghuis et al., 2014; Borghuis et al., 2013) or in MATLAB using the Psychophysics Toolbox (Brainard, 1997) and were displayed by a DLP projector (HP Notebook Projection Companion) with 60-Hz refresh rate modified to project UV light (peak wavelength at 395 nm; Borghuis et al., 2013). This wavelength about equally activates cone photoreceptors along the retina’s dorsal/ventral axis, which have varying ratios of green- and UV-sensitive opsins (Wang et al., 2011). The image of the DLP projector was focused on the retina through the microscope condenser.
Stimuli were presented within a 4 x 3 mm area on the retina and included contrast-reversing spots of variable diameter and drifting sine-wave gratings, both at 100% Michelson contrast relative to the background mean luminance that evoked ∼104 photoisomerizations cone-1 s-1. For the drifting grating stimuli used for calcium imaging, a full-field background of mean luminance was presented for 5 s to let the retina adapt to the combination of laser and mean luminance, and subsequently sinusoidal gratings moving in eight directions were presented sequentially. Each grating stimulus was presented for 2 s with a 5 s interval between stimuli. The grating was presented in a patch of 880 µm (29.3 degrees) in diameter and had a 2-Hz temporal frequency and 0.139 cycles/degree (4.54 cycles/mm) spatial frequency.
Spot tuning experiments were performed with spots of diameter 50 – 1675 µm. Because slightly different spot sizes were presented across experiments, responses were binned based on spot diameters centered around 100 µm (50 – 275 µm), 450 µm (276 – 625 µm), 800 µm (626 – 975 µm), 1150 µm (976 – 1325 µm) and 1500 µm (1326 – 1675 µm) before averaging across cells.
Imaging data analysis
Calcium imaging data were analyzed using custom software written in MATLAB. Data were analyzed within regions of interest (ROIs). To select ROIs, a Gaussian-blurred mean image was first subtracted from each frame for background fluorescence subtraction. Then, a local correlation image was calculated. Each pixel in the local correlation image was the mean of the correlation between the center pixel and the neighboring pixels in the surrounding 3×3 matrix. Rectangular ROIs were manually selected from non-overlapping dendritic segments on the correlated image, with one ROI per isolated dendritic segment. The ROIs were drawn as rectangles with the long side of the rectangle parallel to the dendrite. Next, the raw fluorescence and physical orientation θphy of each ROI were extracted based on the binary ROI mask. For the calcium response to the drifting gratings, R(θk) is calculated as the rectified mean of the baseline-subtracted fluorescence evoked by the grating moving in the direction θk for each ROI or image pixel. The baseline was defined as the average fluorescence 2.5 - 5 second after laser onset and just before stimulus onset.
An orientation selective index (OSI) was defined as follows:
which is equivalent to 1 − circular variance in circular statistics. A direction selective index (DSI) was defined as follows:
The preferred orientation was calculated as follows:
The circular Pearson correlation (rcirc) between θphy and θpref in orientation space was defined as follows:
To demix spatially overlapping dendrites and visualize spatial components of apparent dendritic structures that were otherwise indistinguishable in the correlation image, we utilized non-negative matrix decomposition to decompose 3d video Y(t) into products of spatial footprint Ai and corresponding temporal fluorescence traces Ci and background noise B(t):
using the alternative least square algorithm (Berry et al., 2007).
Optogenetics
For optogenetic experiments, conventional photoreceptor-mediated input was pharmacologically blocked via bath application of the following drug cocktail (in µM): 50-100 D-AP5 (Alomone Labs), 50-100 DNQX (Alomone Labs), 20 L-AP4 (Alomone Labs) and 2-5 ACET (Tocris) (Park et al., 2015; Park et al., 2020; Park et al., 2018; Pottackal et al., 2020; Pottackal, Singer, & Demb, 2021; Pottackal, Walsh, et al., 2021). ReaChR-mediated responses were evoked by an LED (λ peak = 450 or 470 nm; maximum intensity ∼5 x 1017 photons s-1 cm-2) focused through the condenser onto a square (220 µm side) as described (Park et al., 2015; Park et al., 2020; Park et al., 2018) and presented for 250-msec duration (average of 4 pulses, 1-s inter-pulse interval, repeated three times) The efficiency of ReaChR stimulation with blue light is within a factor of two as compared to the peak sensitivity using red light (Lin et al., 2013). Optogenetically-evoked responses were measured as postsynaptic currents in RGCs. The response amplitude was the average current over 120 msec surrounding the peak response minus the baseline current, measured over 200 msec before light onset.
Histology
For whole-mount staining of filled cells, the retina was fixed after recording with 4% PFA for 1 h at room temperature. The retina was incubated in 6% donkey serum and 0.5% triton X-100 in PBS for 1 h at room temperature; and then incubated with 2% donkey serum and 0.5% trion X-100 in PBS with primary antibodies for 1 day at 4° C, and with secondary antibodies for 2 h at room temperature.
For staining with the Bhlhe22 antibody, the retina was fixed the tissue in 4% paraformaldehyde in PBS for 1 hour at room temperature; incubated in blocking solution overnight at 4°C; in primary antibody solution for 5 days at 4°C; and in secondary solution for 2 days at 4°C. Retinas were washed three times in PBS for 20 minutes each at room temperature after primary and secondary antibodies were applied. For imaging, the retinas were mounted on slides in Fluoromount G (Electron Microscopy Sciences) with the retinal ganglion cell layer facing the cover slip.
For staining retinal sections, eyecups were placed in 4% paraformaldehyde in PBS for 1-2 hours at room temperature. Next, eyecups were rinsed twice in PBS and embedded in 7% low-melt agarose (Precisionary, SKU VF-AGT-VM). The agarose-embedded eyecups were Vibratome-sectioned into 100 µm-thick slices (VT1200; Leica) and blocked in PBS containing 5% donkey serum and 0.5% Triton X-100 for 1 hour at room temperature. Sections were incubated with primary antibodies overnight at 4°C, washed three times with PBS for 10 minutes, and incubated for 2 hours at room temperature with appropriate secondary antibodies. Stained slices were washed three times in PBS for 10 minutes, mounted onto slides in Fluoromount G (Electron Microscopy Sciences) and cover slipped.
Primary antibodies were used at the following concentrations: rat anti-Bhlhe22 (Bhlhb5; 1:2000) (Ross et al., 2010), goat anti-ChAT (1:200, Millipore AB144P, RRID:AB_2079751, rabbit anti-Lucifer Yellow (1:2000, ThermoFisher Scientific A-5750, RRID:AB_2536190), mouse anti-calretinin (1:5000, Millipore MAB1568). Secondary antibodies were conjugated to Alexa Fluor 488, C3 and C5 (Jackson ImmunoResearch or ThermoFisher Scientific) and diluted at 1:500.
Confocal imaging and analysis
For both retinal sections and flat mounts, confocal images were taken with a Zeiss LSM800 laser scanning confocal microscope. For filled cells, a whole-mount image of the dendritic tree was acquired using a 10X (NA = 0.3) or 20X air objective (NA = 0.8). A high-resolution z-stack was obtained to determine the relative depth of the filled dendrites relative to the ChAT bands, using a 40X oil objective (NA = 1.4); a digital zoom of 0.8 was used to expand the imaged area. Custom software written in MATLAB (MathWorks) was used to determine dendrite stratification relative to the ChAT bands.
The program and methods used were similar to those described previously (Beaudoin et al., 2019; Manookin et al., 2008; Park et al., 2015). The depth of labeled processes is reported relative to the fluorescence peaks of the ChAT bands, in the z dimension, aligned to 0 (peak of inner ChAT band) and 1 (peak of outer ChAT band) in normalized units.
For the antibody staining in retinal flat mounts and sections, the 40X oil objective was used, and image analysis and processing was performed with ImageJ software (Fiji: ImageJ2 Version 2.3.0/1.53q).
Connectomic analysis with Scanning Block Face Electron Microscopy
Reconstructions were performed in an existing volume k0725 (mouse retina; 50×210×260 µm; voxel size 13.2×13.2×26 nm) (Ding et al., 2016); and with a new SBEM volume with labeled B/K WAC processes (61×24×40 µm; voxel size 5×5×40 nm). Manual skeletonization and annotation were performed using Knossos (Max Planck Institute for Medical Research; Heidelberg, Germany) (Helmstaedter et al., 2011). For analyses using k0725, voxel coordinates were tilt-corrected and normalized to the positions of the ON and OFF Starburst amacrine cells. Connectivity analysis was performed using custom-written Python scripts, and data were visualized using Paraview (Kitware, Inc.) (Graydon et al., 2018; Grimes et al., 2022; Park et al., 2020).
To prepare the new SBEM volume with labeled B/K WAC processes, a Bhlhe22-Flpo;KOR-Cre;dual-dAPEX2 mouse was euthanized and the retina was immersed for 30 minutes in 0.2M cacodylate buffer (pH 7.4) containing 4% paraformaldehyde, 2% gluteraldehyde, and 4 mM CaCl2. Tissue was washed with 0.2M cacodylate buffer on ice (4×20 min), immersed for 30 min in 0.2 M cacodylate buffer containing 20 mM glycine on ice, and washed with 0.1M cacodylate buffer (pH 7.4) on ice (5×10 min). To visualize APEX2 expression, tissue was immersed for 30 minutes in 0.2 M cacodylate containing 0.5 mg/ml diaminobenzidine (DAB) on ice and protected from light. H2O2 was added to a final concentration of 0.03% and tissue was incubated for 20 minutes. Finally, tissue was washed in 0.2 M cacodylate buffer (pH 7.4) on ice (3×1 minutes, then 2×20 minutes). To prepare the tissue for SBEM, the sample was post-fixed at room temperature with 2% osmium tetroxide in 0.15 M cacodylate containing 2.5% potassium ferrocyanide, washed in three changes of distilled H2O, incubated at 60° C in 1% aqueous thiocarbohydrazide, washed again in of distilled H2O, incubated in distilled H2O containing 1% aqueous uranyl acetate and 25% ethanol, rinsed in distilled H2O, incubated in lead aspartate at 60° C, and rinsed in distilled H2O. Then, dehydration was performed in a graded series of ethanol (35–100%) and infiltration in a propylene oxide/epoxy resin series was followed by embedding and polymerization in epoxy resin.
Transcriptomic analysis
Transcriptomic data came from a published study of single-cell RNA sequencing of amacrine cells (Yan et al., 2020). The database, available from the website of the Broad Institute (https://singlecell.broadinstitute.org/single_cell), was downloaded and analyzed in custom software written in Matlab. The distribution of Bhlhe22 expression was analyzed for each of the 63 previously assigned amacrine cell groups (i.e., presumed cell types).
Quantification and Statistical Analysis
Data are reported as mean ± SEM. P-values were calculated based on a two-tailed Student’s T-test (paired or unpaired as appropriate). No statistical methods were used to predetermine sample sizes.
Acknowledgements
This research was supported by NIH grants EY029323 (J.B.D.), EY014454 (J.B.D.), EY017836 (J.H.S.), P30 EY026878 (J.B.D.), T32 EY022312 (Z. Jimmy Zhou), DC-IDDRC P50HD105328 (C.C.-P., A.P.), F32 EY032389 (C.H.-G.), F31 EY034776 (T.E.Z.); a National Science Foundation Graduate Fellowship (J.P.) and a Gruber Science Fellowship (J.P.).
Declaration of Interests
The authors declare no competing interesting.
Resource Availability
Further information and requests for resources and reagents should be directly to Jonathan Demb (email: jonathan.demb@yale.edu).
Supplementary Figure Legends
References
- Neural Mechanisms Generating Orientation Selectivity in the RetinaCurr Biol 26:1802–1815https://doi.org/10.1016/j.cub.2016.05.035
- Quantitative analysis of neuronal morphologies in the mouse retina visualized by using a genetically directed reporterJ Comp Neurol 480:331–351https://doi.org/10.1002/cne.20304
- The functional diversity of retinal ganglion cells in the mouseNature 529:345–350https://doi.org/10.1038/nature16468
- Digital Museum of Retinal Ganglion Cells with Dense Anatomy and PhysiologyCell 173:1293–1306https://doi.org/10.1016/j.cell.2018.04.040
- Selective synaptic connections in the retinal pathway for night visionJ Comp Neurol 527:117–132https://doi.org/10.1002/cne.24313
- Algorithms and applications for approximate nonnegative matrix factorizationComputational statistics & data analysis 52:155–173
- Orientation-sensitive amacrine and ganglion cells in the rabbit retinaJ Neurophysiol 71:1672–1691https://doi.org/10.1152/jn.1994.71.5.1672
- Morphological diversity and connectivity of hippocampal interneuronsCell Tissue Res 373:619–641https://doi.org/10.1007/s00441-018-2882-2
- Kainate receptors mediate signaling in both transient and sustained OFF bipolar cell pathways in mouse retinaJ Neurosci 34:6128–6139https://doi.org/10.1523/JNEUROSCI.4941-13.2014
- Two-photon imaging of nonlinear glutamate release dynamics at bipolar cell synapses in the mouse retinaJ Neurosci 33:10972–10985https://doi.org/10.1523/JNEUROSCI.1241-13.2013
- The Psychophysics ToolboxSpat Vis 10:433–436
- Wiring specificity in the direction-selectivity circuit of the retinaNature 471:183–188https://doi.org/10.1038/nature09818
- Generation of a KOR-Cre knockin mouse strain to study cells involved in kappa opioid signalingGenesis 54:29–37https://doi.org/10.1002/dvg.22910
- Bhlhb5::flpo allele uncovers a requirement for Bhlhb5 for the development of the dorsal cochlear nucleusDev Biol 414:149–160https://doi.org/10.1016/j.ydbio.2016.04.028
- A color-coding amacrine cell may provide a blue-off signal in a mammalian retinaNat Neurosci 15:954–956https://doi.org/10.1038/nn.3128
- A dedicated circuit links direction-selective retinal ganglion cells to the primary visual cortexNature 507:358–361https://doi.org/10.1038/nature12989
- Intrinsic properties and functional circuitry of the AII amacrine cellVis Neurosci 29:51–60https://doi.org/10.1017/S0952523811000368
- Functional Circuitry of the RetinaAnnu Rev Vis Sci 1:263–289https://doi.org/10.1146/annurev-vision-082114-035334
- Inhibitory Interneurons in the Retina: Types, Circuitry, and FunctionAnnu Rev Vis Sci 3:1–24https://doi.org/10.1146/annurev-vision-102016-061345
- Species-specific wiring for direction selectivity in the mammalian retinaNature 535:105–110https://doi.org/10.1038/nature18609
- Form and function of the M4 cell, an intrinsically photosensitive retinal ganglion cell type contributing to geniculocortical visionJ Neurosci 32:13608–13620https://doi.org/10.1523/JNEUROSCI.1422-12.2012
- Directionally selective calcium signals in dendrites of starburst amacrine cellsNature 418:845–852https://doi.org/10.1038/nature00931
- Retinal bipolar cells: elementary building blocks of visionNat Rev Neurosci 15:507–519https://doi.org/10.1038/nrn3783
- Ambient illumination toggles a neuronal circuit switch in the retina and visual perception at cone thresholdNeuron 78:325–338https://doi.org/10.1016/j.neuron.2013.02.014
- Requirement for Bhlhb5 in the specification of amacrine and cone bipolar subtypes in mouse retinaDevelopment 133:4815–4825https://doi.org/10.1242/dev.02664
- Synaptic currents generating the inhibitory surround of ganglion cells in the mammalian retinaJ Neurosci 21:4852–4863https://doi.org/10.1523/JNEUROSCI.21-13-04852.2001
- Inhibition decorrelates visual feature representations in the inner retinaNature 542:439–444https://doi.org/10.1038/nature21394
- Distinct Developmental Mechanisms Act Independently to Shape Biased Synaptic Divergence from an Inhibitory NeuronCurr Biol 30:1258–1268https://doi.org/10.1016/j.cub.2020.01.080
- Unified classification of mouse retinal ganglion cells using function, morphology, and gene expressionCell Rep 40https://doi.org/10.1016/j.celrep.2022.111040
- Synaptic Transfer between Rod and Cone Pathways Mediated by AII Amacrine Cells in the Mouse RetinaCurr Biol 28:2739–2751https://doi.org/10.1016/j.cub.2018.06.063
- A polyaxonal amacrine cell population in the primate retinaJ Neurosci 34:3597–3606https://doi.org/10.1523/JNEUROSCI.3359-13.2014
- Dendro-somatic synaptic inputs to ganglion cells contradict receptive field and connectivity conventions in the mammalian retinaCurr Biol 32:315–328https://doi.org/10.1016/j.cub.2021.11.005
- Hierarchical retinal computations rely on hybrid chemical-electrical signalingCell Rep 42https://doi.org/10.1016/j.celrep.2023.112030
- Strategies and Tools for Combinatorial Targeting of GABAergic Neurons in Mouse Cerebral CortexNeuron 91:1228–1243https://doi.org/10.1016/j.neuron.2016.08.021
- High-accuracy neurite reconstruction for high-throughput neuroanatomyNat Neurosci 14:1081–1088https://doi.org/10.1038/nn.2868
- Causal evidence for retina-dependent and -independent visual motion computations in mouse cortexNat Neurosci 20:960–968https://doi.org/10.1038/nn.4566
- Dual-channel circuit mapping reveals sensorimotor convergence in the primary motor cortexJ Neurosci 35:4418–4426https://doi.org/10.1523/JNEUROSCI.3741-14.2015
- Bhlhb5 is required for the subtype development of retinal amacrine and bipolar cells in miceDev Dyn 243:279–289https://doi.org/10.1002/dvdy.24067
- The functional organization of excitation and inhibition in the dendrites of mouse direction-selective ganglion cellsElife 9https://doi.org/10.7554/eLife.52949
- A sign-inverted receptive field of inhibitory interneurons provides a pathway for ON-OFF interactions in the retinaNat Commun 14https://doi.org/10.1038/s41467-023-41638-3
- Intersectional Strategies for Targeting Amacrine and Ganglion Cell Types in the Mouse RetinaFront Neural Circuits 12https://doi.org/10.3389/fncir.2018.00066
- Reconstruction of genetically identified neurons imaged by serial-section electron microscopyElife 5https://doi.org/10.7554/eLife.15015
- Adaptation to background light enables contrast coding at rod bipolar cell synapsesNeuron 81:388–401https://doi.org/10.1016/j.neuron.2013.10.054
- Feature Detection by Retinal Ganglion CellsAnnu Rev Vis Sci 8:135–169https://doi.org/10.1146/annurev-vision-100419-112009
- Molecular identification of a retinal cell type that responds to upward motionNature 452:478–482https://doi.org/10.1038/nature06739
- Space-time wiring specificity supports direction selectivity in the retinaNature 509:331–336https://doi.org/10.1038/nature13240
- An excitatory amacrine cell detects object motion and provides feature-selective input to ganglion cells in the mouse retinaElife 4https://doi.org/10.7554/eLife.08025
- Origins of direction selectivity in the primate retinaNat Commun 13https://doi.org/10.1038/s41467-022-30405-5
- Neurons of the human retina: a Golgi studyJ Comp Neurol 318:147–187https://doi.org/10.1002/cne.903180204
- Cross-compartmental Modulation of Dendritic Signals for Retinal Direction SelectivityNeuron 95:914–927https://doi.org/10.1016/j.neuron.2017.07.020
- Four alpha ganglion cell types in mouse retina: Function, structure, and molecular signaturesPLoS One 12https://doi.org/10.1371/journal.pone.0180091
- Sidekick 2 directs formation of a retinal circuit that detects differential motionNature 524:466–470https://doi.org/10.1038/nature14682
- An unconventional glutamatergic circuit in the retina formed by vGluT3 amacrine cellsNeuron 84:708–715https://doi.org/10.1016/j.neuron.2014.10.021
- Role of ACh-GABA cotransmission in detecting image motion and motion directionNeuron 68:1159–1172https://doi.org/10.1016/j.neuron.2010.11.031
- Populations of wide-field amacrine cells in the mouse retinaJ Comp Neurol 499:797–809https://doi.org/10.1002/cne.21126
- ReaChR: a red-shifted variant of channelrhodopsin enables deep transcranial optogenetic excitationNat Neurosci 16:1499–1508https://doi.org/10.1038/nn.3502
- Glycinergic transmission shaped by the corelease of GABA in a mammalian auditory synapseNeuron 57:524–535https://doi.org/10.1016/j.neuron.2007.12.010
- The shapes and numbers of amacrine cells: matching of photofilled with Golgi-stained cells in the rabbit retina and comparison with other mammalian speciesJ Comp Neurol 413:305–326
- Extreme diversity among amacrine cells: implications for functionNeuron 20:971–982https://doi.org/10.1016/s0896-6273(00)80478-x
- A circuit suppressing retinal drive to the optokinetic system during fast image motionNat Commun 14https://doi.org/10.1038/s41467-023-40527-z
- Disinhibition combines with excitation to extend the operating range of the OFF visual pathway in daylightJ Neurosci 28:4136–4150https://doi.org/10.1523/JNEUROSCI.4274-07.2008
- Distinctive receptive field and physiological properties of a wide-field amacrine cell in the macaque monkey retinaJ Neurophysiol 114:1606–1616https://doi.org/10.1152/jn.00484.2015
- Different mechanisms generate maintained activity in ON and OFF retinal ganglion cellsJ Neurosci 27:5994–6005https://doi.org/10.1523/JNEUROSCI.0130-07.2007
- Amacrine cells of the rhesus monkey retinaJ Comp Neurol 301:382–400https://doi.org/10.1002/cne.903010305
- The tasks of amacrine cellsVis Neurosci 29:3–9https://doi.org/10.1017/s0952523811000344
- Direction selectivity in retinal bipolar cell axon terminalsNeuron 109:2928–2942https://doi.org/10.1016/j.neuron.2021.07.008
- Receptive Fields of Optic Tract Axons and Lateral Geniculate Cells: Peripheral Extent and Barbiturate SensitivityJ Neurophysiol 27:1154–1173https://doi.org/10.1152/jn.1964.27.6.1154
- The Synaptic and Morphological Basis of Orientation Selectivity in a Polyaxonal Amacrine Cell of the Rabbit RetinaJ Neurosci 35:13336–13350https://doi.org/10.1523/JNEUROSCI.1712-15.2015
- Cardinal Orientation Selectivity Is Represented by Two Distinct Ganglion Cell Types in Mouse RetinaJ Neurosci 36:3208–3221https://doi.org/10.1523/JNEUROSCI.4554-15.2016
- Electrical synapses convey orientation selectivity in the mouse retinaNat Commun 8https://doi.org/10.1038/s41467-017-01980-9
- Long-range synchronization of oscillatory light responses in the cat retina and lateral geniculate nucleusNature 379:728–732https://doi.org/10.1038/379728a0
- Inhibitory inputs tune the light response properties of dopaminergic amacrine cells in mouse retinaJ Neurophysiol 110:536–552https://doi.org/10.1152/jn.00118.2013
- Emergence of simple-cell receptive field properties by learning a sparse code for natural imagesNature 381:607–609https://doi.org/10.1038/381607a0
- Function and Circuitry of VIP+ Interneurons in the Mouse RetinaJ Neurosci 35:10685–10700https://doi.org/10.1523/JNEUROSCI.0222-15.2015
- Connectomic analysis reveals an interneuron with an integral role in the retinal circuit for night visionElife 9https://doi.org/10.7554/eLife.56077
- Convergence and Divergence of CRH Amacrine Cells in Mouse Retinal CircuitryJ Neurosci 38:3753–3766https://doi.org/10.1523/JNEUROSCI.2518-17.2018
- Displaced amacrine cells of the mouse retinaJ Comp Neurol 505:177–189https://doi.org/10.1002/cne.21487
- Simultaneous Denoising, Deconvolution, and Demixing of Calcium Imaging DataNeuron 89:285–299https://doi.org/10.1016/j.neuron.2015.11.037
- Receptoral Mechanisms for Fast Cholinergic Transmission in Direction-Selective Retinal CircuitryFront Cell Neurosci 14https://doi.org/10.3389/fncel.2020.604163
- Computational and Molecular Properties of Starburst Amacrine Cell Synapses Differ With Postsynaptic Cell TypeFront Cell Neurosci 15https://doi.org/10.3389/fncel.2021.660773
- Photoreceptive Ganglion Cells Drive Circuits for Local Inhibition in the Mouse RetinaJ Neurosci 41:1489–1504https://doi.org/10.1523/JNEUROSCI.0674-20.2020
- Loss of inhibitory interneurons in the dorsal spinal cord and elevated itch in Bhlhb5 mutant miceNeuron 65:886–898https://doi.org/10.1016/j.neuron.2010.02.025
- Organization and emergence of a mixed GABA-glycine retinal circuit that provides inhibition to mouse ON-sustained alpha retinal ganglion cellsCell Rep 34https://doi.org/10.1016/j.celrep.2021.108858
- Arousal Modulates Retinal OutputNeuron 107:487–495https://doi.org/10.1016/j.neuron.2020.04.026
- Rapid multi-directed cholinergic transmission in the central nervous systemNat Commun 12https://doi.org/10.1038/s41467-021-21680-9
- A non-canonical pathway for mammalian blue-green color visionNat Neurosci 15:952–953https://doi.org/10.1038/nn.3127
- Physiology of the A1 amacrine: a spiking, axon-bearing interneuron of the macaque monkey retinaVis Neurosci 14:507–522https://doi.org/10.1017/s0952523800012165
- Origin and Segmental Diversity of Spinal Inhibitory InterneuronsNeuron 97:341–355https://doi.org/10.1016/j.neuron.2017.12.029
- Immunohistological studies of metabotropic glutamate receptor subtype 6-deficient mice show no abnormality of retinal cell organization and ganglion cell maturationJ Neurosci 19:2568–2579https://doi.org/10.1523/JNEUROSCI.19-07-02568.1999
- Response properties of long-range axon-bearing amacrine cells in the dark-adapted rabbit retinaVis Neurosci 13:599–604https://doi.org/10.1017/s0952523800008506
- Single-Cell Profiles of Retinal Ganglion Cells Differing in Resilience to Injury Reveal Neuroprotective GenesNeuron 104:1039–1055https://doi.org/10.1016/j.neuron.2019.11.006
- Classification of Mouse Retinal Bipolar Cells: Type-Specific Connectivity with Special Reference to Rod-Driven AII Amacrine PathwaysFront Neuroanat 11https://doi.org/10.3389/fnana.2017.00092
- Synaptic Mechanisms Generating Orientation Selectivity in the ON Pathway of the Rabbit RetinaJ Neurosci 36:3336–3349https://doi.org/10.1523/JNEUROSCI.1432-15.2016
- Nonnegative Matrix Factorization: A Comprehensive ReviewIEEE Transactions on Knowledge and Data Engineering 25:1336–1353
- Spectral and temporal sensitivity of cone-mediated responses in mouse retinal ganglion cellsJ Neurosci 31:7670–7681https://doi.org/10.1523/JNEUROSCI.0629-11.2011
- Functional architecture of the mammalian retinaPhysiol Rev 71:447–480https://doi.org/10.1152/physrev.1991.71.2.447
- Glycinergic transmission in the Mammalian retinaFront Mol Neurosci 2https://doi.org/10.3389/neuro.02.006.2009
- Neural Mechanisms of Motion Processing in the Mammalian RetinaAnnu Rev Vis Sci 4:165–192https://doi.org/10.1146/annurev-vision-091517-034048
- Orientation and direction tuning align with dendritic morphology and spatial connectivity in mouse visual cortexCurr Biol 32:1743–1753https://doi.org/10.1016/j.cub.2022.02.048
- Lateral interactions at inner plexiform layer of vertebrate retina: antagonistic responses to changeScience 175:1008–1010https://doi.org/10.1126/science.175.4025.1008
- Mouse Retinal Cell Atlas: Molecular Identification of over Sixty Amacrine Cell TypesJ Neurosci 40:5177–5195https://doi.org/10.1523/JNEUROSCI.0471-20.2020
- Spatially asymmetric reorganization of inhibition establishes a motion-sensitive circuitNature 469:407–410https://doi.org/10.1038/nature09711
- Multiplexed peroxidase-based electron microscopy labeling enables simultaneous visualization of multiple cell typesNat Neurosci 22:828–839https://doi.org/10.1038/s41593-019-0358-7
- Genetically targeted binary labeling of retinal neuronsJ Neurosci 34:7845–7861https://doi.org/10.1523/JNEUROSCI.2960-13.2014
Article and author information
Author information
Version history
- Preprint posted:
- Sent for peer review:
- Reviewed Preprint version 1:
Copyright
© 2024, Park et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.