Animal care and experimental procedures were approved by the Ethical and Animal Studies Committee of the University of Murcia (UM; Spain; Codes: A13171103, A13170110, and A13170111) and followed the Spanish and European Union regulations for the use of animals in research (Council Directive 86/609/EEC) and the ARVO statement for the use of animals in ophthalmic and vision research. In this work, we have used pigmented C57BL/6J mice (20–24 g) provided by the animal facility of the UM and housed in temperature-controlled rooms with 12-h light/dark cycles and with food and water provided ad libitum. To perform surgery on the tracing of the RGCs from the optic nerve (see below), mice were anesthetized with a mixture of intraperitoneal ketamine (60 mg/kg body weight, Ketolar; Pfizer, Alcobendas, Madrid, Spain) and xylazine (10 mg/kg Rompun; Bayer, Kiel, Germany).

A total of 37 pigmented mice were used in this study. To study the entire population of RGCs in the pigmented mouse retina, three mice (six retinas) were used to detect the total population of RGCs, both traced from the optic nerve and detected by immunohistochemistry (see below). To study the population and distribution of αRGCs and their subtypes, 30 mice (60 retinas) were used for the study of the retina in whole mounts and four additional mice (eight retinas) were used for the study of the retina in radial sections.

In anesthetized mice, the ONs were exposed intra-orbitally following previously described methods that are standard in our laboratory (Avilés-Trigueros et al., 2000; Vidal-Sanz et al., 2002, 2015, 2017; Nadal-Nicolás et al., 2015a,b). A gelatin sponge (Spongostan Film, Ferrosan A/S, Denmark) impregnated with the tracer hydroxyl stilbamidine methanesulfonate (OHSt, Molecular Probes, Leiden, The Netherlands), a Fluorogold analog, [10% dimethyl-sulfoxide (DMSO) in 0.9% saline], was applied ~1–2 mm around the ocular portion of the optic nerve as described (Nadal-Nicolás et al., 2015a,b). The eye fundus was inspected after the procedure and mice were sacrificed 3 days after tracing.

All mice were sacrificed at the same time range (10.00–11.00 am) to avoid diurnal fluctuation in protein expression, by intraperitoneal lethal injection of pentobarbital (Vetoquinol Dolethal, Especialidades Veterinarias, S.A., Alcobendas, Madrid, Spain). Mice were then transcardially perfused with saline and 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (pH 7.4; Nadal-Nicolás et al., 2009; Valiente-Soriano et al., 2014). The eyes were enucleated and postfixed for a further 1 h in 4% PFA and divided for retinal study in flat mounts or radial sections. For the study of the flat mounts, the retinas were dissected by making four radial cuts, with the deepest superior cut to maintain the orientation of the retina at each moment, following standard protocols in our laboratory (Nadal-Nicolás et al., 2012; Valiente-Soriano et al., 2014). For the study of radial sections, the cornea and crystalline lens were discarded while maintaining the optical cups which were cryoprotected in increasing concentrations of sucrose before being embedded in an optimal cutting temperature (OCT) compound (Sakura Finetek, Torrance, USA) for cryostat sectioning (16 μm; Nadal-Nicolás et al., 2015b).

To study the total RGC population, αRGCs and their subtypes, both in flat retinas and in radial sections, immunohistochemical techniques were performed as described previously (Valiente-Soriano et al., 2014; Gallego-Ortega et al., 2020). Briefly, retinas previously permeabilized in 0.5% Triton X100 PBS were incubated overnight with primary antibodies diluted in blocking buffer (PBS, 2% normal donkey or goat serum, 2% Triton X100). The retinas were then washed in PBS 0.5% Triton X100 and incubated at room temperature for 2 h with appropriate secondary antibodies (1:500 concentration) diluted in PBS 0.5% Triton X100. Finally, the retinas were washed in PBS and mounted vitreous side up on slides and covered with an antifading solution (Galindo-Romero et al., 2013b; Valiente-Soriano et al., 2014).

To detect the total population of RGCs, the RNA-binding protein with multiple splicing (RBPMs) antibody was used (Rodriguez et al., 2014). To detect the αRGC population, SMI32 and OPN antibodies were used (Figure 1; Sonoda et al., 2020; Tan et al., 2022). For the study of the αRGCs subtypes, ONs-αRGCs, ONt-αRGCs, OFFs-αRGCs, and OFFt-αRGCs, different combinations of antibodies detailed in Figure 1 were used (Krieger et al., 2017; Gallego-Ortega et al., 2021; Tan et al., 2022). Previous studies have described that ONs-αRGCs express Calbindin (Krieger et al., 2017) and Tbr2 (Chen et al., 2021), so their combination with the αRGCs panmarkers SMI32 or OPN antibodies, which identify the total αRGC population, allows us to immunodetect these cells (Figure 1; Gallego-Ortega et al., 2021). Moreover, because ONs-αRGCs correspond to M4, one of the ipRGC types, we also investigated if ONs-αRGCs could be detected with OPN4 melanopsin antibody (Gallego-Ortega et al., 2021; Vidal-Villegas et al., 2021a). To identify the ONt-αRGC population, we used the combination of the αRGCs panmarkers SMI32 or OPN positive signal with a negative signal of Tbr2 (only expressed in ONs-αRGC; Chen et al., 2021) and Brn3a (only expressed in OFF-αRGCs; Krieger et al., 2017; Figure 1). To identify OFF-αRGCs, we used the brn3a antibody in combination with Brn3c, negative for OFFs-αRGCs and positive for OFFt-αRGCs (Krieger et al., 2017), and with the αRGCs panmarkers SMI32 or OPN (Figure 1). All primary and secondary antibodies used in this work are listed in Table 1.

Flat mounts and radial sections previously immunostained were analyzed and photographed with an epifluorescence microscope (Leica DM6-B; Leica Microsystems, Wetzlar, Germany) as previously described (Gallego-Ortega et al., 2021). In brief, in reconstructions of retinal flat mounts and radial retinal sections, multiple frames were acquired for each filter in a raster scan pattern (×20) contiguously, side-by-side with no overlap or spacing between images. Individual images were focused on before acquisition and obtained with the same focus for each of the specific filters used. To acquire retinal magnifications, ×40 and ×60 objectives were used (Gallego-Ortega et al., 2021).

To determine in flat-mounted retinas the total numbers of RGCs traced retrogradely with OHSt from the optic nerve or immunolabeled with RBPMs⁺RGCs, standard computer routines developed in our laboratory were used (Nadal-Nicolás et al., 2015a; Vidal-Sanz et al., 2015, 2017). For the study of the population of αRGCs immunodetected against SMI32 or OPN, these were manually dotted on each retinal photomontage, and then the total number of dots per retina was quantified using Image Pro Plus software (IPP 5.1 for Windows^®; Media Cybernetics, Silver Spring, MD, USA) as previously described (Gallego-Ortega et al., 2021). To investigate retinal co-expression of SMI32 and OPN in flat-mounted retinas with no trace of blood to avoid artifacts or undesirable markings (mice were perfused transcardially with saline and PFA, see the previous section), individual frames from eight different retinal areas, covering the central and peripheral region of the supero-temporal, supero-nasal, infero-temporal, and infero-nasal quadrants, were superimposed and merged with Adobe Photoshop (21.2.1, Adobe System Incorporated, USA). To assess the proportion of αRGCs expressing SMI32, OPN, or both, an expert investigator dotted manually each RGC that was clearly stained with each antibody, and RGCs that changed color when images were merged were considered positive for both antibodies. The same protocol was used to identify each αRGC subtype. OPN⁺ or SMI32⁺, and Tbr2⁺ or Calbindin⁺ cells were considered ONs-αRGCs, OPN⁺Brn3a⁻Tbr2⁻ cells were considered ONt-αRGCs, OPN⁺Brn3a⁺Brn3c⁻ or SMI32⁺Brn3a⁺Brn3c⁻ cells were considered OFFs-αRGCs and OPN⁺Brn3a⁺Brn3c⁺ or SMI32⁺Brn3a⁺Brn3c⁺ cells were considered OFFt-αRGCs (Figure 1).

To study αRGCs in radial sections, three transverse sections in the nasotemporal axis containing the optic disc from each retina were selected. From each section, eight magnifications (four from the nasal and four from the temporal, 570 × 570 μm) were acquired at approximately 25%, 50%, 75%, and 95% distance from the optic disc towards the periphery as previously described (Di Pierdomenico et al., 2018) and the same protocol used in flat-mounted retinas were used to dot manually αRGCs and all their subtypes.

The topographic distribution of the entire population of αRGCs and their subtypes were studied by neighbor maps using manually dotted photomontages using previously described methods (Galindo-Romero et al., 2013a; Valiente-Soriano et al., 2014). All maps were plotted using SigmaPlot (SigmaPlot11.0 for Windows; Systat Software, Inc., Richmond, CA, USA) using a color code.

Previous studies have indicated the expression of both SMI32 and OPN in αRGCs, identifying them as effective panmarkers for these cells (Duan et al., 2015; Krieger et al., 2017; Lee and Schmidt, 2018; Sonoda et al., 2020; Tan et al., 2022). In this study, we have used anti-OPN and anti-SMI32 antibodies to identify αRGCs and to analyze its distribution in the adult pigmented mouse retina (Figure 2). As shown in panels A, C, and E of Figure 2, the SMI32 antibody labels the entire cell body of αRGCs, including somas, dendrites, and axons. Although, the labeling is very specific and precise, the accumulation of labeled axons near the optic nerve head makes it problematic to locate their somas in this part of the retina (Figures 2, 3). In comparison, labeling of αRGCs with OPN antibody strongly labels somas and dendritic arbors, and lightly labels axons (Figures 2D,F, and 3), making it easier to identify αRGCs across the entire retinal surface (Figure 2B).

The colocalization study of the two αRGCs panmarkers, SMI32, and OPN, is shown in Figure 3 and Table 2. The results show that both markers colocalized in most instances but it was not homogeneous throughout the retina, with peripheral regions showing a higher percentage of colocalization (89.7%) than central regions (82.3%). However, no such differences were found between the superior (86.1%), inferior (85.7%), temporal (86.2%), or nasal (85.7%) regions (Table 2). This is due to the difficulty in perceiving the soma of SMI32-labeled RGCs in the central retinal regions, as the marker is highly immunoreactive in the nerve fibers of the retinal ganglion cells and does not allow for detecting the somas underneath these nerve fibers (Tan et al., 2022). However, the OPN marker was not as immunoreactive in the nerve fibers, so we could see the cell bodies and primary dendrites of αRGCs (Figure 3).

The total number of αRGCs detected with both antibodies was very similar; 2,252 ± 306 SMI32⁺αRGCs vs. 2,315 ± 175 OPN⁺αRGCs, representing approximately 5.2% of the total number of RGCs labeled with OHSt applied around the uninjured optic nerve head (44,302 ± 3,197, n = 6) or immunodetected with RBPMs (44,956 ± 1,600, n = 6; Figure 4). The topographic maps of the SMI32⁺αRGCs and the OPN⁺αRGCs show an identical distribution. The αRGCs are distributed throughout the retina, from the center to the periphery, with a higher density in the temporal hemiretina (Figures 4A,B).

In this work, we have used immunohistochemical techniques to identify αRGCS and its subtypes which are classified depending on the type of central response (ON-OFF) and whether this response is sustained or transient (ONs-αRGCs, corresponding to the population of M4-ipRGC, ONt-αRGCs, OFFs-αRGCs, and OFFt-αRGCs) in the pigmented mouse retina. Following previous studies, SMI32 or OPN antibodies were used to identify the total αRGC population (Krieger et al., 2017; Lee and Schmidt, 2018; Tan et al., 2022). These antibodies are expressed throughout the αRGC cell body, labeling somas, dendrites, and axons, enabling these cells to be localized throughout the retina. However, OPN labels less intensely the axons and this facilitates the identification of the somas located near the optic nerve head (Figure 2).

The mean population of SMI32- or OPN-labeled RGCs was very similar in mouse retinas (2,252 ± 306 and 2,315 ± 175, respectively). These data are slightly higher than those reported by Schmidt et al. (2014) which counted a total of 1,613 αRGCs per mouse retina. Another study showed the mean density of OPN⁺RGCs in the mouse retina with a mean density of ≈250 cells/mm² (Honda et al., 2019), slightly higher than our results (≈160 cells/mm²). These discrepancies may be explained because densities were estimated by taking photographs of the middle regions of the retina, omitting the central regions close to the optic nerve and the peripheral regions. This may overestimate the result as these areas, as shown in the distribution graph (Figure 4B), clearly have a lower density of cells. Other studies in rat retina show lower density data (60–130 cells/mm²) for SMI32⁺ (Tan et al., 2022), OPN⁺ (Gallego-Ortega et al., 2021) or SMI32⁺OPN⁺ (Mayer et al., 2018) αRGCs than our data, although this may be because the RGC density in rat retina is lower than in mouse retina (Nadal-Nicolás et al., 2018). Tan et al. (2022) indicate a limitation in counting SMI32⁺αRGCs in flat retinas due to the intense axon labeling that hinders the visualization of somas in the region close to the optic nerve head. We also experienced this difficulty in our study and therefore we performed the study of αRGC subtypes with the OPN antibody. Our data show that 5.08% of all RGCs are SMI32⁺αRGCs and 5.23% are OPN⁺αRGCs. These data differ slightly from those reported by Schmidt et al. (2014) (3.7% SMI32⁺αRGCs) and Honda et al. (2019) (≈8.9% OPN⁺αRGCs). A possible reason for these discrepancies may be related to the fact that we count the entire population of RGC labeled cells over the entire retina whereas the other studies sample small regions of the retina.

However, there is more consensus on the description of the topographical distribution of αRGCs in the retina. The color maps show that αRGCs are located throughout the retina although they acquire a higher density in the temporal region (Figures 4A,B), suggesting, as previously indicated by Bleckert et al. (2014), that they have greater activity in the central visual field. These results are also consistent with previously documented distribution (Dreher et al., 1985; Peichl, 1989; Sonoda et al., 2020; Tan et al., 2022) except in the work of Schmidt et al. (2014), which suggests a more homogeneous distribution throughout the retina.

In this work, we studied the population and distribution of αRGCs subtypes using immunohistochemical techniques. Previous studies had identified three types of αRGCs, ONs-αRGCs, OFFs-αRGCs, and OFFt-αRGCs (Cleland et al., 1975; Pang et al., 2003; Van Wyk et al., 2009) until Krieger et al. (2017) characterized physiologically and molecularly the fourth type of αRGCs (ONt-αRGC) in the mouse retina.

The most studied αRGC subtype in the literature is the ONs-αRGC population because it constitutes the M4-ipRGCs, a subtype of ipRGCs that is not detectable with the classical anti-melanopsin antibody (OPN4) that only detects M1, M2, and M3 ipRGCs (Estevez et al., 2012; Schmidt et al., 2014; Abud et al., 2017; Berg et al., 2019; Honda et al., 2019; Sonoda et al., 2020), an observation that we have confirmed in the present studies, there were no ONs-αRGC stained with OPN4 antibodies (Figure 6). In this work, we have designed an immunodetection protocol to detect the ONs-αRGC (M4-ipRGC) population using the αRGCs pan-markers SMI32 or OPN and calbindin (expressed in ONs-αRGCs and amacrine cells; Krieger et al., 2017) or Tbr2, a key transcriptional regulator for the development and maintenance of ipRGCs (expressed in ipRGCs and GABA-displaced amacrine cells; Berg et al., 2019; Tran et al., 2019; Chen et al., 2021). Therefore, these protocols indicate that this αRGC subtype expresses these four proteins, making them easily identifiable. In this work we have detected an average of ~840 ONs-αRGCs (M4-ipRGCs) per mouse retina, representing approximately 36% of all αRGCs (1.9% of total RGCs), and this is in agreement with a recent study (Gao et al., 2022) reporting a density of 52 ONs-αRGCs/mm². Our data, however, differ slightly from those published in previous work in which the percentage of ONs-αRGCs (M4-ipRGCs) was slightly lower. Krieger et al. (2017) reported that approximately 27% of αRGCs were ONs-αRGCs. Another study documented that ON-αRGCs (without differentiating between sustained or transient type) represented approximately 50% of αRGCs (Schmidt et al., 2014), while our data estimates the total percentage of ON-αRGCs of approximately 70% of all αRGCs. These differences could be explained because the above-mentioned studies estimate total RGC numbers based on retinal densities obtained from sampling specific retinal regions, whereas our numbers are based on total counts over full photomontages. Sampling retinal regions may lead to calculation errors because RGCs are not distributed homogenously across the retina, as shown before for the total RGC population of the mice retina (Salinas-Navarro et al., 2009; Galindo-Romero et al., 2011; Ortin-Martinez et al., 2014) and the αRGC population in this work.

Regarding the population of ONt-αRGCs, our data show a 34% proportion (797 ± 146 per retina) of the total number of αRGCs and this is in agreement with Krieger et al. (2017) reporting that 25% of αRGCs expressed neither calbindin, Brn3a nor Brn3c and are therefore classified as ONt-αRGCs. Concerning the OFF-αRGCs population, our data show that they constitute 32% of the αRGCs, slightly less than the 48% documented by Krieger et al. (2017) or the 50% documented by Schmidt et al. (2014). Our data do agree with data provided by Krieger et al. (2017) on the proportion of OFF-αRGCs of sustained or transient response that divides at exactly 50% (48% of OFFs-αRGCs and 52% of OFFt-αRGCs in the study of Krieger et al. (2017). Wang et al. (2021), using the molecular marker Kcnip2 to identify OFFt-αRGCs, document that these show a density recovery profile typical of a distinctive type of RGCs and that their densities accounted for approximately 1.5% of the total RGC population, a slightly larger percentage than that found in our present study.

Although OFF-αRGCs are less numerous than ON-αRGCs, their distribution parallels the two types depending on the type of sustained or transient response (Figure 4). Sustained-response αRGCs are mostly distributed in the superior and temporal quadrants of the retina, with a higher density in the mid-peripheral part of the retina (Figures 4C,E). However, transient-response αRGCs are preferentially located in the central-temporal region of the retina with a lower presence in the peripheral region (Figures 4D,F). While the underlying functionality of this distribution of αRGCs, located throughout the retina, with a higher density in the central-peripheral temporal area (especially sustained response αRGCs), is unknown, SMI32⁺RGCs have been previously described to have asymmetric arbors of normal or slightly enlarged size, similar to those described in melanopsin cells, but have no dendritic contact between homotypic cells, although dendritic contact is necessary for cell-cell spacing and for controlling the size of dendritic arbors (Lin et al., 2004; Tan et al., 2022). This network of extensive dendritic fields suggests that these cells may have a functional action throughout the visual field, although they are most prominent in the central visual field.

Our present work is limited to the αRGCs identified with immunohistochemistry using a combination of antibodies known to be expressed by different α-RGCs subtypes. However, we did not use functional techniques to determine their response to light, nor did we use morphological techniques to measure their soma and dendritic field sizes, or study their laminar distribution within the inner plexiform layer, and thus, perhaps it would be more appropriate to refer to these cells as α-like RGCs (Bleckert et al., 2014; Ou et al., 2016; Tan et al., 2022). The present studies preclude reliable identification of all αRGCs and thus we cannot state that all cells immunodetected in our present work are those with the largest soma in the RGC layer, e.g., mouse retinal neurons in the Eyewire Digital Museum¹ (Bae et al., 2018) as big as the OFFt-αRGCs, neither can we state that our study includes the miniature αRGCs (Baden et al., 2016).

Here, using a simple protocol to identify each of the αRGC subtypes, we further characterize the topographic distribution and the total numbers of each of these subtypes in the mouse retina. This approach may be relevant in future work on the study of the response of each cell type to a retinal pathology or a specific treatment.