These studies were conducted under protocols approved by the University of California at Los Angeles (UCLA) Animal Research Committee. All experiments were carried out in accordance with the guidelines for the welfare of experimental animals issued by the U.S. Public Health Service Policy on Human Care and Use of Laboratory Animals and the University of California, Los Angeles (UCLA) Animal Research Committee. Wild-type C57BL/6J mice (20–30 g; Jackson Laboratory, Bar Harbor, ME, USA) of both sexes were used for these studies. Animals were 2–3 months old at the time of the experiments. Animals were deeply anesthetized with 1%–3% isoflurane (Abbott Laboratories, North Chicago, IL, USA) and euthanized by cervical dislocation. To prepare vertical cryostat sections of the retina, the eyecups were fixed in 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (PB), pH 7.4, for 15–60 min at room temperature (RT). Eyecups were then transferred to 30% sucrose in PB overnight at 4°C. The eyecups were embedded in optimal cutting temperature medium (Sakura Finetek, Torrance, CA, USA) and sectioned at 12–14 μm with a Leica CM3050S (Leica Microsystems, Buffalo Grove, IL, USA). Tissue sections were mounted onto gelatin-coated slides and sections were stored at −20°C until immunostaining.

Retinal sections were processed for immunohistochemical labeling using an indirect immunofluorescence method (Pérez de Sevilla Müller et al., 2013, 2015). Frozen retinal sections were thawed for 10–15 min at 37°C on a warming plate, then washed three times for 10 min each with 0.1 M PB (pH 7.4). Retinal sections were then incubated in 10% normal goat serum (NGS) and 0.3%–0.5% Triton X-100 in 0.1 M PB for 1–2 h at RT. Following removal of the blocking solution, sections were then placed in the primary antibodies (see Table 1), diluted in PB with 0.3%–0.5% Triton X-100 and 0.1% NaN₃, overnight at 4°C. After incubation with the primary antibodies, the sections were washed three times for a total of 30 min in 0.1 M PB and placed in their corresponding secondary antibodies: Alexa Fluor goat anti-rabbit 488, goat anti-mouse 594 IgG, Alexa Fluor 568 goat anti-mouse IgG, or Alexa Fluor 568 goat anti-rat IgG (1:1000; Invitrogen, Grand Island, NY, USA) for 1–2 h at RT. The secondary antibodies were removed and sections were washed three times in 0.1 M PB for 10 min per wash. Sections were air-dried and mounted using Aqua Poly/Mount (Polysciences, Warrington, PA, USA), Vectashield (Vector Laboratories), or Citifluor (Citifluor, London, UK).

All antibodies employed in this study have been used previously with PFA-fixed tissue; our immunostaining patterns in the mouse retina were identical to those previously reported in mouse or rat retina (Haverkamp and Wässle, 2000; Deng et al., 2001; Johnson et al., 2003; Martínez-Navarrete et al., 2008; Pérez de Sevilla Müller et al., 2013). Control experiments for nonspecific binding of the secondary antibodies were performed in both single and double-labeling studies.

Whole-mounted retinas were processed for immunohistochemical labeling with a protocol similar to that used for the vertical sections. The retinas were removed from the eyecups and four small incisions were made on each retina to lay the tissue flat. Retinas were mounted onto nitrocellulose membrane filters, with the GCL facing upward (Millipore Corporation, Billerica, MA, USA), and fixed for 15 min in 4% PFA in 0.1 M PB at RT. The whole-mounted retinas were then washed in PB three times for a total of 90 min and incubated in 10% NGS with 0.3%–0.5% Triton X-100 at 4°C overnight. The retinas were subsequently incubated in primary antibody (see Table 1) for 7 days at 4°C and then washed three times for a total of 90 min in 0.1 M PB. The retinas were then placed in the appropriate secondary antibody overnight at 4°C. After three washes for a total of 90 min in PB, the retinas were mounted in Vectashield mounting medium (Vector Laboratories, Burlingame, CA, USA). Coverslips were sealed with nail polish and the slides were stored at 4°C protected from light.

Retinal sections and whole mounts were processed with the following primary antibodies (Table 1) and dilutions: rabbit polyclonal antibody against Prox1 (1:1000–1:2000, PRB-238C, BioLegend, San Diego, CA, USA), mouse monoclonal antibody against calbindin (1:1000, C9848, cl. CB-955; Sigma-Aldrich, St. Louis, MO, USA), mouse polyclonal antibody against calretinin (1:5000, 010399 clone 6B3; Swant, Bellinzona, Switzerland), mouse monoclonal antibody to glutamic acid decarboxylase 67 (GAD₆₇; 1:1000, MAB5406; Millipore, Temecula, CA, USA), mouse monoclonal antibody to Goα (1:300, MAB3073; Millipore, Temecula, CA, USA) and rat polyclonal antibody against glycine (1:1000; IG1002; ImmunoSolution, Everton Park, QLD, Australia). Prox1 antiserum was generated against the C-terminal 15 amino acids of mouse Prox1 (manufacturer’s technical information).

Intracellular injections were performed as described previously (Pérez de Sevilla Müller et al., 2007, 2010a,b; Vuong et al., 2015). Borosilicate glass electrodes (#60200; A-M Systems; Sequim, WA, USA) were pulled and filled at their tips with 0.5% Lucifer Yellow (Sigma–Aldrich) 4% N-(2-aminoethyl)-biotinamide hydrochloride (Neurobiotin; Vector Laboratories, Burlingame, CA, USA), and back-filled with 0.1 M Tris buffer, pH 7.4. In retinal whole mounts, amacrine cell bodies located in the proximal INL at the border of the IPL were targeted for injection. Lucifer Yellow was iontophoresed (−1 nA) into a single cell body and when the bistratified morphology of the AII amacrine cell was recognized, the polarity of the current was reversed (+1 nA) and Neurobiotin was injected for 3 min. The retinas were then fixed in 4% PFA for 10 min and washed for 30 min in 0.1 M PB. Neurobiotin was visualized by incubating the retinas with the injected cells overnight at 4°C with streptavidin–FITC (1:500; Jackson ImmunoResearch, West Grove, PA, USA) in 0.1M PB containing 0.3% Triton X-100 (Sigma–Aldrich). Retinas were washed in PB three times for a total of 30 min. The retinas were subsequently processed for immunohistochemical staining.

Immunostaining was evaluated with a Zeiss laser scanning microscope 710 or 880 (Zeiss LSM 710/Zeiss LSM 880; Carl Zeiss, Thornwood, NY, USA; RRID: SciEx_11637) with a Zeiss C-Apochromat 40× 1.2 NA corrected water objective or Zeiss C-Apochromat 63×/1.4 corrected oil objective at a resolution of 1024 × 1024 or 2048 × 2048 pixels. Images are presented as projection images of two to fifteen image scans (z-axis step between 0.3 and 1 μm). Confocal images were analyzed using Zeiss LSM 510 proprietary software (version 3.2). The intensity levels and contrast of the final images were adjusted in Adobe Photoshop CS2 v.9.02 (Adobe Systems, San Jose, CA, USA).

Strongly expressing Prox1-immunoreactive amacrine cell density was determined from five whole-mounted retinas obtained from three mice (2 months old). Digital images for cell counting were collected at 100 μm intervals from the optic nerve head to the peripheral retina in the superior, inferior, temporal and nasal retinal quadrants. Three retinal fields (500 × 500 μm) per quadrant were collected for each retina using a Plan Apochromat 20×/0.8 NA corrected air objective with a 0.6 magnification factor. Cells were manually counted from the digital images using the cell counter in ImageJ¹ to determine cell number and density. The total number of cells was determined in each sample area and expressed as the number of cells/mm². Nearest-neighbor analysis (Wässle and Riemann, 1978) was performed on the cells located in the nasal area of the retina at 500 μm from the optic nerve head using the plugin “NND” in ImageJ.

For statistical testing we used a one-way ANOVA with p < 0.05 considered significant. Descriptive statistics and Gaussian fit were calculated using GraphPad Prism 4.0 (GraphPad Software, Inc, La Jolla, CA, USA) and presented as a mean ± standard deviation of the mean (SD).

A polyclonal antibody that specifically recognizes the C-terminal 15 amino acids of mouse Prox1 (manufacturer’s technical information) labeled multiple cell bodies in the INL (Figure 1A). There were numerous Prox1-immunoreactive cell bodies in the proximal INL adjacent to the IPL that were similar in size and strongly immunostained (Figure 1A, arrows). Their location and size suggest that they are amacrine cells. There were also numerous weakly stained Prox1-immunoreactive cell bodies in the middle and distal INL; based on their size and position, they likely correspond to bipolar cells (Figure 1A, arrowheads). In addition, there were large and sparsely distributed Prox1-immunoreactive cell bodies in the distal INL at the border of the OPL (Figure 1A, thin arrows) that are horizontal cells (Belecky-Adams et al., 1997; Dyer et al., 2003; Cid et al., 2010). A few Prox1-immunoreactive somata were also in the GCL.

In whole-mounted retinas, strong Prox1 immunostaining was in cell somata in the proximal INL in all retinal regions (Figure 1B). These cell bodies were round in shape and their mean cell body diameter was 7.20 ± 0.44 μm (n = 100 cells; N = 5 retinas). Their cell diameter is consistent with their identity as amacrine cells in the rodent retina (Perry, 1981; Pérez de Sevilla Müller et al., 2007). Prox1 antibodies also labeled numerous somata in the middle and distal INL (Figure 1C). Overall, the immunolabeling of the cells in these regions of the INL was weaker compared to the immunolabeling of the amacrine cell bodies in the proximal INL. The weakly immunostained cell bodies were also round in shape, but their somal diameters were smaller, averaging 5.0 ± 0.3 μm (n = 100 cells; N = 5 retinas), consistent with their identity as bipolar cells (Ghosh et al., 2004; Pignatelli and Strettoi, 2004).

The majority of amacrine cells are either GABA or glycine immunoreactive, while a few amacrine cells do not contain either of these transmitters (for review see Vaney, 1990; Wässle and Boycott, 1991; Pourcho, 1996; Kay et al., 2011). To characterize the neurotransmitter used by Prox1-immunoreactive amacrine cells, we performed double-labeling experiments in retinal sections, using antibodies directed against GAD₆₇, a GABA-synthesizing enzyme (Schnitzer and Rusoff, 1984), or glycine (Pourcho and Goebel, 1985).

In the GCL and proximal INL, numerous small-diameter somata were GAD₆₇-immunoreactive (Figure 2B). However, the GAD₆₇-immunoreactive amacrine cells did not contain Prox1 immunoreactivity (n = 0/87 cells; N = 4 retinas; Figures 2A–C).

In the INL, strong glycine immunostaining was found in multiple amacrine cells and weak immunostaining was observed in bipolar cells (Figures 2D–F), consistent with earlier findings (Menger et al., 1998; Vaney et al., 1998). Glycine immunoreactivity in bipolar cells is due to its diffusion from glycine-containing amacrine cells to bipolar cells through gap junctions (Vaney et al., 1998). All Prox1-immunoreactive amacrine cell bodies also contained glycine immunoreactivity (n = 131/131 cells; N = 5 retinas; Figure 2F); however, not all glycine-immunoreactive amacrine cells expressed Prox1 immunoreactivity (n = 31/87 cells; N = 2 retinas). Prox1-immunoreactive cell bodies comprise 35 ± 3% of all glycine-immunoreactive amacrine cells.

Together, these findings indicate that Prox1-expressing amacrine cells contain glycine immunoreactivity but not GAD₆₇ immunoreactivity, consistent with Prox1-immunoreactive amacrine cells forming a subgroup of the glycinergic amacrine cells.

In the mouse retina, ~35% of the amacrine cells are glycine-immunoreactive (Voinescu et al., 2009; Zhang and McCall, 2012). The glycine-immunoreactive amacrine cells in the rat retina consist of at least eight narrow-field amacrine cell types. The most common type is the AII amacrine cell, which constitutes 20%–30% of the glycinergic amacrine cell population (Menger et al., 1998). On this basis, we tested if the Prox1 and glycine-immunoreactive amacrine cells were AII amacrine cells in the mouse retina.

To test for the expression of Prox1 immunoreactivity in AII amacrine cells, amacrine cells were randomly selected and injected with Lucifer Yellow and Neurobiotin (N = 3 retinas). Labeled AII amacrine cells (n = 5 cells) were identified by their distinct bistratified morphology and their thick lobules in the OFF sublayer of the IPL. The AII amacrine cells exhibited two different gap junctional couplings: AII amacrine to AII amacrine cells, and AII amacrine to bipolar cells (Famiglietti and Kolb, 1975; Strettoi et al., 1992; Chun et al., 1993; Urschel et al., 2006).

The Neurobiotin-injected AII amacrine cell and every tracer-coupled AII amacrine cell were Prox1-immunoreactive (Figures 3A–C, arrows). Furthermore, no Prox1-immunostained cell bodies were in the immediate vicinity of the Neurobiotin-injected AII amacrine cell, suggesting that AII amacrine cells are the predominant cell type expressing Prox1.

In addition, only some of the AII amacrine cell coupled bipolar cells were not Prox1-immunoreactive, indicating that not all bipolar cell types express Prox1 (Figures 3D–F, asterisks). These results are consistent with immunohistochemical findings of subpopulations of glycine-immunoreactive bipolar cell bodies lacking Prox1 immunoreactivity (Figures 2D–F, arrows).

An earlier study of the adult mouse retina reported Prox1 expression in weakly immunostained calretinin amacrine cells (Cid et al., 2010). We tested if Prox1 immunoreactivity is in calretinin-expressing amacrine cells. The Prox1-immunoreactive cell bodies in the proximal INL adjacent to the IPL did not contain calretinin immunoreactivity in retinal sections (Figures 4A–C). These findings are consistent with other studies that showed that AII amacrine cells in the mouse retina are not calretinin-immunoreactive (Haverkamp and Wässle, 2000; Vuong et al., 2015; Meyer et al., 2016). Some cell bodies in the middle INL were weakly immunostained for Prox1 and calretinin (Figures 4A–C, arrows), which is consistent with an earlier study (Cid et al., 2010).

Our experiments with glycine antibodies indicate that some, but not all, bipolar cells displayed weak Prox1 immunoreactivity (Figures 2D–F). Studies in the postnatal and adult mouse retina (Dyer et al., 2003; Cid et al., 2010) showed weak Prox1 immunostaining in bipolar cells. Building on these findings, we performed double labeling experiments for Prox1 and Goα, a marker for ON-cone bipolar cells and rod bipolar cells in the mouse retina (Vardi, 1998; Haverkamp and Wässle, 2000). All Goα-expressing bipolar cells exhibited weak Prox1 immunoreactivity. Prox1-immunolabeled cell bodies that lacked Goα-immunoreactivity were also observed (Figures 5A–C, arrows). Together, these experiments indicate that ON-cone and rod bipolar cells are the predominant bipolar cell types that express Prox1. In addition, some Prox1-expressing bipolar cells are likely OFF-cone bipolar cells, based on the lack of Goα immunostaining (Vardi, 1998; Haverkamp and Wässle, 2000).

We confirmed Prox1 expression in horizontal cells using antibodies to Prox1 and calbindin, a specific marker for horizontal cells (Röhrenbeck et al., 1987; Chun and Wässle, 1993; Massey and Mills, 1996; Haverkamp and Wässle, 2000; Hirano et al., 2005, 2011). Large Prox1 and calbindin-immunostained somata were located in the distal INL at the OPL border, consistent with their identity as horizontal cells (Figures 5D–F, arrows).

The density of Prox1-immunoreactive amacrine cells was measured from the superior to inferior retina and from the nasal to temporal retina in whole-mount preparations (N = 5). Prox1-immunoreactive amacrine cell bodies were in all regions of the retina, with little variation in their distribution and no significant differences (P = 0.556, one-way ANOVA; N = 5 retinas) between the different retinal quadrants (Figures 6A,C,D). The highest average Prox1-containing amacrine cell density (3887 ± 160 cells/mm²) was found 0.5 mm from the optic disc. Cell density (3717 ± 152 cells/mm²) was only slightly lower at 1 mm from the optic nerve head. The mid-peripheral retina, 2 mm from the optic nerve head, had a lower cell density (3133 ± 350 cells/mm²).

In addition, we evaluated the regularity of the mosaic of Prox1-immunoreactive amacrine cells (Figure 6B). We measured the nearest-neighbor distance for each cell body in an area of 250 × 500 μm in the nasal retinal quadrant, 0.5 mm from the optic disc. The area contained a total of 540 Prox1-immunoreactive amacrine cells and their mean nearest-neighbor distance was 10.1 ± 3 μm (mean ± SD). The frequency of nearest-neighbor distances fits a Gaussian distribution, indicating that the Prox1-immunoreactive cell bodies formed a regular mosaic in this region. Furthermore, their regularity index (RI), measured as the ratio between the mean of the nearest-neighbor distances and its standard deviation, was 3.4 (Eberhardt, 1967; Wässle and Riemann, 1978). Together these findings suggest that Prox1-labeled amacrine cells comprise a single amacrine cell population.

Prox1-immunoreactive cell body density did not vary between the different retinal quadrants (P = 0.556, one-way ANOVA; N = 5 retinas). Prox1-immunoreactive cell density was highest (3887 ± 160 cells/mm²) 0.5 mm from the optic nerve head and there was a slight reduction of cell density (3717 ± 152 cells/mm²) 1 mm from the optic nerve head. Cell density (3133 ± 350 cells/mm²) was approximately 20% lower in mid-peripheral retina, 2 mm from the optic nerve head. Our estimates of AII amacrine cell density differ from an earlier report based on Disabled 1 immunostaining of AII amacrine cells (Rice and Curran, 2000); Disabled 1-immunoreactive cell density (4086 cells/mm²) was highest in central retina and lowest (1560 cells/mm²) in dorsal peripheral retina. Although peak cell densities are similar overall, the lower cell density of Disabled 1-immunoreactive cells in the dorsal peripheral retina may be due to an absence or very low levels of Disabled 1 expression. The lower cell density of Disabled 1-immunoreactive cells could also be due to tissue preparation or immunostaining protocols.

The AII amacrine cell distribution based on Prox1 immunostaining in the mouse retina is similar to the AII amacrine cell distribution in other mammalian retinas. In mouse, as well as in rat, rabbit, cat, macaque and human retinas, AII amacrine cell density was greatest in central retinal regions and lowest in the peripheral retina (Vaney, 1985; Wässle et al., 1995; Massey and Mills, 1999; Lee et al., 2004). Our studies of the mouse retina showed a ~20% decrease in cell density between central to mid-peripheral retinal regions. In contrast, there is a ~60%–80% decrease in cell density in rat, rabbit, cat, macaque and human retinas, depending on the species (Vaney, 1985; Mills and Massey, 1999; Lee et al., 2004). The difference of mouse AII amacrine cell density in central and mid-peripheral retinal regions is similar to the modest difference between central and peripheral retinal densities of VIP-Cre-expressing and ChAT-immunoreactive amacrine cells (Keeley et al., 2007; Pérez de Sevilla Müller et al., in press). TH-immunoreactive amacrine cell density does not vary between the central and peripheral mouse retina (Versaux-Botteri et al., 1984; Whitney et al., 2009; Keeley et al., 2014).

Assuming ~39,700 amacrine cells/mm² in the INL of the C57BL/6 retina (Jeon et al., 1998), we estimate that AII amacrine cells comprise ~9.8% of the total amacrine cell population. This percentage is similar to the proportion of AII amacrine cells found in other mammalian retinas. For instance, using parvalbumin as a marker, AII amacrine cells are estimated to make up 10% of the amacrine cells in the rat retina (Wässle et al., 1993) and 11% of amacrine cells in the rabbit retina (Casini et al., 1995; Strettoi and Masland, 1996; Massey and Mills, 1999). Using calretinin as a marker, AII amacrine cells account for nearly a quarter of all amacrine cells in the cat retina (Vaney, 1985; Macneil et al., 2009) and 11% of all amacrine cells in the macaque retina (Wässle et al., 1995; Mills and Massey, 1999).

The AII amacrine cell population forms the largest identified amacrine cell population in the mouse retina, accounting for ~10% of the total amacrine cell population. In comparison, the ChAT- and VIP-immunoreactive amacrine cell populations in the INL account for ~3.0%–5.0% and ~1.4% of the amacrine cell population, respectively (Jeon et al., 1998; Whitney et al., 2008; Keeley et al., 2014; Pérez de Sevilla Müller et al., in press). Furthermore, the dopamine-containing or TH-immunoreactive amacrine cell population is considerably smaller with ~450–600 cells per retina in the C57BL/6J mouse strain (Versaux-Botteri et al., 1984; Masland et al., 1993; Gustincich et al., 1997; Whitney et al., 2009; Keeley et al., 2014).

Prox1-immunoreactive amacrine cell bodies form a nonrandom mosaic, suggested by the fit of their nearest-neighbor distance distribution to a Gaussian distribution (Wässle and Riemann, 1978). Their RI of 3.4 is indicative of a regular distribution of cell bodies in the Prox1-immunoreactive amacrine cell mosaic; a ratio of 1.0 indicates a random distribution and higher ratios indicate a more regular distribution (Eberhardt, 1967; Wässle and Riemann, 1978). The RI of Prox1-immunoreactive amacrine cells in the mouse retina is similar to the regularity indices of AII amacrine cells in rat (RI = 5.1, Wässle et al., 1993; Lee et al., 2004), rabbit (RI = 3.23, Casini et al., 1995), cat (RI = 3.55, Vaney, 1985; Lee et al., 2004), monkey (RI = 2.7, Wässle et al., 1995) and human (RI = 3.76, Lee et al., 2004) retinas.

In conclusion, we report that Prox 1 immunoreactivity can be used as a marker to identify AII amacrine cells in the mouse retina. On this basis, Prox1 antibodies can be used to identify AII amacrine cells to study AII amacrine cell number and distribution in experimental and genetic (Keeley et al., 2014; Reese and Keeley, 2016) models. Additionally, the Prox1 promoter can be useful for developing genetic tools to label a subpopulation of retinal neurons, including AII amacrine cells to study AII amacrine cell structure, connectivity and physiology.