Two adult, male macaques (Macaca fascicularis) were used in this study. All procedures performed on macaques were approved by Institutional Animal Care and Use Committees at the University of Alabama at Birmingham (UAB) and performed in accordance with the Association for Research in Vision and Ophthalmology Statement for the use of animals in ophthalmic and vision research. Animal AV263 (Age ~20 years) received no in-life treatment. Retinal cells were labeled with micro-ruby™ post sacrifice. Animal SA76A (Age ~8 years) received in-life bilateral subretinal injections of AAV to label photoreceptors and bilateral LGN injections of micro-ruby™ to label RGCs (Tables 1, 2).

The AAV vector comprising the 292 base pair human rhodopsin kinase promoter (hGRK1) driving GFP (Beltran et al., 2010) packaged in AAV serotype 5 has been previously described (Zolotukhin, 2005; Jacobson et al., 2006a). Virus stock at a titer of 8.0 × 10¹³ vector genomes per milliliter (vg/ml) was diluted in the same buffer used for AAV storage, balanced salt solution (BSS) (Alcon, Fort Worth, TX) with 0.014% Tween 20 (JT Baker, Philipsburg, NJ) to the delivered dose concentration of 1 × 10¹² vg/ml ~1 h prior to subretinal injections. The vector dosing solution was allowed to warm to room temperature to prevent off-gassing during injection.

Tetramethylrhodamine and biotinylated dextran 3000 MW, lysine fixable (micro-ruby™, #D-7162; Thermo Fisher Scientific) was reconstituted to 10% in sterile saline.

Subretinal injections in macaque eyes were performed according to our previously published methods (Boye et al., 2012). For a more detailed description, see Supplementary Materials.

In-life GFP expression was observed 20 days post-injection using a Topcon TRC 50EX fundus camera equipped with a Fundus Photo adapter, Canon EOS 6D digital camera, and New Vision Ophthalmic Imaging Software (Fundus Photo, LLC). This system employs interchangeable, custom excitation and barrier filters, and images of GFP expression were captured with a 469 nm excitation filter (Semrock FF01-469/35-32-D) combined with a 525 nm barrier filter (Semrock FF02—525/40).

Twenty five days post-subretinal injection and 5 days post- fundus imaging, the lateral geniculate nuclei of animal SA76A were injected bilaterally with micro-ruby™ to retrogradely label RGCs as described previously (Dacey et al., 2003, 2005). For a more detailed description, see Supplementary Materials.

At the time of sacrifice (32 days post-subretinal injections for SA76A), animals were deeply anesthetized and eyes were enucleated. The animals were euthanized immediately after enucleation. The anterior chamber and the vitreous were removed. The resulting eyecup was immersed in oxygenated Ames media while the retina was carefully and thoroughly isolated from the retinal pigment epithelia (RPE). Retina from the OS eye of animal AV263 was then dissected into four quadrants (Figure 1). Retinal tissue from quadrant 1 was set aside for use in an unrelated experiment. Quadrant 2 tissue was immersed in RNA later solution (Qiagen, CA, Cat #76104). RNA was extracted and cDNA prepared from this quadrant to validate macaque-specific primers to retinal expressed genes. Retinal tissue from quadrants 3 and 4 were used to optimize cell sorting conditions. Animal AV263's OD eye was hemisected and submerged into oxygenated media as described above. To retrogradely label RGCs, the eye was stabilized with the optic nerve end up, a small piece of tygon tubing was attached to the back of the eye cup, creating a well around the severed end of the optic nerve. 2.5 mg/mL micro-ruby diluted in oxygenated Ames media was added to the well and allowed to incubate for 3 h at room temperature, after which the eye was dissected (maintaining orientation) and the retina cut into 4 quadrants. Quadrant 1 (superior nasal retina) was set aside for use in an unrelated experiment. Retina from quadrant 2 (inferior nasal) was fixed for 1 h at room temperature in 4% PFA, washed and mounted for imaging (data not shown). Retinal tissue from quadrants 3 (superior temporal) and 4 (inferior temporal) were used to optimize cell sorting conditions as described below. The dissociated cells from quadrants 3 and 4 were incubated with 5 μg/ml Alexa-488 fluorophore-conjugated peanut agglutinin (PNA) in 1x PBS/5% FBS for 15 min at room temperature. A detailed summary of experimental design for animal AV263 can be found in Figure 1.

From the eyecups of animal SA76A, 4 mm punches were made with sterile disposable biopsy punch (Sklar surgical instrument, PA Cat #SK96-1115) to isolate the macula/fovea under a dissecting microscope. The remaining OS retina was divided into superior and inferior hemispheres. The OD retina minus the macula was divided into four equal quadrants. Quadrant 1 (temporal superior retina) was kept for an unrelated experiment. Quadrant 2 (temporal inferior retina) was fixed for 1 h at room temperature in 4% PFA and mounted for imaging with a Zeiss Axioscope wide field fluorescence microscope with a 40 × 1.4 NA objective. Quadrants 3 and 4 (nasal inferior and nasal superior retina, respectively) were combined with the inferior and superior regions from the OS retina, respectively to constitute the “Superior” and “Inferior” samples. The macula/foveal punches from both OD and OS eyes were combined to constitute the “Macula/Fovea” sample. Samples were dissociated and analyzed by FACS. A detailed summary of experimental design for animal SA76A can be found in Figure 2.

SA76A was perfused through the aorta with 2 L 1% Sodium nitrite/0.9% Sodium chloride, and fixed with 4L of 4% paraformaldehyde in 0.1 M phosphate buffer. The brain was stereotaxically blocked in the coronal plane at the level of the brainstem, placed in 30% sucrose in 0.1 M phosphate buffer for 3–5 days, and then sectioned at 40 μm on a freezing, sliding microtome (AO 860). All sections from the level of the optic chiasm through the posterior LGN were collected for later processing. Selected sections though the LGN were rinsed and incubated in streptavidin-HRP (Rockland, 1:1000) in PBS plus 0.5% triton X-100. They were then rinsed and reacted with 3′, 3′ diaminobenzidine (DAB), Pierce metal-enhanced solution for 12 min. Sections were rinsed, mounted on gel-subbed slides, and allowed to dry. They were then defatted in Xylene, and coverslipped with Permount. Whole brain sections were imaged at 2.65 μm resolution on a Canon 9000F mark II slide scanner. Contrast and brightness of images were subsequently adjusted in Photoshop for optimal clarity.

The various retinal samples from animals AV263 and SA76A were dissociated with papain (Worthington Biochemical Corporation, NJ, Cat #3150) according to the manufacture's protocol. In brief, papain was pre-incubated in 5 ml of Earle's Balanced Salt Solution (EBSS) for 10 min at 37°C. After pre-incubation, 250 μl of DNase (dissolved in either 500 μl or 250 μl of EBSS) was added to a final concentration of ~20 or 40 units/ml papain and 0.005% DNase. Dissected retina samples were placed in 15 ml falcon tubes containing 700 μl of papain/DNase and equilibrated with 95% O₂:5% CO_2. Retina blocks were dissociated by incubation with activated papain at 37°C either for 20 or 45 min with constant agitation followed by trituration. Dissociated cells were spun down for 5 min at 2000 rpm and resuspended in 500 μl of resuspension media (430 μl EBSS, 50 μl albumin-ovomucoid inhibitor, 25 μl DNase). To prepare the density gradient, 600 μl of albumin-ovomucoid inhibitor was added to a 15 ml Falcon tube, and the cell suspension layered on top. Following centrifugation for 6 min at 1000 rpm, the cells were resuspended with 1x PBS/5% FBS. Dissociated cells from the retinal blocks of animal AV263 were incubated in 5 μg/ml of Alexa-488 fluorophore-conjugated PNA in 1x PBS/5% FBS for 15 min at RT before cell sorting. Sorting of PNA-positive or GFP-positive (GFP+), micro-ruby™ -positive (Ruby+) and unlabeled cells was performed on a BD FACS ARIA SORP equipped with BD FACS Diva software 8.0.1 and a 100 micron nozzle. The filters used to detect the PNA or GFP positive fraction were 505LP and 530/30BP (range 515–545 nm) off the Blue 488 nm laser. The filters used to detect micro-ruby™ were the 600LP and the 610/20BP (range 600–620 nm) off the Yellow Green 561 nm laser.

FACS sorted cells were collected in 1x PBS/5% FBS, spun down for 5 min at 2000 rpm, and resuspended in lysis buffer. Total RNA was extracted from the cell lysate using the RNeasy mini RNA extraction kit according to the manufacturer's protocol (Qiagen, CA, Cat #74104). The RNA samples were treated with DNase (to remove genomic DNA contamination) for 30 min at 37°C followed by 10 min at 75°C to deactivate DNase. The quantity and purity of RNA were determined using a NanoDrop™ 2000 spectrophotometer (Thermo Scientific, MA) and 200 ng of total RNA used as template for reverse transcription using iScript Reverse transcription Supermix (Bio-Rad, CA, Cat #1708890) at 25°C for 5 min, followed by 42°C treatment for 30 min and then 85°C treatment for 5 min. No RT control qPCR reactions were performed with primer pairs that targeted a discrete exon to confirm that samples were free of genomic DNA contamination (data not shown).

qPCR was performed using iTaq™ universal SYBR Green supermix (Bio-Rad, CA, Cat #172-5121) and CFX96 real-time PCR system (Bio-Rad laboratories, Inc., USA). Primers were validated in tenfold serial dilutions of sample. The cycling conditions were as follows: initial denaturation for 5 min at 95°C; 39 cycles of 15 s at 95°C, 30 s at 60°C, and a melting curve of 65°C to 95°C at an increment of 0.5°C per second. Data analyses, which included the determination of a standard curve, quantification cycle (C_q) value, PCR efficiencies, slope of the standard curve and melt curve were automatically performed with the CFX Manager™ Software system (Version 1.6; Bio-Rad Laboratories, Inc., USA). Relative expression of five different primate retinal cell- specific genes including Rhodopsin (RHO; rod specific), guanine nucleotide binding protein (G protein) alpha transducin activity polypeptide 2 (GNAT2; cone specific), glutamate receptor Metabotropic 6 (GRM6; ON bipolar cell specific), glutamate-ammonia ligase/glutamine synthetase (GLUL; Muller cell specific), and Thy-1 cell surface antigen (THY1; retinal ganglion cell-specific) was evaluated. Additionally a subset of samples was evaluated with primer pairs to the different cone opsins, M/L opsin, opsin1 (cone pigments), medium, or long-wave-sensitive (OPN1MW or OPN1LW) and S opsin, opsin1 (cone pigments), short-wave-sensitive (OPN1SW) and other RGC expressed genes, BRN3A, POU class 4 homeobox (POU4F1), melanopsin, Opsin 4 (OPN4) and Retinal pigment epithelium specific protein 65 kDa (RPE65) each assay was carried out using 2 μl (1:10 dilution) of cDNA in a total reaction volume of 20 μl containing 500 nM of each Forward and Reverse primers. A complete list of primers, their sequences, accession number used as reference sequence and their location on the respective cDNA can be found in Supplementary Table 1. All primers were designed to be identical matches to mRNA targets of M. fascicularis (used in this study) and M. mulatta unless otherwise noted. The primer pair for M/L opsin is predicted to equally amplify both M and L opsin cDNA based on annealing location and the existing polymorphisms between the two respective genes. All samples, including the standards, and negative control were run in triplicate. PCR cycling conditions were as follows: one cycle of initial denaturation at 95°C for 5 min; 40 cycles of 30 s at 95°C, 30 s at 58°C and 1 min at 72°C and a final extension step of 72°C for 7 min. GAPDH was used as an internal reference since all the samples used were retina and no comparison to other tissue types were made. Relative quantification was calculated with the ΔΔCq method. Prior to transcript analysis by qPCR, validation of all primer pair was performed conforming to MIQE guidelines (Bustin et al., 2009). Specifically standard curve analysis and confirmation of target specific by melting curve analysis (single peak) was performed utilizing the whole retina from the OS eye of AV263. No signals were detected in any of the “no template” controls (Supplementary Figure 1).

Retina samples from animal AV263 were subjected to two different papain concentrations and incubation times that were selected based on conditions used previously by Han et al. in primate retina (Han et al., 2000). The dissociated samples then underwent FACS to isolate Ruby+ positive (561 nm), PNA+ (488 nm) and unlabeled cells. The scatter plots and the total number of micro-ruby™, PNA positive and un-labeled cells collected under both conditions is presented in Figure 3. Regardless of the method used (20 vs. 45 min incubation in papain), some separation was achieved based on the differential profile of gene expression of THY1, RHO, and GNAT2 between the collected Ruby+, PNA+, and unlabeled cell populations (Figure 3). Ruby+ positive cells were enriched for THY1 expression regardless of the initial duration of papain incubation but GNAT2 expression was only enriched in PNA positive cells when retina was dissociated for 45 min. Based on these preliminary results, we selected 45 min as the optimal duration with which to dissociate macaque retina in papain prior to sorting. However, due to the overlap of THY1 and GNAT2 in the PNA+ and Ruby+ captured cells relative to captured un-labeled cells, we concluded that labeling post-enucleation was insufficient to effectively isolate cones from RGCs. Additionally, PNA labeling only enables capture of cone photoreceptors. We therefore chose to utilize strategies that labeled RGCs and all photoreceptors in-life.

Subretinal delivery of AAV5-GRK1-GFP has previously been shown to drive GFP expression exclusively in rod and cone photoreceptors of all regions of macaque retina (Boye et al., 2012). Utilizing this same AAV vector, we performed a series of subretinal injections in both eyes of animal SA76A with the goal of transducing photoreceptors over a large retinal area. Prior to sacrifice, fluorescence fundus imaging was performed to assess GFP expression. En face fluorescent images revealed robust GFP expression at 20 days post-injection in both the central and peripheral retina that was restricted to the area of each respective subretinal bleb. GFP expression was observed in the macular blebs of each eye however its intensity was decreased relative to that seen in the peripheral blebs, an observation possibly due to natural pigmentation of the macular retina (Figure 4). Notably, by restricting individual bleb volumes to <100 μl, we were able to place at least 5 separate subretinal blebs in different regions of macaque retina, including the macula. Based on clinical observations and fundus images, there was no apparent damage to the retina.

Electrophysiologically-guided injections of the LGN were successful in labeling substantial regions of the LGN and many fibers of passage in the adjacent optic tract. Coronal brain sections through LGN confirmed successful, bilateral micro-ruby™ delivery (Figure 5). Labeling was most intense in the right LGN with horseradish peroxidase (HRP)-stained injection tracks easily visualized (yellow arrows, Figure 5). AAV5-hGRK1-GFP mediated GFP expression in photoreceptors and micro-ruby™ labeling in RGCs were also verified post mortem by direct visualization of cells in a retinal flat mount from the inferior temporal portion of SA76A's OD eye (Figure 6).

Using the optimal dissociation condition described above, GFP positive cells (putative photoreceptors), micro-ruby™ positive cells (putative RGCs) and unlabeled cells from macular, superior and inferior regions of SA76A's retina were isolated using FACS. Scatter plots and the total numbers of cells captured from each population and region are presented in Figure 7. The GFP+ population within all anatomical regions was easily identifiable on scatterplots as an isolated cluster. In contrast, more efficient clustering of Ruby+ cells was observed in the macula/fovea relative to the inferior and superior region samples.

The captured populations of Ruby+, GFP+, and unlabeled cells were first analyzed for expression of RPE65 to confirm that manual separation of neural retina from the underlying RPE efficiently removed the latter cell type from samples. RPE65 expression in the unlabeled cell populations was ~3 log levels lower than that seen in whole retina and was below the level of detection in the Ruby+ and GFP+ cell samples (data not shown).

A comparison of the expression levels of selected retinal genes (THY1, RHO, GNAT2, GLUL, and GRM6) normalized to GAPDH in the Ruby+ and GFP+ cell populations showed a pattern consistent with enrichment for RGCs or photoreceptors, respectively in each of the anatomical regions sampled (Figure 8). For example, in the macula/fovea sample, THY1 expression was highest in the Ruby+ cells while RHO and GNAT2 expression were highest in the GFP+ cells. In addition, expression of all non-RGC- and non-photoreceptor- specific genes were highest in the unlabeled cell populations. When gene expression within each respective fluorescent cell population was normalized to levels observed in the unlabeled cells from that same anatomical region, the magnitude of enrichment was even more apparent (Figure 8). The enrichment factors (fold increase or decrease) for each gene in the Ruby+ and GFP+ cell populations from each anatomical region are presented in Table 3.

We evaluated the expression of M/L opsin and S opsin in GFP+ cell populations sorted from macula/fovea, superior and inferior regions as well as in the corresponding unlabeled cells from the same regions (Figure 9). As expected, opsin expression was much higher in the GFP+ cell populations than in unlabeled cells. Based on the distribution of cone subclasses in the primate retina, it was anticipated that M/L opsin would be enriched in the GFP+ cells from the macula/fovea relative to those isolated from the superior and inferior retina, as was seen for the unsorted cell populations (Curcio et al., 1987). However, in GFP+ cells isolated from macular retina, M/L opsin expression was slightly lower than that seen in GFP+ cells from superior or inferior retina (Figure 9). We believe this result is due either to a lower recovery efficiency of GFP+ cells in the macula/fovea relative to peripheral retina and/or inefficient gating conditions in the macular sample due to small cone size. M/L opsin expression in GFP+ cells from the superior retina was higher than that seen in the inferior sample (Figure 9). Increased M/L opsin expression in superior relative to inferior retina is consistent with the gradient distribution of this cone subclass in the macaque retina (Wikler and Rakic, 1990). S-opsin expression was higher in GFP+ cells of the inferior retina relative to that seen in superior and macula/fovea samples (Figure 9).

A comparison of OPN4 expression in Ruby+ RGCs revealed modest enrichment in the macula/fovea and inferior retina relative to that seen in the superior. In contrast, BRN3A expression was clearly enriched in Ruby+ cells from superior retina relative to that seen in macula/fovea and inferior samples (Supplementary Figure 2).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Despite hosting the research topic together with one of the authors of this manuscript, the handling Editor states that the process met the standards of a fair and objective review.