{"paper_id":"2306e282-d54c-43fc-9115-8efd0c9ac976","body_text":"1 \nSingle-cell analysis reveals cellular heterogeneity and limits of marker-based 1 \nassessment in retinal ganglion cell-enriched organoid cultures 2 \nJessica Yuen Wuen Ma1, Dulce B. Vargas-Landin2, Janya Grainok2, Alice Pébay1,3,4 3 \n1 Department of Anatomy and Physiology, The University of Melbourne, Parkville, VIC 4 \n3010, Australia 5 \n2 PYC Therapeutics, Nedlands, WA 6009, Australia 6 \n3 Department of Surgery, Ophthalmology, The University of Melbourne, East Melbourne, 7 \nVIC 3002, Australia 8 \n4 CellTellus Laboratory, Melbourne, VIC 3000, Australia 9 \n 10 \nCorrespondence: jessica.ma@unimelb.edu.au;  apebay@unimelb.edu.au  11 \n 12 \nAbstract  13 \nHuman pluripotent stem cell (hPSC)-derived retinal organoids provide an  in vitro system for 14 \ngenerating retinal ganglion cells (RGCs), yet the cellular composition and developmental 15 \nfidelity of RGC-enriched cultures remain insufficiently characterised. Here, we tested an 16 \nRGC-enriched approach involving dissociation of hPSC-derived retinal organoids at day 40, 17 \ncorresponding to peak expression of RGC markers, followed by two-dimensional culture 18 \nconditions intended to enrich for RGC survival. Flow cytometry was used to assess the 19 \nexpression of RGC markers, including POU4F, ISL1, SNCG, and THY1. Across four 20 \nsamples, POU4F expression ranged from 79-95%, ISL1 from 18-58%, SNCG from 22%-91% 21 \nand THY1 from 3%-29%, indicating substantial variability between markers and samples. 22 \nSingle-cell RNA sequencing analysis of 73,642 cells identified multiple retinal lineages, 23 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n2 \nincluding retinal progenitors, RGCs, photoreceptor-committed cells, amacrine and horizontal 24 \ncells, and retinal pigment epithelium (RPE), as well as off-target populations comprising 25 \nHOX-enriched posterior neural cells and other cell types. Cellular composition varied across 26 \nsamples. Transcriptomically defined RGCs accounted for 19-45% of cells across samples, 27 \nwith different subtypes identified. These findings indicate that marker-based assessments 28 \nalone may overestimate RGC identity and provide a detailed single-cell characterisation of 29 \ncellular heterogeneity in RGC-enriched retinal organoid cultures. 30 \n 31 \nKeywords 32 \nRetinal ganglion cells, single-cell RNA sequencing, differentiation, retinal organoids, hPSC 33 \nIntroduction 34 \nThe human retina offers a unique and accessible window into the organisation of the central 35 \nnervous system. Retinal ganglion cells (RGCs), the sole projection neurons of the retina, 36 \nconvey visual information to central targets via the optic nerve. Loss of RGCs is a defining 37 \npathological feature of optic neuropathies such as glaucoma, which remains a leading cause 38 \nof irreversible blindness worldwide [1]. Current therapies slow disease progression by 39 \nlowering intraocular pressure but cannot restore RGCs once degenerated [2] , highlighting a 40 \ncritical need for physiologically relevant human models capable of recapitulating RGC 41 \ndevelopment, maturation, and vulnerability to support disease modelling, drug screening, and 42 \npotential cell-based replacement therapies. However, achieving this goal requires reliable 43 \naccess to human RGCs, which are difficult to obtain and maintain in vitro. Indeed, RGCs 44 \nconstitute less than 1% of all retinal cells in the adult eye [3,4], making the prospect of 45 \nisolating large numbers of human RGCs from donor tissue challenging. Human pluripotent 46 \nstem cells (hPSCs) including both embryonic stem cells (hESCs) and induced pluripotent 47 \nstem cells (hiPSCs), represent a promising alternative source for generating human RGCs. 48 \nFor RGCs, two challenges remain: (i) generating sufficiently pure and high yield of RGCs, 49 \nand (ii) maintaining them in culture in a manner that preserves their developmental 50 \ntrajectories and diverse subtypes. Previous studies identified 40-46 RGC subtypes in the 51 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n3 \nmouse retina [5,6]  and 18 subtypes in primates [7], although the exact number in humans 52 \nremains unknown. This diversity is biologically important: RGC subtypes differ in their 53 \nmorphology, function, and potentially disease vulnerability [8]. Hence, heterogeneity must be 54 \ncaptured to allow the reliable study of disease mechanisms and therapeutic responses. 55 \nMany RGC differentiation protocols rely on using small molecules to induce two-56 \ndimensional (2D) differentiation with positive selection based on a limited set of surface 57 \nmarkers [9–12], yet such strategies may bias cultures toward particular molecular subtypes. 58 \nAdditionally, these cultures lack the three-dimensional (3D) organisation and signalling 59 \ngradients that drive retinogenesis in vivo , hence miss the spatial cues required for 60 \nphysiologically appropriate specification and maturation. Furthermore, flow-cytometric 61 \nmarker-based enrichment is rarely complemented by transcriptome-level validation, leaving 62 \nuncertainties about subtype composition and cellular heterogeneity. Retinal organoids offer 63 \nan opportunity to overcome these limitations by providing an  in vivo -like environment that 64 \nsupports the emergence of laminated retinal architecture and intrinsic patterning cues that 65 \nshape early RGC identity. Further, retinal organoids recapitulate the developmental hierarchy 66 \nof retinogenesis, in which RGCs are the first neuronal population to emerge [13]. This 67 \ntemporal advantage allows the generation of comparatively enriched RGC populations when 68 \norganoids are harvested at early stages. Building on this principle, we applied an RGC-69 \nenriched approach to examine RGC differentiation and overall cellular composition in hPSC-70 \nderived retinal organoids. Organoids were dissociated at day 40, corresponding to peak 71 \nexpression of POU4F (BRN3) transcription factors [14], which are key regulators of RGC 72 \ndifferentiation, survival and function [15–17], and replated for an additional 14 days in 73 \nmedium formulated to support RGC survival and maturation. We used flow cytometry and 74 \nsingle-cell RNA sequencing (scRNA-seq) to assess enrichment efficiency, sample-to-sample 75 \nvariability, and the spectrum of retinal and off-target cell types present in the resulting 76 \ncultures. Our findings demonstrate that protein marker-based assessment alone can 77 \noverestimate RGC identity and highlight the necessity of single-cell transcriptomic validation 78 \nfor accurate evaluation of hPSC-derived RGC differentiation. 79 \nMethodology 80 \nEthics  81 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n4 \nThis work was approved by the office of Research Ethics and Integrity of the University of 82 \nMelbourne (2026-32991-75214-7), as per the requirements of the NHMRC, in accordance 83 \nwith the Declarations of Helsinki.  84 \nhPSC maintenance 85 \nHESC H9 (WiCell) and hiPSC WAB-0222 [18] lines were maintained on Matrigel-coated 86 \nplates (Corning, #354230) in mTeSR™ Plus medium (STEMCELL Technologies, #100-87 \n0276). Medium was changed every other day, and cells were passaged weekly for routine 88 \nmaintenance. 89 \nRetinal organoid differentiation 90 \nRetinal organoids were generated following a previously published protocol [14], with some 91 \nmodifications. On Day-1, hPSC cultures at 70% confluency were dissociated using ReLeSR 92 \n(STEMCELL Technologies, #100-0484), and 4,000-5,000 cells were seeded into each well of 93 \na 96-well low adhesion U-bottom plate (Corning, #7007) in mTeSR™ Plus containing 20 µM 94 \nY-27632 (STEMCELL Technologies, #72304). Plates were centrifuged at 300 × g for 3 95 \nminutes to promote aggregate formation. On Day 0, each well received 50 µL mTeSR™ Plus 96 \nand 50 µL Neural Induction Medium (NIM; DMEM/F12 [1:1; Thermo Fisher Scientific, 97 \n#11320033] supplemented with 1% N2 [Thermo Fisher Scientific, #17502001], 1% MEM 98 \nnon-essential amino acids [Sigma, #M7145], 1× penicillin-streptomycin [Gibco, 99 \n#15140122], and 2 µg/mL heparin [STEMCELL Technologies, #7980]), supplemented 100 \nwith 20 µM Y-27632. On Day 1, 40-45 aggregates were transferred directly into 10-101 \ncentimeter polystyrene dishes (Corning, #430591) and fed with 6 mL fresh mTeSR™ Plus 102 \nand 6 mL NIM. Medium was refreshed with NIM on Days 2 and 3. On Day 6, cultures were 103 \nchanged to NIM containing 50 ng/mL BMP4 (R&D systems, #314-BP-010). On Day 8, wells 104 \nof a 6-well plate were pre-coated with 500 µL fetal bovine serum (FBS; Cytiva HyClone,  105 \n#SH30084.03), and 20-30 aggregates were plated per well. On Day 9, half of the BMP4-106 \ncontaining medium was replaced with NIM to achieve a final concentration of 10% FBS, 107 \nfollowed by half-medium changes on Days 12 and 15. On Day 16, organoids were lifted and 108 \ntransferred into 10-centimeter dishes containing Retinal Differentiation Medium (RDM; 109 \nDMEM/F12 [3:1], 2% B27 supplement [Life Technologies, #17504044], 1% MEM non-110 \nessential amino acids, and 1× penicillin-streptomycin) containing 1% FBS. RDM was 111 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n5 \nreplaced on Days 18 and 20 using media supplemented with 3% and 5% FBS, respectively. 112 \nOn Day 22, organoids were transitioned to Advanced RDM (ARDM; DMEM/F12 [3:1], 113 \nsupplemented with 2% B27, 1% MEM non-essential amino acids, 1× penicillin-streptomycin, 114 \n1× GlutaMAX, 10% FBS, and 100 µM taurine). Half-medium changes were performed every 115 \n2-3 days until Day 40. 116 \nOn Day 40, retinal organoids were dissociated using the Papain Dissociation Kit 117 \n(Worthington Biochemical Corporation, #LK003153). Following a 30-minute incubation at 118 \n37 °C, organoids were triturated 20 times with a 1-mL pipette to generate a suspension 119 \nenriched for small aggregates. The suspension was transferred to a 15-mL tube containing an 120 \nequal volume of 10 mg/mL ovomucoid protease inhibitor and centrifuged at 300 × g for 5 121 \nminutes. The resulting pellet was resuspended in 1 mL Neurobasal-based neuronal 122 \ndifferentiation medium (NDM; Neurobasal Plus [Thermo Fisher Scientific, #A3582901], 123 \nsupplemented with 1% MEM non-essential amino acids, 1% GlutaMAX  [Thermo Fisher 124 \nScientific, #35050061], 1% 45% glucose [Merck , #G8769], 1× penicillin-streptomycin, 1× 125 \nB27, 1× N2, 1× CultureOne [Thermo Fisher Scientific, #A3320201], and 1× Normocin 126 \n[InvivoGen, #ant-nr-2]). All supplements were added fresh immediately before use. Viable 127 \ncells were counted using trypan blue exclusion and plated onto 12-well plates coated with 128 \npoly-D-lysine (2 µg/cm²; Sigma-Aldrich, #P0899-10MG) and laminin (1 µg/cm²; Sigma-129 \nAldrich, #L2020-1MG) at a density of 200,000 cells/cm². Following dissociation, cells were 130 \ncultured in NDM containing 20 µM Y-27632, 10 ng/mL CNTF (PeproTech; #450-13-50UG), 131 \n40 ng/mL BDNF (PeproTech, #450-02-50UG), 10 µM forskolin (Biogem/Lonza, #6652995), 132 \nand 3 µM DAPT (Abcam, #AB120633), applied as a full medium change. Y-27632 was 133 \nadded immediately after plating. On Day 41, a half-medium change with NDM containing 134 \nCNTF, BDNF, forskolin, and DAPT was performed, with medium added gently along the 135 \nwell perimeter to avoid dislodging adherent cells. A full medium change was performed on 136 \nDay 43 to remove debris and eliminate residual Y-27632. From Days 44-47, half-medium 137 \nchanges were performed every 2-3 days while maintaining CNTF, BDNF, forskolin, DAPT, 138 \nand CultureOne. From Days 48-54, medium changes continued as above, with forskolin 139 \nincreased to 25 µM; Day 48 involved a full medium change, followed by half-medium 140 \nchanges every 2-3 days. 141 \n  142 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n6 \nFlow cytometry 143 \nhPSC-RGCs cultured for 14-16 days on PDL/laminin-coated 24-well plates were dissociated 144 \nusing either TrypLE Express (Thermo Fisher Scientific; 12604-021, 10 minutes, 37 °C) or the 145 \nPapain Dissociation Kit (Worthington Biochemical Corporation; LK003153, 20 minutes, 37 146 \n°C), depending on cell numbers. The papain kit was more effective at preserving cell viability 147 \nin cultures with lower cell yields. Cells were collected by centrifugation at 340 × g for 5 148 \nminutes at 4 °C, and the resulting pellets were resuspended in DPBS- 2% BSA (FACS 149 \nbuffer). Live/dead staining was performed using the Fixable Violet 405 dye kit (Thermo 150 \nFisher Scientific, #L34964, 30 minutes, room temperature). Cells were then washed in the 151 \nFACS buffer, and centrifuged at 340 × g for 5 minutes at 4 °C. For extracellular staining, 152 \ncells were resuspended in FACS buffer and incubated with Anti-CD90 BV510 (BD Horizon, 153 \n#563070) for 45 minutes on ice. For intracellular staining, cells were first resuspended in 154 \nDPBS, fixed with 4% paraformaldehyde for 10 minutes at room temperature, and 155 \npermeabilized in 0.1% BSA in DPBS-0.05% Triton X-100-0.1% Tween-20 for 15 minutes at 156 \nroom temperature. Cells were subsequently resuspended in 0.1% BSA- DPBS- 0.1% Tween-157 \n20 and incubated with the following antibodies for 45 minutes at room temperature: Anti-158 \nSNCG Alexa Fluor 488 (Santa Cruz, #sc-65979), Anti-ISL1 PE (BD Pharmingen, #562547), 159 \nAnti-GFAP Alexa Fluor 647 (BD Pharmingen, #560298), and Anti-POU4F Alexa Fluor 647 160 \n(Santa Cruz, catalog #sc-390780). Antibody selection was based on markers that had been 161 \ndetected in hPSC-derived RGCs. The pan-RGC marker RBPMS was therefore not included 162 \ndue to its variable and often low expression in these cultures [19]. Stained cells were analysed 163 \nusing a CytoFLEX LX flow cytometer, and data were processed with FlowJo v10.10 164 \nsoftware. A negative control using hPSC-derived RPE cells is shown in Fig. S1. 165 \nSingle cell preparation of iPSC-RGCs 166 \nhPSC-RGCs cultured for 14-16 days on PDL/laminin-coated 24-well plates were dissociated 167 \nusing the Papain Dissociation kit according to the manufacturer’s protocol. Briefly, cells were 168 \nincubated with papain/DNase solution at 37 °C for 20 minutes, gently triturated, and the 169 \nenzymatic reaction was quenched using the albumin-ovomucoid inhibitor solution. Following 170 \ndissociation, cells were centrifuged and washed in 1% BSA, then sequentially filtered 171 \nthrough 30 µm pre-separation filters (Miltenyi Biotec; 130-041-407) and kept on ice. Cell 172 \nviability and concentration were determined by Trypan Blue exclusion using a Countess 3 FL 173 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n7 \nAutomated Cell Counter (Thermo Fisher; AMQAF2000). Pelleted cells were fixed for long-174 \nterm storage using the GEM-X Flex Sample Preparation v2 Kit (10x Genomics; 1000781) 175 \naccording to the manufacturer’s CG000782 protocol for GEM-X Flex Gene Expression. 176 \nGeneration of single cell GEMs and sequencing libraries 177 \nSingle-cell suspensions were processed by the Australian Genome Research Facility using 178 \nthe 10x Genomics Chromium GEM-X Flex Gene Expression Human 4-plex assay according 179 \nto the manufacturer’s protocol. For each sample, 3 × 10 /i1  fixed cells were hybridised with 180 \nuniquely barcoded probe sets (BC001-BC004) for 20 hours at 42 °C, washed, and pooled at 181 \nequal concentrations. Pooled cells were loaded onto a Chromium X instrument with GEM-X 182 \nFX Chips, combining barcoded Gel Beads, master mix, and Partitioning Oil B to generate 183 \nsingle-cell Gel Beads-in-Emulsion (GEMs) targeting recovery of ~80, 000 cells. GEMs were 184 \ntransferred to a thermal cycler to ligate the left-hand and right-hand probes that remained 185 \nhybridised to their target RNA, hybridise Gel Bead primers to the capture sequence of the 186 \nligated probe pairs, and extend barcode sequences. Following emulsion breaking with 187 \nRecovery Reagent, the ligated and extended products were PCR-amplified, cleaned, and 188 \nindexed. Libraries were quality-assessed on a TapeStation D1000, quantified by qPCR, and 189 \nsequenced on an Illumina NovaSeq X Plus (10B flow cell; 150 bp paired-end + 10 % PhiX). 190 \nMapping of reads to transcripts and cells 191 \nBase-call files were processed using Cell Ranger v7.1.0 (10x Genomics) configured for the 192 \nChromium Single Cell 3 ′  v3.1 chemistry. Reads were aligned to the Homo sapiens  reference 193 \ngenome (GRCh38, Ensembl release 109). Cell Ranger performed default barcode and UMI 194 \ncorrection to generate unfiltered gene-by-cell count matrices. No library aggregation was 195 \nperformed. 196 \n  197 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n8 \nscRNA-seq processing, integration, and visualisation 198 \nCell recovery and ambient RNA correction 199 \nRaw gene-barcode count matrices were processed to remove empty droplets and correct 200 \nambient RNA contamination. Ambient RNA contamination was corrected using the SoupX 201 \npackage [20]. For each sample, the ambient RNA profile was estimated from barcodes not 202 \npresent in the cell-containing matrix (raw-only barcodes) with total UMI counts in the range 203 \n1-100 UMIs. Non-empty putative cells were pre-clustered in Seurat (v5) using LogNormalize 204 \n→  FindVariableFeatures →  ScaleData →  Run principal component analysis (PCA; 30 PCs) 205 \n→  FindNeighbours →  FindClusters (Leiden, resolution = 0.4). These quick cluster labels 206 \nwere supplied to SoupX, contamination fraction ( ρ ) was estimated using autoEstCont, and 207 \ncorrected counts were generated with adjustCounts. 208 \nPost-processing and doublet detection 209 \nBased on the pre-filtering quality-control metrics (Fig. S2), Seurat objects were reconstructed 210 \nfrom SoupX-corrected counts and filtered using thresholds of nFeature_RNA 1000-8000 and 211 \nmitochondrial RNA percentage < 30%. Doublets were detected on raw counts using 212 \nscDblFinder (SingleCellExperiment backend; serial execution) [21], and only singlets were 213 \nretained. 214 \nNormalisation and Harmony Integration 215 \nEach sample was normalised independently using SCTransform v2 (glmGamPoi backend) 216 \n[22], regressing out mitochondrial transcript fraction. After merging samples, principal 217 \ncomponent analysis was performed on SCT residual features, and sample-associated 218 \ntechnical variation was mitigated using Harmony. Harmony aligns transcriptionally similar 219 \ncell states across samples without assuming technical replicates or enforcing alignment of 220 \nnon-overlapping populations. Harmony embeddings were used for UMAP visualisation and 221 \nconstruction of the shared nearest-neighbour graph (dims 1-30). Clustering was performed 222 \nusing the Leiden algorithm across a range of resolutions (0.1-1.0) to assess cluster stability 223 \nand granularity. 224 \nDimensionality reduction and clustering 225 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n9 \nPrincipal component analysis (PCA; 50 components) was performed on the Harmony-226 \nintegrated object. To determine the number of biologically informative dimensions, we 227 \nprimarily examined the Elbow plot, which revealed a clear inflection point at approximately 228 \n20-25 principal components ( Fig. S3A). This observation was supported by the cumulative 229 \nvariance plot, which showed that the majority of structured variation was captured within the 230 \nfirst ~20-25 PCs, with progressively smaller gains thereafter ( Fig. S3B). To ensure retention 231 \nof potentially informative higher-order components while avoiding excessive noise, a 232 \nconservative cutoff of the first 30 PCs was selected for all subsequent neighbour graph 233 \nconstruction, UMAP embedding, and clustering. 234 \nCluster stability assessment and resolution selection 235 \nUMAP embeddings were generated using Harmony-corrected embeddings (min.dist = 0.3, 236 \nspread = 1.0) with fixed random seeds to ensure reproducibility [23]. Graph-based clustering 237 \nwas performed using the Leiden algorithm (algorithm = 4) across a range of resolutions from 238 \n0.1 to 1.0 in increments of 0.1 [24]. Cluster stability across resolutions was assessed by 239 \ncalculating the Adjusted Rand Index (ARI) between clustering solutions at consecutive 240 \nresolutions. ARI values increased rapidly between low resolutions (0.2-0.4) and reached a 241 \nhigh, stable range from resolution 0.4 onwards, indicating robust preservation of cluster 242 \nstructure across increasing granularity ( Fig. S4A). Consistent with this, cluster visualisation 243 \ndemonstrated coherent cluster propagation with minimal fragmentation across intermediate 244 \nresolutions, forming a stable plateau between resolutions 0.4 and 0.5 ( Fig. S4B). Based on 245 \nthe convergence of ARI stability and clustree topology, a resolution of 0.5 was selected for 246 \nall downstream analyses. 247 \nDifferential gene expression and marker identification 248 \nDifferential expression analysis was performed on the Harmony-integrated object using 249 \nSeurat v5. Prior to differential expression analysis, clusters containing fewer than 100 cells 250 \nwere excluded to ensure sufficient statistical power. The object was prepared for DE using 251 \nPrepSCTFindMarkers(). Cluster-specific markers were identified using FindAllMarkers 252 \n(test.use = \"wilcox\", only.pos = TRUE, min.pct = 0.10, log2FC threshold = 0.25), comparing 253 \neach cluster against all other cells. Gene identifiers were standardised to HGNC symbols 254 \nusing Ensembl v109 via biomaRt. 255 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n10 \nCell type annotation and marker validation 256 \nGene symbols were standardised to HGNC conventions using HGNChelper [25] and 257 \nbiomaRt [26]. For each Seurat cluster, average gene expression (SCT assay) was z-scored 258 \nand visualised with ComplexHeatmap to confirm marker specificity and to guide manual 259 \nannotation. Clusters containing fewer than 100 cells were excluded. The remaining clusters 260 \nwere assigned to retinal progenitor cells, RGCs, amacrine cells, horizontal cells, 261 \nphotoreceptors, retinal pigment epithelium (RPE), Multilineage (stressed), Other ( HOX-262 \nenriched), or Other based on the expression of established retinal marker genes from human 263 \ndatasets (Table 1) [27–29].  264 \n 265 \nTable 1: Established retinal marker genes from human datasets. 266 \nCell Identity  Marker genes \nRetinal progenitor cell  SOX2, PAX6, VSX2, LHX2, MKI67, ATOH7, FABP7, HES6, \nDLL3, PTF1A \nRGC POU4F2 (BRN3B), ISL1/2, EOMES, RBPMS, SNCG, THY1, \nNEFL, NEFM, GAP43, SLC17A6, POU6F2, ELAVL3, ELAVL4 \nAmacrine cell GAD1, TFAP2A, PRDM13, CHAT, SLC6A9  \nHorizontal cell PROX1, LHX1, ONECUT1, ONECUT2, GJA10, RORB \nPhotoreceptor OTX2, CRX, THRB, NRL, RHO, GNAT1, NR2E3, ARR3, \nGNAT2, PDE6H \nRPE RPE65, MITF, TTR, TYR, TYRP1, RDH5 \nMultilineage (stressed) FOS, JUN, ATF3, DDIT3, CDKN1A, HSPA1B \nOther (HOX-enriched) HOXB5-8 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n11 \n  267 \nProjection of Multilineage (stressed) cells 268 \nTo infer the lineage affiliation of “Multilineage (stressed)” cells, we applied Seurat’s label-269 \ntransfer framework using non-stressed cells as the reference population. Cells in cluster 19, 270 \ndefined based on Harmony-integrated clustering, were treated as the query set. Transfer 271 \nanchors were computed using PCA-based reduction on normalised expression profiles, and 272 \nlineage predictions for each stressed cell were generated using TransferData.   273 \nTrajectory inference, lineage reconstruction, and pseudotime estimation 274 \nThe developmental root was defined by proliferative activity within retinal progenitor cell 275 \nclusters. Cell-cycle phase was assigned using Seurat’s CellCycleScoring with canonical S-276 \nphase and G2/M gene sets. Clusters 15 and 17 exhibited high proliferative activity; however, 277 \ncell-cycle phase analysis revealed distinct cycling states. Cluster 15 comprised a mixed 278 \ncycling population (50.9% S phase, 49.1% G2/M), whereas cluster 17 was almost exclusively 279 \nG2/M (98.9%), indicating a highly synchronised mitotic progenitor population. Cluster 17 280 \nwas therefore selected as the trajectory root, as it represents the most actively cycling 281 \nprogenitor state. In contrast, differentiated neuronal populations, including RGCs (61-95.8% 282 \nG1), interneurons (75.8-84.6% G1), and photoreceptor-committed cells (53.8-71.2% G1), 283 \nwere predominantly in G1, consistent with post-mitotic states. 284 \nCell differentiation trajectories were inferred using Slingshot  (v2.16.0) applied to the 285 \nPCA embeddings derived from the Harmony-integrated object (R v4.5.1) [30]. Trajectory 286 \ninference was performed in a family-wise manner, with separate analyses conducted for 287 \nmajor retinal lineages without cross-lineage interference. For each lineage family, Slingshot 288 \ninferred smooth principal curves through the cluster topology, generating continuous 289 \npseudotime values for all included cells. Pseudotime distributions for downstream 290 \ncomparison and visualisation were summarised using ridge plots (ggridges). 291 \nRGC Subclustering and Subtype Annotation  292 \nClusters 2, 3, 11, and 14 were identified as RGCs and extracted for downstream analysis. All 293 \nanalyses were performed in R (v4.5.1) using Seurat (v5) together with harmony, mclust, 294 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n12 \nclustree, dplyr, and ggplot2. This RGC population was reprocessed independently. The RNA 295 \nassay was re-normalised using SCTransform, and PCA was performed on the resulting SCT 296 \nassay, computing 50 PCs. Dimensionality selection was guided by PCA elbow plots as well 297 \nas variance explained and cumulative variance plots ( Fig. S5). Based on these metrics, the 298 \nfirst 20 PCs were retained for downstream neighbour graph construction and dimensionality 299 \nreduction. A shared nearest-neighbour graph was constructed using the Harmony embeddings 300 \n(dimensions 1-20), and Leiden clustering was applied across a resolution grid ranging from 301 \n0.1 to 1.0 (step size = 0.1). Clustering stability was assessed by calculating the ARI between 302 \nadjacent resolutions using mclust and by visual inspection of cluster relationships using 303 \nclustree. Based on ARI profiles and clustree topology, a final clustering resolution of 0.3 was 304 \nselected (Fig. S6).  305 \nRGC subtype markers were selected based on previously reported human RGC 306 \nmarker genes associated with major RGC subtypes (Table 2), including α /Parasol, Midget, 307 \ndirection-selective ganglion cells (DSGC), orientation-selective ganglion cells (OSGC; J-308 \nRGC-like), large sparse RGCs, and intrinsically photosensitive RGCs (ipRGCs) [7,29,31–309 \n33]. 310 \nTable 2: Established marker genes for human RGC subtypes. 311 \nSubtypes Markers \nα /Parasol SPP1 CAV2 CHRNA2 POU6F2 FABP4 \nMidget  TPBG TBR1 GUCY1A1 RBPMS2  \nDSGC DCX CDH6 FSTL4     \nOSGC (J-RGC) JAM2         \nLarge sparse SATB2         \nipRGC OPN4         \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n13 \nResults 312 \nGeneration of RO and incorporation of RGC enrichment strategies 313 \n 314 \nFigure 1. Generation of retinal organoids and subsequent RGC enrichment. (A) 315 \nSchematic diagram illustrating the generation of retinal organoids, and ( B) the subsequent 316 \nenrichment of RGCs following dissociation of retinal organoids at day 40 (D40). Created 317 \nwith BioRender.com (C-F) Brightfield images showing retinal organoids at D40 (left panel) 318 \nand the corresponding RGC-enriched enriched cell populations at D54 (right panel) for ( C) 319 \nH9_RGC1, (D) H9_RGC2, (E) WAB_RGC2, (F) WAB_RGC1. Scale bar: 100 µm. 320 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n14 \n  321 \nThe RGC enrichment approach was developed by integrating elements from prior studies 322 \n(Fig. 1A,B). Retinal organoids were first generated using a modified version of a previously 323 \nestablished protocol [14], then dissociated and plated for further differentiation, a step that 324 \nhas been shown to favour enrichment of neuronal populations expressing CD90 [10]. Retinal 325 \norganoids were dissociated at day 40, a developmental stage reported to exhibit peak POU4F 326 \nexpression, a canonical RGC marker [14]. Finally,  cultures were maintained under 327 \nconditions previously shown to support the survival and maturation of emerging RGC-like 328 \ncells [9](Fig. 1B). 329 \nRetinal organoids derived from two hPSC lines were produced in duplicate 330 \nindependent cultures (H9_RGC1, H9_RGC2, WAB_RGC1, and WAB_RGC2). All samples 331 \nunderwent the same differentiation procedure from the hPSC stage to retinal organoids 332 \nformation. Typically, lifted retinal organoids after day 16 acquire a smooth, rounded 333 \nmorphology with a distinct bright halo along the periphery, indicating the establishment of 334 \nlaminated retinal layers. This characteristic morphology was observed in H9_RGC1, 335 \nH9_RGC2, and WAB_RGC2 ( Fig. 1C-E, left panels). In contrast, WAB_RGC1 organoids 336 \nbecame flattened during the plating-down period, likely as a result of disruption to the 337 \nembryoid body surface during manual transfer. Despite the flattening, these organoids were 338 \nlifted and retained for downstream culture and analysis to assess whether such morphological 339 \nchanges could serve as a potential morphological selection criterion in future protocols. Upon 340 \nlifting, WAB_RGC1 organoids developed irregular, clumped edges around a central dense 341 \ncore, and lacked the characteristic bright peripheral halo, suggesting disrupted lamination 342 \n(Fig. 1F, left panel). In addition, regions of folded or thickened tissue were observed on the 343 \nsurface of WAB_RGC1 organoids, indicating a failure to establish proper neuroepithelial 344 \npolarity ( Fig. 1F , left panel). We therefore included WAB_RGC1 as a suboptimal 345 \ndifferentiated control to assess whether altered organoid morphology would influence RGC 346 \nyield.  347 \nAfter 14 days of RGC enrichment following organoid dissociation, cells from 348 \nH9_RGC1, H9_RGC2, and WAB_RGC2 formed multiple well-defined neuronal clusters 349 \ninterconnected by thin, long dendritic processes, typical of neuronal networks ( Fig. 1C-E , 350 \nright panels). By contrast, WAB_RGC1-derived cultures displayed multiple dense aggregates 351 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n15 \ncontaining several large soma-like centres with multiple extended multiple neurites ( Fig. 1F, 352 \nright panel; large soma-like centres indicated by arrowheads).  353 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n16 \nFlow cytometric assessment of RGC marker expression 354 \n 355 \n 356 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n17 \nFigure 2. Flow cytometry assessment of RGC-associated marker expression. (A) 357 \nRepresentative flow cytometry dot plots showing expression of POU4F, ISL1, SNCG, and 358 \nTHY1 across four samples (H9_RGC1, H9_RGC2, WAB_RGC2, and WAB_RGC1). Rows 359 \ncorrespond to individual samples and columns to individual markers. Gates were defined 360 \nusing unstained populations. Percentages denote the fraction of marker-positive cells within 361 \nthe live singlet population. (B) Violin plots showing log-normalised mRNA expression  of 362 \ncanonical RGC markers POU4F family members ( POU4F1, POU4F2, POU4F3), ISL1, 363 \nSNCG, and THY1  across H9_RGC1, H9_RGC2, WAB_RGC2, WAB_RGC1. Each violin 364 \nrepresents the distribution of transcript abundance across individual cells. 365 \n  366 \nTo quantify the abundance of RGC-like cells within each enriched culture, flow cytometry 367 \nanalysis was performed using four markers, which in combination are indicative of RGC 368 \nidentity: POU4F (recognising POU4F family proteins), ISL1, SNCG, and THY1 ( Fig. 2A; 369 \nFig. S7). All samples demonstrated high proportions of POU4F+ cells (79.0 to 95.1%), 370 \nconsistent with efficient induction of sensory neurons/ RGC lineage identity. SNCG, a small 371 \ncytosolic protein highly enriched in RGCs [34], was used as an additional marker of RGC 372 \ndifferentiation. ISL1, a transcription factor expressed in RGCs and other inner retinal neurons 373 \n[35,36], showed greater variability across samples (18.0-58.0%), reflecting the known stage-374 \ndependent temporal dynamic expression of ISL1 during RGC differentiation [35]. SNCG 375 \nexpression was similarly high in H9_RGC1, H9_RGC2 and WAB_RGC2 (81.3-91.0%), but 376 \nwas markedly reduced in WAB_RGC1 (21.5%), consistent with its less organised neurite 377 \narchitecture. THY1, a surface protein historically used as an RGC antigen in the retina [37], 378 \ndisplayed the greatest variability, ranging from 3.7% in WAB_RGC2 to 29.3% in H9_RGC1. 379 \nThis variability aligns with the recognised instability of THY1 as a surface marker as it labels 380 \nonly a subset of RGCs and is also detectable in non-RGC retinal cell types [38]. Together, 381 \nthese data indicate the presence of RGC-associated marker expression across all samples, 382 \nwith some variability among individual markers.  383 \n All RGC markers used for flow-cytometry validation were also detected at the 384 \ntranscript level in the scRNA-seq dataset; however, the mRNA expression of these individual 385 \nmarkers varied substantially across all samples ( Fig. 2B; Fig. S8 ). POU4F1, POU4F2, and 386 \nPOU4F3 transcripts were detected in only a small fraction of cells across all samples, with 387 \nthe majority of cells exhibiting zero or near-zero expression. ISL1 mRNA was detected in a 388 \nsmall subset of cells primarily in H9_RGC1, while H9_RGC2, WAB_RGC2, and 389 \nW\nAB_RGC1 showed little to no detectable expression. SNCG  transcripts were detected in 390 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n18 \nsubsets of cells in H9_RGC1, H9_RGC2, and WAB_RGC2, with a distinct population 391 \nexhibiting elevated expression, whereas SNCG expression was minimal in WAB_RGC1, 392 \nwhere most cells showed near-zero transcript levels. THY1 transcripts were present in subsets 393 \nof cells in H9_RGC1 and H9_RGC2 but were largely absent in WAB_RGC samples, with 394 \nonly rare cells showing detectable expression. These findings indicate that reliance on flow 395 \ncytometry, through detection of stable or residual protein, may overestimate RGC abundance 396 \nrelative to scRNA-seq.  397 \n  398 \n  399 \n 400 \n  401 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n19 \n 402 \n 403 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n20 \nFigure 3. Single-cell transcriptomic profiling of RGC-enriched dissociated retinal 404 \norganoid cells. (A) UMAP visualisation showing the distribution of the 22 transcriptionally 405 \ndefined clusters identified in the integrated dataset. (B) Heatmap displaying z-scored average 406 \nlog-normalised RNA expression of different marker genes across clusters. Rows correspond 407 \nto marker genes grouped by major retinal cell classes, and columns represent the 22 clusters. 408 \nLeft annotation indicates the retinal cell type associated with each marker set, and the top 409 \nannotation reflects the assigned cell-type identity for each cluster. ( C) Stacked bar charts 410 \nshowing the proportional composition of annotated retinal cell types in each sample. ( D) 411 \nRidge plots illustrating pseudotime distributions for the major developmental lineages. Each 412 \nridge reflects the relative density of cells and the progression of cells along pseudotime 413 \nwithin a given lineage. ( E) Lineage tree reconstructed from pseudotime analysis, showing 414 \nhierarchical branching of retinal progenitor populations into downstream neuronal and 415 \nepithelial fates. Each node is labelled with its cluster identifier, with bracketed numbers 416 \nindicating the corresponding cluster IDs. RPC: retinal progenitor cell, RGC: retinal ganglion 417 \ncell, AC: amacrine cell, HC: horizontal cell, PR: photoreceptor, RPE: retinal pigment 418 \nepithelium. 419 \nIdentification and characterisation of 22 subpopulations 420 \nAcross the four samples, a total of 73,642 high-quality single cells were retained for 421 \ndownstream integrated analysis (Table S1 ). At a clustering resolution of 0.5, a total of 22 422 \ntranscriptionally distinct clusters were identified ( Fig. 3A). However, cluster 22 contained 423 \nfewer than 100 cells (n = 91) and was therefore excluded from downstream analyses. Based 424 \non marker expression, clusters were annotated into different cell types ( Fig. 3B ). Sample-425 \nlevel comparisons revealed notable differences in differentiation bias across the four datasets 426 \n(Fig. 3C). In H9_RGC1, RGCs constituted the largest population (45%), followed by 427 \namacrine cells (15%), retinal progenitor cells (14%), HOX-enriched cells (13%), and 428 \nhorizontal cells (7%). In H9_RGC2, HOX-enriched cells were the most abundant (30%), with 429 \nRGCs and amacrine cells each accounting for 19%, followed by retinal progenitor cells 430 \n(11%), and photoreceptor-committed cells (10%). WAB_RGC2 displayed 38% of cells 431 \nclassified as photoreceptor-committed cells and 27% as RGCs, alongside horizontal cells 432 \n(12%), other minor populations (8%), retinal progenitor cells (7%), and amacrine cells (6%). 433 \nWAB_RGC1 exhibited the highest proportion of retinal progenitor cells (47%), with RGCs 434 \n(10%), amacrine cells (10%), RPE (9%), photoreceptor-committed cells (7%), HOX-enriched 435 \ncluster (5%), and other populations (5%) (Table S2).  436 \nSix retinal progenitor cells clusters (4, 10, 15, 17, 18, 21) were annotated based on the 437 \nexpression of canonical retinal progenitor cell  markers such as VSX2, along with key 438 \ntranscription factors regulating neuronal fate, including PAX6 and SOX2 [39–41]. They also 439 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n21 \nexpressed LHX2, which is essential for maintaining open chromatin during retinogenesis and 440 \nfor gliogenesis [42,43]. Cell cycle-related genes were unevenly distributed among progenitor 441 \nsubpopulations, with the G2/M phase marker MKI67 showing predominant expression in 442 \nclusters 15 and 17.  443 \nRGC clusters were annotated based on the expression of canonical markers including 444 \nPOU4F2, ISL1, ISL2, EOMES, SNCG, THY1, NEFL, NEFM, and RBPMS [5,44–47]. Using 445 \nthese markers, four transcriptionally distinct RGC subpopulations were identified ( Fig. 3B), 446 \nAll of these clusters (2, 3, 11, and 14) expressed POU4F2 . Cluster 3 represented the earliest 447 \nRGC-fated state, characterised by high expression of ATOH7, HES6, and DLL3. Clusters 2 448 \nand 14 showed strong co-expression of POU4F2, ISL1, RBPMS, NEFL, NEFM, SNCG, 449 \nTHY1, SLC17A6, and GAP43, indicative of maturing RGCs. Clusters 11 and 14 exhibited 450 \npartial expression of horizontal cell-associated markers such as  ONECUT1 and ONECUT2 , 451 \nindicating transcriptional heterogeneity within the enriched RGC population.  452 \nTwo amacrine cell clusters (5 and 7) were enriched with an inhibitory neuron marker 453 \nGAD1 [27], while a horizontal cell cluster (6) expressed PROX1, ONECUT1, ONECUT2 and 454 \nGJA10 [48–50]. Photoreceptor-committed cells identity was defined using established lineage 455 \nmark\ners, including OTX2, CRX, THRB, and NRL, together with rod-specific markers ( RHO, 456 \nGNAT1, NR2E3) and cone-specific markers ( ARR3, GNAT2, PDE6H) [51–55]. Four clusters 457 \n(8, 9, 13 and 20) showed strong expression of OTX2 and CRX , confirming photoreceptor 458 \nlineage commitment. Among these, cluster 9 displayed the most advanced phototransduction 459 \nprogramme, co-expressing cone-specific markers ( ARR3, GNAT2, PDE6H) together with 460 \nrod-associated genes ( NRL, PDE6B, GNAT1), indicating a population of mixed rod-cone 461 \nphotoreceptor precursors. Additionally, cluster 16 showed the expression of RPE 65, MITF, 462 \nTYR, TYR and RDH5, classical markers of RPE [56].  463 \n       Cluster 19 exhibited a stress-associated transcriptional profile characterised by high 464 \nexpression of FOS, JUN, ATF3, DDIT3, CDKN1A, and HSPA1B [57–61]. This cluster was 465 \nannotated as a Multilineage (stressed) population, and it was less than 5% of the total cells 466 \nacross all samples respectively. Differential expression analysis revealed induction of genes 467 \nassociated with apoptosis initiation, cellular stress signalling and early cell-death pathways, 468 \nalthough no coherent set of canonical lineage markers was detected. To determine the 469 \nunderlying lineage composition masked by stress-induced transcriptional reprogramming, 470 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n22 \nlabel-transfer analysis was performed (Table 4 ). This analysis revealed that cluster 19 471 \ncomprised mainly of retinal progenitor cells, followed by “Other” cell types, RPE, HOX-472 \nenriched cells, and amacrine cells (Table 4). 473 \nTable 4: Lineage composition of the Multilineage (stressed) cluster identified by label-474 \ntransfer analysis. 475 \nFamily # of cells Percentage of cells (%) \nRetinal progenitor cell 483 35.3 \nRGC 43 3.1 \nAmacrine cell 125 9.1 \nHorizontal cell 14 1.0 \nPhotoreceptor-committed cell 28 2.0 \nRPE 144 10.5 \nOther (HOX-enriched) 136 9.9 \nOther 397 29.0 \n 476 \nCluster 1 was annotated as HOX-gene-enriched cells, marked by high expression of 477 \nHOX-related genes such as HOXB5-8. These cells also expressed neuronal structural genes 478 \nsuch as NEFL, NEFM and ELAVL4  but lacked retinal identity markers, indicating the 479 \nformation of off-target posterior neural cell types not belonging to anterior retinal tissue. 480 \nHOX-enriched populations were most abundant in the two H9-derived samples. In addition, 481 \ncluster 12 was annotated as “Other” as it did not exhibit retinal cell markers.  482 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n23 \nPseudotime analysis reveals progressive retinal lineage trajectories 483 \nTo reconstruct differentiation pathways across retinal lineages, we performed pseudotime and 484 \nlineage trajectory analysis on the Harmony-integrated dataset comprising all four samples. 485 \nClusters annotated as “Multilineage (stressed)” and “Other” were excluded as they exhibited 486 \nmixed lineage marker expression and/or off-target transcriptional programs, rather than a 487 \nsingle coherent retinal identity. Inclusion of these cells could distort transcriptional similarity 488 \nrelationships and lead to artificial trajectory splits that do not reflect genuine developmental 489 \nlineage decisions. Cells were ordered along developmental trajectories based on 490 \ntranscriptional similarity. Ten major lineages were identified, corresponding to retinal 491 \nprogenitor, RGC, amacrine, horizontal, photoreceptor, and RPE fates, each branching from 492 \nthe retinal progenitor cell cluster 17, which showed the highest expression of the proliferation 493 \nmarker MKI67 (Fig. 3D). The earliest pseudotime positions were occupied by the RPC1-494 \nRPC3 populations, which together formed the central developmental trunk. RPC4 495 \npredominantly gave rise to two RGC lineages and to RPC5, which forms the root of all other 496 \nmajor retinal cell lineages, including an additional retinal progenitor cell lineage, an amacrine 497 \ncell lineage, one horizontal lineage, two photoreceptor lineages, and one RPE lineage ( Fig. 498 \n3E).  499 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n24 \n 500 \nFigure 4. Subclustering and molecular characterisation of RGC subpopulations.  ( A) 501 \nUMAP showing the clusters (in purple) selected for subsequent reclustering of the RGC 502 \nsubpopulation. ( B) UMAP showing the seven distinct clusters identified. ( C) Heatmap 503 \nshowing z-scored average log-normalised RNA expression of RGC subtypes markers. Rows 504 \nrepresent marker genes grouped by major RGC subtypes, and columns represent the seven 505 \nclusters. The left annotation indicates the subtype associated with each marker set. (D ) 506 \nUMAP visualisation of RGC subtype marker expression using log-normalised RNA 507 \nexpression values. For each marker, cells with detectable expression (log-normalised 508 \nexpression > 0) are highlighted in purple, while non-expressing cells are shown in grey. Each 509 \npanel represents a single marker gene, with gene symbols displayed at the top of each plot. 510 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n25 \n 511 \nTranscriptional characterisation of RGC subtypes 512 \nTo further resolve the molecular diversity within the RGC lineage, we re-clustered cells from 513 \nRGC-associated clusters (2, 3, 11 and 14) at a resolution of 0.3, resulting in seven 514 \ntranscriptionally distinct RGC subclusters ( Fig. 4A , B). Canonical human RGC markers 515 \n[7,29,31–33] (Table 2) were detected across these subclusters, including markers for parasol, 516 \nmidget, DSGC, OSGC, and large sparse RGCs  (Fig. 4C ), suggesting that the cultures 517 \ngenerate multiple RGC subtypes. Most subclusters showed overlapping expression of several 518 \nRGC subtype marker genes, suggesting incomplete transcriptional segregation. Although 519 \nclusters 1 and 4 showed elevated OPN4 signal (Fig. 4C), it was detected in only 49 cells for 520 \ncluster 1, and 11 cells in cluster 4 in the log-normalised RNA matrix. As this proportion fell 521 \nbelow the minimum detection threshold (≥ 5% of cells expressing the marker), OPN4 was not 522 \nidentified in the differential expression list (Fig. 4C, D).  For each selected marker gene, cells 523 \nwith normalised expression greater than zero were considered positive, and Harmony-derived 524 \nUMAP embeddings were used to visualise the spatial distribution of marker-positive cells 525 \n(Fig. 4D). 526 \nGiven that OPN4 marks a rare subtype of RGC, we assessed whether additional rare 527 \nmelanopsin-expressing cells were missed due to normalisation and subsequent subclustering. 528 \nExamination of the unnormalised, SoupX-adjusted RNA count matrix, where low-abundance 529 \ntranscripts remain easier to detect than in SCTransform-normalised data [62], identified 128 530 \nOPN4+ cells, comprising 120 cells with low transcript abundance and 8 cells with higher 531 \nabundance (Fig. S9 and Table S3 ). Mapping these OPN4+ cells onto the Harmony-derived 532 \nUMAP embedding revealed that approximately half were located within the horizontal 533 \ncluster (cluster 6 of the full integrated RGC-enriched retinal organoid dataset; Fig. 3), while 534 \n51 cells were located within the RGC cluster (cluster 2; Fig. 3). 535 \nDiscussion 536 \nThis work provides an in-depth analysis of RGC differentiation within hPSC-derived retinal 537 \norganoid cultures using complementary flow cytometry and single-cell transcriptomic 538 \napproaches. Discrepancies between protein- and transcript-based readouts highlight the 539 \nlimitations of marker-based assessments and underscore the importance of transcriptomic 540 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n26 \nbenchmarks for defining RGC identity in heterogeneous cultures. Across samples, protein-541 \nbased assays suggested robust RGC enrichment, whereas transcriptomic profiling revealed a 542 \nsmaller RGC population alongside significant contributions from non-RGC neuronal 543 \npopulations. For instance, over 80% of cells were deemed POU4F+ and SNCG+ in 544 \nH9_RGC2, yet scRNA-seq identified more than one-third of cells belonging to HOX-545 \nenriched cell clusters. These HOX-enriched clusters may represent posterior CNS-like 546 \nneurons, such as spinal interneurons or hindbrain sensory populations, which are known to 547 \ntransiently express POU4F family proteins during development [63]. Similarly, WAB_RGC2 548 \nshowed high proportions of POU4F+ cells and SNCG+ cells by flow cytometry, whilst 549 \nscRNA-seq showed that the major cell types within this differentiation were from the 550 \nphotoreceptor lineage. These can also transiently express SNCG [64], which is widely 551 \nexpressed across multiple peripheral and central nervous system neuronal populations [65].  552 \nISL1 expression further illustrates the limitations of protein-based classification, as it 553 \nvaries with cellular maturation and is not restricted to RGCs, being detectable in other 554 \nneuronal populations [66]. This likely contributes to the variable proportion of ISL1+ cells 555 \nobserved across samples. In contrast, THY1+ cells consistently represented the smallest 556 \nfraction, reflecting the dynamic nature of THY1 expression, its downregulation in stressed 557 \nRGCs [67], and its presence in subsets of amacrine cells [68]. In H9_RGC1 and 558 \nWAB_RGC2, scRNA-seq identified a greater proportion of RGCs than suggested by flow 559 \ncytometry, highlighting the limited reliability of THY1 as a commonly used standalone 560 \nmarker for estimating RGC content [10]. These findings indicate that RGC-associated 561 \nproteins can be detected outside the canonical RGC lineage, whereas scRNA-seq more 562 \naccurately excludes non-RGC cells. This discrepancy reflects the broad and often transient 563 \nexpression of commonly used RGC markers during early neurogenesis, together with their 564 \npersistence in off-target neuronal lineages. As a result, protein marker positivity alone can 565 \noverestimate RGC identity when lineage resolution is incomplete. Future flow cytometry 566 \napproaches could improve specificity by combining POU4F and SNCG with exclusion 567 \nmarkers for photoreceptors and/or GAD1 for amac rine cells, thereby ensuring that marker-568 \npositive populations more faithfully represent RGCs. 569 \nMethodological differences between flow cytometry and scRNA-seq further 570 \ncontribute to these contrasting readouts. Flow cytometry enriches larger and healthier 571 \ndissociated cells through gating and is highly sensitive to stable or residual protein, 572 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n27 \npotentially enriching early neurons that transiently express RGC-associated markers, as 573 \ndiscussed above. In contrast, scRNA-seq captures a broader and less selectively filtered 574 \npopulation, including fragile or transcriptionally immature cells, and classifies identity based 575 \non integrated gene expression programs rather than individual markers. Although transcript 576 \ndropout is an inherent limitation of scRNA-seq with typical protocols capturing only ~10-577 \n20% of transcripts [69–71], reliance on multi-gene signatures reduces the impact of false 578 \nnegatives [72] and provides a more conservative and lineage-resolved definition of RGC 579 \nidentity.  580 \nSingle-cell analysis resolved multiple RGC transcriptional states, alongside relatively 581 \nhigher proportions of retinal progenitor cells, off-target posterior neural populations, and 582 \nphotoreceptor-committed cells. Rather than forming a uniform population, RGCs in enriched 583 \ncultures occupied a continuum of transcriptional states. Pseudotime reconstruction supported 584 \nthe coexistence of early, intermediate, and maturing RGC populations. Multiple RGC 585 \nsubtypes were also identified. Rare subtypes, including melanopsin-expressing RGCs, were 586 \ndetected at low abundance and were localised to specific transcriptional clusters, consistent 587 \nwith genuine low-level expression. However, these populations were particularly sensitive to 588 \nanalytical thresholds and normalisation strategies, underscoring the limitations of single-cell 589 \napproaches for resolving rare neuronal populations and highlighting the importance of 590 \ncautious interpretation when assessing low-abundance transcripts in developing systems.  591 \nIn the presence of multiple developmental states within the culture, the observed 592 \noverlap between RGC transcriptional programs and markers associated with other early-born 593 \nretinal neurons can be understood in the context of retinal development. RGCs, amacrine and 594 \nhorizontal cells arise from shared ATOH7 + progenitors, and fate commitment during early 595 \nretinogenesis occurs gradually rather than through abrupt transitions [73]. Transitional cells 596 \nmay therefore transiently activate transcriptional programs associated with multiple retina 597 \nlineages before terminal differentiation.  Consistent with this, developing RGCs or RGC-like 598 \nneurons have been found to express markers associated with amacine and horizontal cells 599 \n[74,75]. Hence, when interpreting the scRNAseq analysis, low-level or partial expression of 600 \nlineage-associated markers should not be interpreted as definitive evidence of fate switching, 601 \nbut rather as a reflection of developmental immaturity or incomplete fate resolution. 602 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n28 \nThis developmental heterogeneity complicates marker-based assessments of RGC 603 \nenrichment. As enriched cultures contain early RGCs, maturing neurons, and diverse subtype 604 \nidentities, no single surface or intracellular marker can reliably distinguish true RGCs from 605 \ntransient intermediates or off-target neurons with overlapping antigen expression. This 606 \nlimitation is particularly relevant for THY1-based enrichment strategies, which are 607 \ncommonly used for immunopanning and positive selection [9–12]. In our study, THY1 608 \nexpression was variable and lacked specificity, being detected across multiple retinal and 609 \nnon-retinal neuronal populations. These findings suggest that THY1-based approaches are 610 \nlikely to isolate a heterogeneous mixture rather than a well-defined RGC population and that 611 \nmore stringent multi-marker or transcriptomically informed selection strategies will be 612 \nnecessary to obtain high-fidelity RGC enrichment. However, identifying surface markers that 613 \nare both specific and stable across RGC subtypes remains an important and unresolved 614 \nchallenge. 615 \nAnother feature of our dataset is the presence of HOX-enriched cell populations, 616 \nparticularly in hESC-derived organoids. From a developmental perspective, this is 617 \nunexpected, as the retina originates from the anterior neuroectoderm, whereas HOX genes 618 \npattern posterior regions of the CNS [76,77] . However, our findings are consistent with 619 \nprevious studies reporting upregulation of HOX gene programs in hPSC-derived retinal 620 \norganoids [18,78,79]. Although the abundance and composition of these clusters vary across 621 \nstudies, the recurrent detection of posterior HOX -expressing cells suggests that off-target 622 \nneural identities represent a common feature of current retinal organoid differentiation 623 \nprotocols. Modulating retinoid acid availability during early differentiation such as the use of 624 \nB27 supplement without vitamin A can shift neural identity toward anterior domains [80,81], 625 \nand could potentially reduce the proportion of HOX-enriched cells.  626 \nWhile this protocol yielded variable proportions of RGCs across samples, the 627 \nresulting RGC content (19-45%), excluding the poorly differentiated retinal organoids from 628 \nWAB_RGC1, is comparable to previously reported yields from hPSC-to-RGC differentiation 629 \nstrategies validated by scRNA-seq. Earlier studies reported RGC proportions of around 12% 630 \nusing a 2D differentiation protocol and 17% in RGC-enriched retinal organoid systems 631 \n[18,83], indicating that the overall efficiency observed here is above the range of 632 \ntranscriptomically validated approaches. Nevertheless, sample-to-sample variability indicates 633 \nopportunities for further optimisation, as early exclusion of poorly patterned organoids and 634 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n29 \nrefinement of dissociation timing may reduce retention of retinal progenitor and 635 \nphotoreceptor-committed cells [14,84].  636 \nLimitations 637 \nA limitation of this study is that RGC subtype classification is based on transcriptional 638 \nprofiles rather than functional properties. Classical RGC subtypes are defined by 639 \nmorphology, connectivity and electrophysiological behaviour [33], which cannot be resolved 640 \nby scRNA-seq alone. As a result, transcriptionally defined clusters may not map directly onto 641 \nestablished functional RGC classes and may instead reflect developmental states, stress 642 \nresponses or culture-induced divergence. Future studies integrating transcriptomics with 643 \nelectrophysiology, connectivity mapping and functional assays will be required to directly 644 \nmolecular identity with RGC function.  645 \n  646 \nData availability 647 \nAll data have been deposited in the ArrayExpress database 648 \n(https://www.ebi.ac.uk/arrayexpress). 649 \n 650 \nSupplementary Information 651 \nNine supplementary figures, three supplementary tables, and an additional Excel file 652 \ncontaining differential gene expression results with an adjusted p-value < 0.05. 653 \n 654 \nAcknowledgments 655 \nThe authors acknowledge Dr. Magdaline Sakkas of the Melbourne Cytometry Platform, The 656 \nUniversity of Melbourne, for her instruction on using the Beckman Coulter CytoFLEX LX 657 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n30 \nflow cytometer, and Dr. Peter Lau of the Australian Genome Research Facility for his 658 \nassistance in processing samples for scRNA-seq analysis. 659 \n 660 \nAuthors’ Contributions  661 \nConceptualisation: all authors; Methodology: all authors; Investigation: J.Y.W.M.; Data 662 \nanalysis: J.Y.W.M.; Writing original draft: J.Y.W.M., A.P.; Writing review & editing, all 663 \nauthors; Funding Acquisition: all authors; Supervision and project administration: A.P. 664 \n 665 \nFunding 666 \nThis research was supported by funding from PYC therapeutics, a University of Melbourne 667 \nDepartment of Anatomy and Physiology Early Career Seeding Grants (JYWM) and a Dame 668 \nKate Campbell Fellowship (AP).  669 \n 670 \nMaterials & Correspondence. Requests for materials and correspondence should be 671 \naddressed to Jessica Ma and Alice Pébay. 672 \n 673 \nCompeting Interests  674 \nA.P. is a scientific advisor to PYC Therapeutics, and a director and shareholder of CellTellus 675 \nPty Ltd.  676 \n 677 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n31 \nReferences 678 \n1. Vision Loss Expert Group of the Global Burden of Disease Study, GBD 2019 Blindness 679 \nand Vision Impairment Collaborators. Global estimates on the number of people blind or 680 \nvisually impaired by glaucoma: A meta-analysis from 2000 to 2020. Eye (Lond). 681 \n2024;38:2036–46. 682 \n2. Luo X, Salgueiro Y, Beckerman SR, Lemmon VP, Tsoulfas P, Park KK. Three-683 \ndimensional evaluation of retinal ganglion cell axon regeneration and pathfinding in whole 684 \nmouse tissue after injury. Exp Neurol. 2013;247:653–62. 685 \n3. Curcio CA, Allen KA. Topography of ganglion cells in human retina. J Comp Neurol. 686 \n1990;300:5–25. 687 \n4. Curcio CA, Millican CL, Allen KA, Kalina RE. Aging of the human photoreceptor mosaic: 688 \nevidence for selective vulnerability of rods in central retina. Invest Ophthalmol Vis Sci. 689 \n1993;34:3278–96. 690 \n5. Rheaume BA, Jereen A, Bolisetty M, Sajid MS, Yang Y, Renna K, et al. Single cell 691 \ntranscriptome profiling of retinal ganglion cells identifies cellular subtypes. Nat Commun. 692 \n2018;9:2759. 693 \n6. Tran NM, Shekhar K, Whitney IE, Jacobi A, Benhar I, Hong G, et al. Single-Cell Profiles 694 \nof Retinal Ganglion Cells Differing in Resilience to Injury Reveal Neuroprotective Genes. 695 \nNeuron. 2019;104:1039–55.e12. 696 \n7. Peng Y-R, Shekhar K, Yan W, Herrmann D, Sappington A, Bryman GS, et al. Molecular 697 \nClassification and Comparative Taxonomics of Foveal and Peripheral Cells in Primate 698 \nRetina. Cell. 2019;176:1222–37.e22. 699 \n8. Morgan JE, Bevan RJ, Cimaglia G. Retinal Ganglion Cell Subtypes and Their 700 \nVulnerability in Glaucoma. Methods Mol Biol. 2025;2858:191–205. 701 \n9. Chavali VRM, Haider N, Rathi S, Vrathasha V, Alapati T, He J, et al. Dual SMAD 702 \ninhibition and Wnt inhibition enable efficient and reproducible differentiations of induced 703 \npluripotent stem cells into retinal ganglion cells. Sci Rep. 2020;10:11828. 704 \n10. Rabesandratana O, Chaffiol A, Mialot A, Slembrouck-Brec A, Joffrois C, Nanteau C, et 705 \nal. Generation of a Transplantable Population of Human iPSC-Derived Retinal Ganglion 706 \nCells. Front Cell Dev Biol. 2020;8:585675. 707 \n11. VanderWall KB, Huang K-C, Pan Y, Lavekar SS, Fligor CM, Allsop AR, et al. Retinal 708 \nGanglion Cells With a Glaucoma OPTN(E50K) Mutation Exhibit Neurodegenerative 709 \nPhenotypes when Derived from Three-Dimensional Retinal Organoids. Stem Cell Reports. 710 \n2020;15:52–66. 711 \n12. Edo A, Sugita S, Futatsugi Y, Sho J, Onishi A, Kiuchi Y, et al. Capacity of Retinal 712 \nGanglion Cells Derived from Human Induced Pluripotent Stem Cells to Suppress T-Cells. Int 713 \nJ Mol Sci [Internet]. 2020;21. Available from: http://dx.doi.org/10.3390/ijms21217831 714 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n32 \n13. Eiraku M, Takata N, Ishibashi H, Kawada M, Sakakura E, Okuda S, et al. Self-organizing 715 \noptic-cup morphogenesis in three-dimensional culture. Nature. 2011;472:51–6. 716 \n14. Harkin J, Peña KH, Gomes C, Hernandez M, Lavekar SS, So K, et al. A highly 717 \nreproducible and efficient method for retinal organoid differentiation from human pluripotent 718 \nstem cells. Proc Natl Acad Sci U S A. 2024;121:e2317285121. 719 \n15. Wagstaff EL, Ten Asbroek ALMA, Ten Brink JB, Jansonius NM, Bergen AAB. An 720 \nalternative approach to produce versatile retinal organoids with accelerated ganglion cell 721 \ndevelopment. Sci Rep. 2021;11:1101. 722 \n16. Mu X, Klein WH. A gene regulatory hierarchy for retinal ganglion cell specification and 723 \ndifferentiation. Semin Cell Dev Biol. 2004;15:115–23. 724 \n17. Sridhar A, Hoshino A, Finkbeiner CR, Chitsazan A, Dai L, Haugan AK, et al. Single-Cell 725 \nTranscriptomic Comparison of Human Fetal Retina, hPSC-Derived Retinal Organoids, and 726 \nLong-Term Retinal Cultures. Cell Rep. 2020;30:1644–59.e4. 727 \n18. Daniszewski M, Senabouth A, Liang HH, Han X, Lidgerwood GE, Hernández D, et al. 728 \nRetinal ganglion cell-specific genetic regulation in primary open-angle glaucoma. Cell 729 \nGenom. 2022;2:100142. 730 \n19. Gudiseva HV, Vrathasha V, He J, Bungatavula D, O’Brien JM, Chavali VRM. Single 731 \nCell Sequencing of Induced Pluripotent Stem Cell Derived Retinal Ganglion Cells (iPSC-732 \nRGC) Reveals Distinct Molecular Signatures and RGC Subtypes. Genes (Basel) [Internet]. 733 \n2021;12. Available from: http://dx.doi.org/10.3390/genes12122015 734 \n20. Young MD, Behjati S. SoupX removes ambient RNA contamination from droplet-based 735 \nsingle-cell RNA sequencing data. Gigascience [Internet]. 2020;9. Available from: 736 \nhttp://dx.doi.org/10.1093/gigascience/giaa151 737 \n21. Germain P-L, Lun A, Garcia Meixide C, Macnair W, Robinson MD. Doublet 738 \nidentification in single-cell sequencing data using. F1000Res. 2021;10:979. 739 \n22. Hafemeister C, Satija R. Normalization and variance stabilization of single-cell RNA-seq 740 \ndata using regularized negative binomial regression. Genome Biol. 2019;20:296. 741 \n23. McInnes L, Healy J, Saul N, Großberger L. UMAP: Uniform Manifold Approximation 742 \nand Projection. J Open Source Softw. 2018;3:861. 743 \n24. Traag VA, Waltman L, van Eck NJ. From Louvain to Leiden: guaranteeing well-744 \nconnected communities. Scientific Reports. 2019;9:1–12. 745 \n25. Oh S, Abdelnabi J, Al-Dulaimi R, Aggarwal A, Ramos M, Davis S, et al. HGNChelper: 746 \nidentification and correction of invalid gene symbols for human and mouse. F1000Res. 747 \n2020;9:1493. 748 \n26. Durinck S, Spellman PT, Birney E, Huber W. Mapping identifiers for the integration of 749 \ngenomic datasets with the R/Bioconductor package biomaRt. Nat Protoc. 2009;4:1184–91. 750 \n27. Wang SK, Nair S, Li R, Kraft K, Pampari A, Patel A, et al. Single-cell multiome of the 751 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n33 \nhuman retina and deep learning nominate causal variants in complex eye diseases. Cell 752 \nGenom [Internet]. 2022;2. Available from: http://dx.doi.org/10.1016/j.xgen.2022.100164 753 \n28. Xu Z, Liao X, Li N, Zhou H, Li H, Zhang Q, et al. A Single-Cell Transcriptome Atlas of 754 \nthe Human Retinal Pigment Epithelium. Front Cell Dev Biol. 2021;9:802457. 755 \n29. Li J, Wang J, Ibarra IL, Cheng X, Luecken MD, Lu J, et al. Integrated multi-omics single 756 \ncell atlas of the human retina [Internet]. Res Sq. 2023. Available from: 757 \nhttp://dx.doi.org/10.21203/rs.3.rs-3471275/v1 758 \n30. Street K, Risso D, Fletcher RB, Das D, Ngai J, Yosef N, et al. Slingshot: cell lineage and 759 \npseudotime inference for single-cell transcriptomics. BMC Genomics. 2018;19:477. 760 \n31. Langer KB, Ohlemacher SK, Phillips MJ, Fligor CM, Jiang P, Gamm DM, et al. Retinal 761 \nGanglion Cell Diversity and Subtype Specification from Human Pluripotent Stem Cells. Stem 762 \nCell Reports. 2018;10:1282–93. 763 \n32. Yan W, Peng Y-R, van Zyl T, Regev A, Shekhar K, Juric D, et al. Cell Atlas of The 764 \nHuman Fovea and Peripheral Retina. Scientific Reports. 2020;10:1–17. 765 \n33. Kim US, Mahroo OA, Mollon JD, Yu-Wai-Man P. Retinal Ganglion Cells-Diversity of 766 \nCell Types and Clinical Relevance. Front Neurol. 2021;12:661938. 767 \n34. Surgucheva I, Weisman AD, Goldberg JL, Shnyra A, Surguchov A. Gamma-synuclein as 768 \na marker of retinal ganglion cells. Mol Vis. 2008;14:1540–8. 769 \n35. Pan L, Deng M, Xie X, Gan L. ISL1 and BRN3B co-regulate the differentiation of 770 \nmurine retinal ganglion cells. Development. 2008;135:1981–90. 771 \n36. Wu F, Kaczynski TJ, Sethuramanujam S, Li R, Jain V, Slaughter M, et al. Two 772 \ntranscription factors, Pou4f2 and Isl1, are sufficient to specify the retinal ganglion cell fate. 773 \nProc Natl Acad Sci U S A. 2015;112:E1559–68. 774 \n37. Martersteck EM, Hirokawa KE, Evarts M, Bernard A, Duan X, Li Y, et al. Diverse 775 \nCentral Projection Patterns of Retinal Ganglion Cells. Cell Rep. 2017;18:2058–72. 776 \n38. Rodriguez AR, de Sevilla Müller LP, Brecha NC. The RNA binding protein RBPMS is a 777 \nselective marker of ganglion cells in the mammalian retina. J Comp Neurol. 2014;522:1411–778 \n43. 779 \n39. Zou C, Levine EM. Vsx2 controls eye organogenesis and retinal progenitor identity via 780 \nhomeodomain and non-homeodomain residues required for high affinity DNA binding. PLoS 781 \nGenet. 2012;8:e1002924. 782 \n40. Sun J, Zhao Y, McGreal R, Cohen-Tayar Y, Rockowitz S, Wilczek C, et al. Pax6 783 \nassociates with H3K4-specific histone methyltransferases Mll1, Mll2, and Set1a and 784 \nregulates H3K4 methylation at promoters and enhancers. Epigenetics Chromatin. 2016;9:37. 785 \n41. Taranova OV, Magness ST, Fagan BM, Wu Y, Surzenko N, Hutton SR, et al. SOX2 is a 786 \ndose-dependent regulator of retinal neural progenitor competence. Genes Dev. 787 \n2006;20:1187–202. 788 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n34 \n42. de Melo J, Zibetti C, Clark BS, Hwang W, Miranda-Angulo AL, Qian J, et al. Lhx2 Is an 789 \nEssential Factor for Retinal Gliogenesis and Notch Signaling. J Neurosci. 2016;36:2391–405. 790 \n43. Zibetti C, Liu S, Wan J, Qian J, Blackshaw S. Epigenomic profiling of retinal progenitors 791 \nreveals LHX2 is required for developmental regulation of open chromatin. Commun Biol. 792 \n2019;2:142. 793 \n44. Clark BS, Stein-O’Brien GL, Shiau F, Cannon GH, Davis-Marcisak E, Sherman T, et al. 794 \nSingle-Cell RNA-Seq Analysis of Retinal Development Identifies NFI Factors as Regulating 795 \nMitotic Exit and Late-Born Cell Specification. Neuron. 2019;102:1111–26.e5. 796 \n45. Freeman NE, Templeton JP, Orr WE, Lu L, Williams RW, Geisert EE. Genetic networks 797 \nin the mouse retina: growth associated protein 43 and phosphatase tensin homolog network. 798 \nMol Vis. 2011;17:1355–72. 799 \n46. Hu Y, Wang X, Hu B, Mao Y, Chen Y, Yan L, et al. Dissecting the transcriptome 800 \nlandscape of the human fetal neural retina and retinal pigment epithelium by single-cell 801 \nRNA-seq analysis. PLoS Biol. 2019;17:e3000365. 802 \n47. Peirson SN, Oster H, Jones SL, Leitges M, Hankins MW, Foster RG. Microarray analysis 803 \nand functional genomics identify novel components of melanopsin signaling. Curr Biol. 804 \n2007;17:1363–72. 805 \n48. Dyer MA, Livesey FJ, Cepko CL, Oliver G. Prox1 function controls progenitor cell 806 \nproliferation and horizontal cell genesis in the mammalian retina. Nat Genet. 2003;34:53–8. 807 \n49. Hahn J, Monavarfeshani A, Qiao M, Kao AH, Kölsch Y, Kumar A, et al. Evolution of 808 \nneuronal cell classes and types in the vertebrate retina. Nature. 2023;624:415–24. 809 \n50. Shen B, Fang T, Dai M, Jones G, Zhang S. Independent losses of visual perception genes 810 \nGja10 and Rbp3 in echolocating bats (Order: Chiroptera). PLoS One. 2013;8:e68867. 811 \n51. Milam AH, Rose L, Cideciyan AV, Barakat MR, Tang W-X, Gupta N, et al. The nuclear 812 \nreceptor NR2E3 plays a role in human retinal photoreceptor differentiation and degeneration. 813 \nProc Natl Acad Sci U S A. 2002;99:473–8. 814 \n52. Miyamoto M, Aoki M, Hirai K, Sugimoto S, Kawasaki K, Imai R. A nonsense mutation 815 \nin Gnat1, encoding the alpha subunit of rod transducin, in spontaneous mouse models of 816 \nretinal dysfunction. Exp Eye Res. 2010;90:63–9. 817 \n53. Renninger SL, Gesemann M, Neuhauss SCF. Cone arrestin confers cone vision of high 818 \ntemporal resolution in zebrafish larvae. Eur J Neurosci. 2011;33:658–67. 819 \n54. Ronning KE, Allina GP, Miller EB, Zawadzki RJ, Pugh EN Jr, Herrmann R, et al. Loss of 820 \ncone function without degeneration in a novel Gnat2 knock-out mouse. Exp Eye Res. 821 \n2018;171:111–8. 822 \n55. Shimizu-Matsumoto A, Itoh K, Inazawa J, Nishida K, Matsumoto Y, Kinoshita S, et al. 823 \nIsolation and chromosomal localization of the human cone cGMP phosphodiesterase gamma 824 \ncDNA (PDE6H). Genomics. 1996;32:121–4. 825 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n35 \n56. Plaza Reyes A, Petrus-Reurer S, Padrell Sánchez S, Kumar P, Douagi I, Bartuma H, et al. 826 \nIdentification of cell surface markers and establishment of monolayer differentiation to 827 \nretinal pigment epithelial cells. Nat Commun. 2020;11:1609. 828 \n57. Kueng-Hitz N, Grimm C, Lansel N, Hafezi F, He L, Fox DA, et al. The retina of c-fos-/- 829 \nmice: electrophysiologic, morphologic and biochemical aspects. Invest Ophthalmol Vis Sci. 830 \n2000;41:909–16. 831 \n58. Fernandes KA, Harder JM, Kim J, Libby RT. JUN regulates early transcriptional 832 \nresponses to axonal injury in retinal ganglion cells. Exp Eye Res. 2013;112:106–17. 833 \n59. Syc-Mazurek SB, Fernandes KA, Wilson MP, Shrager P, Libby RT. Together JUN and 834 \nDDIT3 (CHOP) control retinal ganglion cell death after axonal injury. Mol Neurodegener. 835 \n2017;12:71. 836 \n60. Piri N, Kwong JMK, Gu L, Caprioli J. Heat shock proteins in the retina: Focus on HSP70 837 \nand alpha crystallins in ganglion cell survival. Prog Retin Eye Res. 2016;52:22–46. 838 \n61. Lazzarini R, Nicolai M, Pirani V, Mariotti C, Di Primio R. Effects of senescent secretory 839 \nphenotype acquisition on human retinal pigment epithelial stem cells. Aging (Albany NY). 840 \n2018;10:3173–84. 841 \n62. Lu S, Yang J, Yan L, Liu J, Wang JJ, Jain R, et al. Transcriptome size matters for single-842 \ncell RNA-seq normalization and bulk deconvolution. Nat Commun. 2025;16:1246. 843 \n63. Nishida K, Matsumura S, Uchida H, Abe M, Sakimura K, Badea TC, et al. Brn3a controls 844 \nthe soma localization and axonal extension patterns of developing spinal dorsal horn neurons. 845 \nPLoS One. 2023;18:e0285295. 846 \n64. Dorgau B, Collin J, Rozanska A, Zerti D, Unsworth A, Crosier M, et al. Single-cell 847 \nanalyses reveal transient retinal progenitor cells in the ciliary margin of developing human 848 \nretina. Nat Commun. 2024;15:3567. 849 \n65. Morato Torres CA, Wassouf Z, Zafar F, Sastre D, Outeiro TF, Schüle B. The Role of 850 \nAlpha-Synuclein and Other Parkinson’s Genes in Neurodevelopmental and 851 \nNeurodegenerative Disorders. Int J Mol Sci [Internet]. 2020;21. Available from: 852 \nhttp://dx.doi.org/10.3390/ijms21165724 853 \n66. Elshatory Y, Deng M, Xie X, Gan L. Expression of the LIM-homeodomain protein Isl1 in 854 \nthe developing and mature mouse retina. J Comp Neurol. 2007;503:182–97. 855 \n67. Huang W, Fileta J, Guo Y, Grosskreutz CL. Downregulation of Thy1 in retinal ganglion 856 \ncells in experimental glaucoma. Curr Eye Res. 2006;31:265–71. 857 \n68. Raymond ID, Vila A, Huynh U-CN, Brecha NC. Cyan fluorescent protein expression in 858 \nganglion and amacrine cells in a thy1-CFP transgenic mouse retina. Mol Vis. 2008;14:1559–859 \n74. 860 \n69. Riemondy KA, Ransom M, Alderman C, Gillen AE, Fu R, Finlay-Schultz J, et al. 861 \nRecovery and analysis of transcriptome subsets from pooled single-cell RNA-seq libraries. 862 \nNucleic Acids Res. 2019;47:e20. 863 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint \n\n36 \n70. Bacher R, Kendziorski C. Design and computational analysis of single-cell RNA-864 \nsequencing experiments. Genome Biol. 2016;17:63. 865 \n71. Hwang B, Lee JH, Bang D. Single-cell RNA sequencing technologies and bioinformatics 866 \npipelines. Exp Mol Med. 2018;50:1–14. 867 \n72. Amezquita RA, Lun ATL, Becht E, Carey VJ, Carpp LN, Geistlinger L, et al. 868 \nOrchestrating single-cell analysis with Bioconductor. Nat Methods. 2020;17:137–45. 869 \n73. Yang Z, Ding K, Pan L, Deng M, Gan L. Math5 determines the competence state of 870 \nretinal ganglion cell progenitors. Dev Biol. 2003;264:240–54. 871 \n74. Sapkota D, Chintala H, Wu F, Fliesler SJ, Hu Z, Mu X. Onecut1 and Onecut2 872 \nredundantly regulate early retinal cell fates during development. Proc Natl Acad Sci U S A. 873 \n2014;111:E4086–95. 874 \n75. Agarwal D, Dash N, Mazo KW, Chopra M, Avila MP, Patel A, et al. Human retinal 875 \nganglion cell neurons generated by synchronous BMP inhibition and transcription factor 876 \nmediated reprogramming. NPJ Regen Med. 2023;8:55. 877 \n76. Fuhrmann S. Eye morphogenesis and patterning of the optic vesicle. Curr Top Dev Biol. 878 \n2010;93:61–84. 879 \n77. Krumlauf R. Hox genes in vertebrate development. Cell. 1994;78:191–201. 880 \n78. Swamy V, McGaughey D. Eye in a Disk: eyeIntegration Human Pan-Eye and Body 881 \nTranscriptome Database Version 1.0. Invest Ophthalmol Vis Sci. 2019;60:3236–46. 882 \n79. Wahlin KJ, Cheng J, Jurlina SL, Jones MK, Dash NR, Ogata A, et al. CRISPR Generated 883 \nSIX6 and POU4F2 Reporters Allow Identification of Brain and Optic Transcriptional 884 \nDifferences in Human PSC-Derived Organoids. Front Cell Dev Biol. 2021;9:764725. 885 \n80. Brown G. Retinoic acid receptor regulation of decision-making for cell differentiation. 886 \nFront Cell Dev Biol. 2023;11:1182204. 887 \n81. Yelon D, Stainier DYR. Pattern formation: swimming in retinoic acid. Curr Biol. 888 \n2002;12:R707–9. 889 \n82. Sanjurjo-Soriano C, Erkilic N, Damodar K, Boukhaddaoui H, Diakatou M, Garita-890 \nHernandez M, et al. Retinoic acid delays initial photoreceptor differentiation and results in a 891 \nhighly structured mature retinal organoid. Stem Cell Res Ther. 2022;13:478. 892 \n83. Teotia P, Niu M, Ahmad I. Mapping developmental trajectories and subtype diversity of 893 \nnormal and glaucomatous human retinal ganglion cells by single-cell transcriptome analysis. 894 \nStem Cells. 2020;38:1279–91. 895 \n84. Matsushita T, Onishi A, Matsuyama T, Masuda T, Ogino Y, Kageyama M, et al. Rapid 896 \nand efficient generation of mature retinal organoids derived from human pluripotent stem 897 \ncells via optimized pharmacological modulation of Sonic hedgehog, activin A, and retinoic 898 \nacid signal transduction. PLoS One. 2024;19:e0308743. 899 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.08.704697doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}