Unsupervised clustering for single-cell transcriptomic analysis to identify retinal cell classes. (A) Sketch of retinal sections showing the main cell types. (B–D) tSNE cluster results of the expression profile of 86,962 developing human whole-retinal cells (B), 9,070 developing human macula cells (C), and 11,005 developing human periphery cells (D). (E) Heatmap of DEGs from each cell type and the GO term enrichment of each set of DEGs. For visualization, the top five GO terms with the lowest p-value were used. Each column and row represent a single cell and gene, respectively. (F) Average expression of known marker genes in each cell class. (G) Statistics on the proportion of major cell types in the adult human whole retina, adult mouse whole retina, and adult zebrafish whole retina. (H) Transcriptional pattern correlation of major retinal types in humans and mice. The top 3,000 highly variable genes were extracted for correlation analysis, with the correlation coefficient >0.6 reserved for presentation, and the line width indicated correlation.

Expression heterogeneity of IRD gene in the retinal cells of three species. IRD genes highly expressed in (A) photoreceptors (human), (B) photoreceptors (mouse), (C) photoreceptors (zebrafish) (D) MGs (human), (E) RPCs (mouse), (F) MGs, astrocyte cells, and microglial cells (zebrafish).

WGCNA reveals the gene network module and hub genes of different retinal cell types. (A) Clustering dendrogram of 283 samples. Samples were clustered according to the similarity of gene expression. (B) Heatmap visualization gene network based on the topological overlap matrix (TOM). Color depth represents the degree of overlap. (C) Module–module and module–cell type correlation statistical analysis, in which the color of the square represents the correlation of modules, the solid line and the dotted line represent positive and negative correlation, respectively, and the color and the thickness of the line represent the significance and correlation of module–cell type, respectively. (D) GO term enrichment by eigengenes of different modules. (E–G) Network and hub genes for brown (E), black (F), and yellow (G) modules. Hub genes were identified from the module genes using a degree analysis method. The depth of the color indicates the rank of the hub genes from low to high.

Identification of the TF regulation network of human retinal cells based on SCENIC. (A) Heatmap of the area under curve (AUC) score estimated by SCENIC. Colors distinguished the cell clusters of the retina. (B) Heatmap of the ON/OFF status of regulons. Brown/white indicates ON/OFF status. (C) Dot plot of the regulon specificity score (RSS) for the retinal core cell cluster. (D) tSNE plot of the AUC score for the major regulons of the retina.

Ligand–receptor-based interactions between human retinal cells. (A–C) Intercellular communication ability among retinal cells of humans (A), mice (B), and zebrafish (C). Line colors represent ligands expressed by retinal cells marked by the same color. Lines connect to the cells expressing the corresponding receptors. Line thickness is proportional to the number of ligands. Loops indicate autocrine circuits. (D) Heatmap shows the gene expression levels of receptor–ligand pairs involved in interactions between different clusters in the human retina. (E–G) Overview of selected ligand–receptor interactions of MG cells (E), RGC (F), and RPCs (G). Differential transcription pattern of macular and peripheral retinal cells.

Differential expression analysis of cell clusters by region. (A–C) Point plot of cell cluster DE analysis by region. The average log2FC and percent difference of each gene were compared between the two regions. (D–E) The violin diagram shows the selected DE genes of the macula and periphery. The red-labeled gene is the IRD gene. (D) Gene expressed in rods and cones; (E) genes expressed in the Müller glia. (F–H) The GSEA plot shows differential pathways in both regions for all cells (F), cones (G), and Müller glial cells (H).

Bifurcation in the transcriptional state of photoreceptors: (A) trajectory manifold of photoreceptors from the developing human retina. Cell trajectories/fates were defined by expression profiles. (B) Expression heatmap of significant (q < 1e-5) genes based on branch expression analysis comparing the two photoreceptors. GO terms are listed on the right. (C) Trace plots showing transcriptional changes in IRD gene expression levels along the pseudotime in the human retina. (D) Trajectory manifold of BCs from developing human retinas. (E) Feature plot on UMAP of select genes in BCs during development.