Overview of the study.

(A) Schematic summary of the study. scRNA-Seq and scATAC-Seq of whole 13LGS retinas were performed at 10 different time points. Major retinal cell types are identified by clustering in both scRNA-Seq and scATAC-Seq datasets (B) UMAP projections of all 13LGS retinal cells from scRNA-Seq and scATAC-Seq. Each point represents a single cell, with cell type and time point indicated by shading. (C) Cell-specific marker gene expression and accessibility are shown across 11 major retinal cell types. (D) The relative abundances of rods, rod precursors, cones and cone precursors are different between developing 13LGS and mouse retina.

Identification of 13LGS–specific regulatory mechanisms in cone development by integrating late-stage retinal scRNA-Seq and scATAC-Seq data with mouse.

(A) UMAP projections of the integrated scRNA-Seq data from 13LGS and mouse. (B) UMAP projections of the scATAC-Seq data, showing cell types re-annotated based on the integrated scRNA-Seq analysis. (C) Heatmaps displaying eight clusters of genes that are differentially expressed between 13LGS and mouse. Column labels correspond to the five major cell types relevant to photoreceptor development: RPCs (late-stage primary retinal progenitor cells), N. RPCs (late-stage neurogenic progenitor cells), BC/photoreceptor precursors (bipolar cell, cone, and rod precursor cells), cone photoreceptors and rod photoreceptors. (D) Heatmaps showing the chromatin accessibility of positively-correlated regulatory elements for each of the eight DEG clusters. (E) Dot plots depicting the relative number of positively correlated regulatory elements for each conserved gene pair between 13LGS and mouse. Each dot represents a conserved gene pair, with the x-axis indicating the number of regulatory elements in mouse and the y-axis showing the number in 13LGS. Pairwise t-tests were used to compare the number of elements in each species for each gene pair, and the median values are indicated below each plot. (F) Differential motif analysis in five cone development–associated cell types between 13LGS and mouse. (G) Zic3 expression motif accessibility is higher in 13LGS Photoreceptor Precursors and mature Cones than in mouse. (H) Heatmap showing the expression levels of the top 30 Cone-promoting genes identified by GRN analysis in 13LGS across retinal progenitor cells (RPCs), neurogenic progenitor cells (N. RPCs), and photoreceptor precursors in both 13LGS and mouse.

Zic3 is necessary for normal mouse cone development and sufficient to promote cone gene expression.

(A) Diagram of overexpression strategy used to test effects of ZIC3 overexpression. (B) Heatmap of log-2 fold change in expression between cone-like precursors and Rods separated by condition for select genes, scaled by gene. (C) Upset plot of significant differentially expressed genes shared between conditions. (D) Diagram of retinal progenitor cell-specific conditional loss of function analysis of Zic3. (E) Immunohistochemistry showing GFP and Arr3 expression in P14 wild type (WT) and Zic3lox/lox (Zic3 cKO) mouse retinas. Scale bars, 20 µm. (F) Box plot of the number of cones per micron along the retinal slice (n = 11 WT, 10 Zic3 cKO). P-values calculated by Wilcoxon rank-sum test. P0, postnatal day 0; P5, postnatal day 5; P14, postnatal day 14; FACS, fluorescence-activated cell sorting; scRNA-Seq, single-cell RNA sequencing; ONL, outer nuclear layer; INL, inner nuclear layer; DAPI, 4′,6-diamidino-2-phenylindole; GFP, green fluorescent protein; OE, overexpression; log2FC, log2 fold change; DEG, differentially expressed gene; WT, wild type; cKO, conditional knockout.

Mef2c is sufficient to promote cone-specific gene expression and repress rod-specific gene expression in mouse.

(A) Diagram of overexpression strategy used to test effects of MEF2C overexpression. (B) UMAP representation of P8 mouse retinal explants electroporated with a plasmid expressing GFP alone (Empty) or GFP in a bicistronic transcript with human MEF2C (MEF2C), (n = 7445 cell Empty, 11949 cells MEF2C). Each point represents a single cell and is colored by cell type as determined by clustering and marker gene expression. (C) Heatmap of expression for select genes for cones, cone-like photoreceptor precursors, rod photoreceptor precursors, and rods in cells overexpressing MEF2C, scaled by gene. (D) Immunohistochemistry showing GFP and GNAT2 expression in P8 mouse retinas from Empty and MEF2C conditions. Scale bars, 50 µm. (E) Box plot of the number of Gnat2+, GFP+ cells divided by the total number of GFP+ cells (n = 8 for both conditions). P-values calculated by Wilcoxon rank-sum test. P0, postnatal day 0; P8, postnatal day 8; FACS, fluorescence-activated cell sorting; scRNA-Seq, single-cell RNA sequencing; ONL, outer nuclear layer; INL, inner nuclear layer; DAPI, 4′,6-diamidino-2-phenylindole; GFP, green fluorescent protein; OE, overexpression.

Conserved TFs bind to species-specific enhancers to promote Cone specification in 13LGS.

(A) Schematic illustrating annotation of cis-regulatory elements in RPCs and Photoreceptor Precursors by integration of scATAC-Seq and CUT&RUN 13LGS and mouse datasets. (B) Heatmaps show annotated accessible regulatory elements in both 13LGS and mouse. Promoters, activated enhancers (AEs), and poised enhancers (PEs) which are associated with histone markers associated with genes in clusters C2 and C3, which are selectively active in 13LGS RPCs and/or photoreceptor precursors. Shading indicates CUT&TAG signal for the corresponding histone modification within 2kb of the scATAC-Seq peak center. Bar plots displaying the number of each category of regulatory element in each species that are conserved or species-specific. (C) Dot plots showing the enrichment of binding sites for Otx2 and Neurod1, TFs which are broadly expressed in both neurogenic RPC and photoreceptor precursors, which are enriched in both evolutionarily-conserved cis-regulatory elements in both species. (D) Bar plots showing the number of evolutionarily-conversed and species-specific enhancers per TSS in four cone-promoting genes between 13LGS and mouse. (E) The gene regulatory networks regulating Thrb expression in 13LGS and mouse late N. RPCs. (F) An example of a Thrb-related regulon and its corresponding scATACseq and CUT&RUN tracks. The arrow indicates the consistent regulatory relationships between GRNs prediction and experimental validations. (G) The epigenetic model of cone specification in 13LGS and mouse.