MG and MG-derived cells maintained high level of Notch signaling.

(a) Schematic representations of the intravitreal injection of AAV7m8-GFAP-cyclinD1-p27Kip1 shRNA-WPRE (CCA). (b) The overview process of MG proliferation induced by CCA. (c) Representative Sox9 and EdU immunostaining on retinal sections from Glast-CreERT2;Sun1:GFP mice treated with CCA at P28 and collected 4 months post-treatment. C1-C2 are the magnified views of the highlighted regions. The white arrows refer to GFP+ EdU+ Sox9- cells. (d) Representative Otx2 immunostaining on retinal sections from Glast-CreERT2;Sun1:GFP mice treated with CCA at P28 and collected 4 months post-treatment. D1-D2 are the magnified views of the highlighted regions. The white arrows refer to GFP+ Otx2+ cells. (e) Hes1 mRNA in situ hybridization in the control and the Glast-CreERT2;Sun1:GFP mouse retinas harvested at four months post CCA injection. (f) Magnified views of the highlighted regions in (e). n=3 mice.

Rbpj deletion in adult MG induces limited dedifferentiation

(a) Schematic illustration of MG dedifferentiation and reprogramming experiment. (b) Representative immunostaining of Sox9 on retinal sections from Glast-CreERT2;Sun1:GFP and Glast-CreERT2;Rbpjflox/flox;Sun1:GFP mice in different timepoints post TAM injection. The white arrows refer to GFP+ Sox9- cells. (c) Magnified views of the highlighted regions in (b). (d) Percentage of GFP+ Sox9- cells in overall GFP+ cells. n=3 mice, data are presented as mean ± SEM. ns=not significant, ****P < 0.0001, by one-way ANOVA with Tukey’s post hoc test. (e) Percentage of GFP+ Otx2+ cells in overall GFP+ cells. n=3 mice, data are presented as mean ± SEM. ns=not significant, **P < 0.01, by one-way ANOVA with Tukey’s post hoc test.

Rbpj KO and CCA synergistically increase Otx2+ cell formation from MG

(a) Schematic illustration of the neurogenesis assessment experiment. (b) Representative immunostaining of EdU and Otx2 on retinal sections. The white arrows refer to tdT+ Otx2+ cells. (c) Magnified views of the highlighted regions in (b). The white arrows refer to tdT+ Otx2+ cells. (d) Percentage of tdT+ Otx2+ cells in overall tdT+ cells. 3W: 3 weeks, 4M: 4 months, n=3 mice, data are presented as mean ± SEM. ns=not significant, ***P < 0.001, by one-way ANOVA with Tukey’s post hoc test. (e) Percentage of tdT+ Otx2+ cells in overall tdT+ cells. n=3 mice, data are presented as mean ± SEM. ns=not significant, *P < 0.05, ***P < 0.001, ****P < 0.0001, by one-way ANOVA with Tukey’s post hoc test. (f) Representative immunostaining of EdU and Otx2 or Crx on retinal sections. (g) Percentage of tdT+ EdU+ Otx2+ cells in ONL tdT+ EdU+ cells. n=3 mice, data are presented as mean ± SEM. ns=not significant, ****P < 0.0001, by one-way ANOVA with Tukey’s post hoc test. (h) Percentage of tdT+ EdU+ Crx+ cells in ONL tdT+ EdU+ cells. n=3 mice, data are presented as mean ± SEM. ns=not significant, ****P < 0.0001, by one-way ANOVA with Tukey’s post hoc test.

snRNA-seq analysis shows that Rbpj KO and CCA promote MG reprogramming.

(a) Schematic illustration of the snRNA-seq experiment. To induce Rbpj deletion and GFP expression in MG, we administered TAM for 7 consecutive days in Glast-CreERT2;Sun1:GFP and Glast-CreERT2;Rbpjflox/flox;Sun1:GFP mice from P28. At P36, CCA was injected into the CCA and Rbpj KO+CCA groups, while control and Rbpj KO mice were uninjected by any AAV vector. 3-4 retinas from mice at indicated ages were pooled together for nuclei extraction, and MG nuclei were then isolated by GFP signal via FACS for snRNA-seq. (b) UMAP plot of snRNA-seq data from four treatment groups (Ctrl, CCA, Rbpj KO, Rbpj KO+CCA) at different timepoints, and separated UMAP based on different timepoints. clusters were identified based on known marker gene expression. (c) Proportions of cell clusters within Ctrl, CCA, Rbpj KO and + groups at different timepoints. (d) Feature plots highlighting the cluster of quiescent MG (Kcnj10), proliferating MG (Mki67), reactivated MG (Gfap), MGPC (Dll1, Ascl1), transitional MG (Mt-Nd4), AC-like MG (Elavl3), and BC-like MG (Otx2). (e) Heatmap showing the expression of top 50 differentially expressed genes (DEGs) of 3-week reactivated MG between CCA and Rbpj KO+CCA (p<0.05). (f) Heatmap showing the expression of top 50 DEGs of 4-month MGPC between Rbpj KO and Rbpj KO+CCA (p<0.05).

CCA and Rbpj deletion drive BC subtype differentiation.

(a) Subclusters of the BC-like population. The BC-like MG (outlined in red) was used for subclustering analysis. (b) Feature plot of BC-like subtypes showing the cluster of OFF-cone BC (Pcdh17, Zfhx4), ON-cone BC (Isl1, Grm6) and rod BC (Isl1, Prkca, Cep112). (c) Dot plot showing gene expression and cell percentages for OFF-cone BC, ON-cone BC and rod BC. (d) Percentage of BC subtypes in different treatment groups. (e-f) Pcdh17 and Grm6 mRNA in situ hybridization in the Ctrl and Rbpj KO+OCCA groups. The white dashed boxes indicate the position of the enlarged images. (g-h) Representative immunostaining of EdU and PKCα on retinal sections of the Ctrl and Rbpj KO+CCA samples. The white dashed boxes indicate the position of the enlarged images. (i) Percentage showing GFP+ Grm6+ cells in overall GFP+ cells, GFP+ Pcdh17+ cells in overall GFP+ cells and GFP+ PKCα+ cells in overall GFP+ cells in the Ctrl and Rbpj KO+CCA groups. n=3 mice, *P < 0.05, **P < 0.01, ****P < 0.0001, by unpaired two-tailed student’s t-test. (j) Representative immunostaining of MG-derived cells with bipolar cell morphology.

snATAC-seq analysis suggests increased chromatin accessibility of the neurogenic genes by CCA.

(a) Schematic illustration of the snATAC-seq experiment. (b) Proportions of cell clusters within Ctrl, CCA, Rbpj KO and Rbpj KO+CCA groups. (c) UMAP plot of snATAC-seq data from four treatment groups (Ctrl, CCA, Rbpj KO, Rbpj KO+CCA). (d) Feature plots highlighting the cluster of quiescent MG (Kcnj10), KO MG (Hes5), active MG (Bal3), MGPC (Dll1), AC-like MG (Caln1), BC-like MG (Otx2). (e) Dot plot showing gene expression and cell percentages for quiescent MG, KO MG, active MG, MGPC, AC-like MG, BC-like MG. (f) Volcano plot showing the differential gene expression of MG and active MG within CCA treatment group (p<0.05). Red dots indicate the genes upregulated in active MG and blue dots indicate genes downregulated in active MG. (g) WikiPathways enrichment analysis of upregulated genes in active MG (p<0.05). (h) Increased chromatin accessibility at the Neurod2 locus was observed in the active MG. The black arrow shows the transcription direction.

Neurons derived from MG exhibited long-term survival capabilities.

(a) Schematic illustration of the experiment assessing long-term cell survival. (b) Representative immunostaining of EdU and Otx2 on retinal sections. The white arrows refer to tdT+ Otx2+ cells. (c) Percentage of tdT+ Otx2+ cells in overall tdT+ cells, n=3 mice, data are presented as mean ± SEM. ns=not significant, **P < 0.01, ***P < 0.001, by one-way ANOVA with Tukey’s post hoc test. (d) Percentage of tdT+ EdU+ Otx2+ cells in overall tdT+ EdU+ cells, ns=not significant, **P < 0.01, by one-way ANOVA with Tukey’s post hoc test. (e) Number of EdU+ cells per 500 µm, ns=not significant, by one-way ANOVA with Tukey’s post hoc test.

Notch signaling remains active in MG and MG-derived progeny.

(a) Schematic illustrating the experimental design for labeling proliferating MG. (b) Schematic illustrating the changes in MG proliferation over time. (c) Representative EdU immunostaining on retinal sections from Glast-CreERT2;Sun1:GFP mice treated with CCA at P28 and collected 4 months post-treatment. (d) Schematic illustration of scRNA-seq experiment. (e) UMAP plot of scRNA-seq data for MG treated with CCA and control virus. (f) Split UMAP plots of the control and CCA groups. (g) Feature plots of normalized Notch related genes, Notch1, Rbpj and Hes1 gene expression in different cell clusters.

Deletion of Rbpj was sufficient to inhibit the canonical Notch signaling pathway.

(a) Schematic of Glast-CreERT2;Rbpjflox/flox;tdT mouse used in this study. (b) Schematic illustration of the examination of Notch inhibition via Rbpj deletion. (c) Hes1 mRNA in situ hybridization in Glast-CreERT2;tdT (Ctrl), Glast-CreERT2;Rbpjflox/wt;tdT and Glast-CreERT2;Rbpjflox/flox;tdT mice received tamoxifen (TAM) injection. (d) Magnified views of the highlighted regions in (c). (e) The average pixel level of Hes1 mRNA per GFP+ cell. n≥3 mice, data are presented as mean ± SEM. ns=not significant, ****P < 0.0001, by one-way ANOVA with Tukey’s post hoc test.

Overproduction of rod photoreceptor cells at expense of MG after Notch signaling inhibition.

(a) Schematic of Glast-CreERT2;Rbpjflox/flox;Sun1:GFP mouse used in this study. (b) Schematic illustration of the experiment examining how Rbpj deletion affects neurogenesis in retinal progenitor cells. Glast-CreERT2;tdT (Ctrl), Glast-CreERT2;Rbpjflox/wt;tdT and Glast-CreERT2;Rbpjflox/flox;tdT mice were received TAM injection at postnatal day 1 (P1) and harvested at P12. (c) Representative immunostaining of Otx2 and Sox9 on retinal. (d) Percentage of tdT+ rod photoreceptor cells in overall tdT+ cells. (e) Percentage of tdT+ Sox9+ cells in overall tdT+ cells. n≥3 mice, ns. not significant, ****P < 0.0001, by one-way ANOVA with Tukey post hoc test.

Deletion of Rbpj in late RPCs would not affect Cone photoreceptor cells formation.

(a) Representative immunostaining of mCAR and Sox2 on retinal sections from Glast-CreERT2;tdT (Ctrl), Glast-CreERT2;Rbpjflox/wt;tdT and Glast-CreERT2;Rbpjflox/flox;tdT mice received tamoxifen (TAM) injection at postnatal day 1 (P1) and harvested at P12. (b) Magnified views of the highlighted regions in (a). (c) Number of cone photoreceptor cells in 250μm. n≥3 mice, data are presented as mean ± SEM. ns=not significant, by one-way ANOVA with Tukey’s post hoc test.

Notch signaling inhibition hindered the BCs generation from late RPCs.

(a) Representative immunostaining of Otx2 on retinal sections from Glast-CreERT2;tdT (Ctrl), Glast-CreERT2;Rbpjflox/wt;tdT and Glast-CreERT2;Rbpjflox/flox;tdT mice received tamoxifen (TAM) injection at postnatal day 1 (P1) and harvested at P12. (b) Magnified views of the highlighted regions in (a). (c) Percentage of tdT+ Otx2+ cells in overall tdT+ cells. n≥3 mice, ns=not significant, *P < 0.05, **P < 0.01, by one-way ANOVA with Tukey post hoc test.

Deletion of Rbpj in late RPCs would not affect ACs formation.

(a) Representative immunostaining of HuC/D and Pax6 on retinal sections from Glast-CreERT2;tdT (Ctrl), Glast-CreERT2;Rbpjflox/wt;tdT and Glast-CreERT2;Rbpjflox/flox;tdT mice received TAM injection at postnatal day 1 (P1) and harvested at P12. (b) Magnified views of the highlighted regions in (a). (c) Percentage of tdT+ HuC/D+ Pax6+ cells in overall tdT+ cells. n≥3 mice, data are presented as mean ± SEM. ns=not significant, by one-way ANOVA with Tukey’s post hoc test.

Deletion of Rbpj in late RPCs would not affect RGCs formation.

(a) Representative immunostaining of Rbpms on retinal sections from Glast-CreERT2;tdT (Ctrl), Glast-CreERT2;Rbpjflox/wt;tdT and Glast-CreERT2;Rbpjflox/flox;tdT mice received tamoxifen (TAM) injection at P1 and harvested at P12. (b) Magnified views of the highlighted regions in (a). (c) Number of Rbpms+ cells per 150μm. n≥3 mice, data are presented as mean ± SEM. ns=not significant, by one-way ANOVA with Tukey’s post hoc test.

Rbpj deletion in adult MG induces limited neuronal conversion.

(a) Schematic illustration of MG dedifferentiation and reprogramming experiment. (b) Representative immunostaining of Sox9 on retinal sections from Glast-CreERT2;Sun1:GFP and Glast-CreERT2;Rbpjflox/flox;Sun1:GFP mice harvested at different timepoints post TAM injection. The white arrows refer to GFP+ Sox9- cells. (c) Magnified views of the highlighted regions in (B).

Rbpj deletion induces transdifferentiation-mediated neurogenesis in adult MG.

(a) Schematic illustration of the experiment examining MG proliferation. (b) Representative immunostaining of Otx2 and EdU on retinal sections from Glast-CreERT2;Sun1:GFP and Glast-CreERT2;Rbpjflox/flox;Sun1:GFP mice harvested at 4 months post TAM injection.

Notch inhibition reduced MG proliferation induced by CCA, yet it did not entirely prevent it.

(a) Schematic illustration of proliferation level comparison experiment. (b) Representative immunostaining of EdU on retinal sections from Glast-CreERT2;tdT (Ctrl), Glast-CreERT2;Rbpjflox/wt;tdT and Glast-CreERT2;Rbpjflox/flox;tdT mice received CCA injection. (c) Number of EdU+ cells in 500μm. n≥3 mice, data are presented as mean ± SEM. ns=not significant, ***P < 0.001; ****P < 0.0001, by one-way ANOVA with Tukey’s post hoc test. (d-e) Schematic illustration of the comparison of proliferation levels between different CCA injection and TAM administration orders. (f) Representative immunostaining of EdU on retinal sections from Glast-CreERT2;Rbpjflox/flox;tdT mice received CCA injection. (g) Number of EdU+ cells in 500μm. n=3 mice, data are presented as mean ± SEM. ns=not significant, ****P < 0.0001, by one-way ANOVA with Tukey’s post hoc test.

The synergistic effect of Rbpj KO and CCA resulted in a robust MG dedifferentiation.

(a) Schematic illustration of MG dedifferentiation experiment. (b) Representative immunostaining of Sox9 on retinal sections from Glast-CreERT2;Sun1:GFP (Ctrl) and Glast-CreERT2;Rbpjflox/flox;Sun1:GFP mice at 3 weeks and 4 months post CCA treatment. The white arrows refer to GFP+ Sox9- cells. (c) Magnified views of the highlighted regions in (b). (d) Percentage of GFP+ Sox9- cells in overall GFP+ cells. 3W: 3 weeks, 4M: 4 months, n=3 mice, data are presented as mean ± SEM. ns=not significant, ****P < 0.0001, by one-way ANOVA with Tukey’s post hoc test. (e) Percentage of GFP⁺ Sox9⁻ cells among total GFP⁺ cells in the 4-month samples. n=3 mice, data are presented as mean ± SEM. ns=not significant, *P < 0.05, ****P < 0.0001, by one-way ANOVA with Tukey’s post hoc test. (f) Number of Sox9+ cells per 500μm in the 4-month samples. n=3 mice, data are presented as mean ± SEM. ns=not significant, *P < 0.05, **P < 0.01, ***P < 0.001, by one-way ANOVA with Tukey’s post hoc test.

The synergistic effect of Rbpj KO and CCA resulted in a robust MG reprogramming.

(a) Percentage of tdT+ EdU+ Otx2+ cells in overall tdT+ EdU+ cells. n=3 mice, data are presented as mean ± SEM. ns=not significant, ****P < 0.0001, by one-way ANOVA with Tukey’s post hoc test. (b) Representative immunostaining of EdU and Nrl on retinal sections. (c) Percentage of tdT+ EdU+ Nrl+ cells in ONL tdT+ EdU+ cells. n=3 mice, data are presented as mean ± SEM. ns=not significant, ***P < 0.001, by one-way ANOVA with Tukey’s post hoc test.

Preprocessing and filtering of snRNA data.

(a) The UMAP before the removal of contamination cells. The original mature neurons affecting the analysis were removal. (b) Number of cells passing quality control and used for snRNA-seq analysis. (c) Dot plot showing the expression of marker genes of each cluster in unfiltered UMAP. (d) The initial annotation of clusters in the unfiltered UMAP.

Additional snRNA-seq analysis of MG at 1 week, 3 weeks and 4 months post-CCA treatment.

(a) Dot plot showing gene expression and cell percentages for quiescent MG, proliferating MG, reactivated MG, MGPC, transitional MG, AC-like MG, BC-like MG. (b) Separated UMAP plot of snRNA-seq data of Ctrl, CCA, Rbpj KO, Rbpj KO+CCA treatment. (c) Separated proportions of cell clusters within Ctrl, CCA, Rbpj KO and Rbpj KO+CCA groups at different timepoints.

Separation of UMAP by different timepoints and treatments.

(a) The split UMAP by the condition of time and treatment. To monitor the progress of MG regeneration, we implemented three time points (1 week (1W), 3 weeks (3W), 4 months (4M)) and four treatment groups (Ctrl, CCA, Rbpj KO, Rbpj KO+CCA).

Pseudotime trajectory analysis of MG reprogramming process.

(a) Pseudotime trajectory analysis showing the MG reprogramming process. (b) Dynamic cell type distribution along the pseudotime. (c) Change of gene expression level along the pseudotime. MG (Hes5, Kcnj10, Slc1a2, Sox9, Hes1, Aqp1, Rlbp1); Reactivated MG (Mx2, Vim); Proliferating MG (Mcm5, Mki67); MGPC (Neurog2, Dll1, Ascl1, Eya2); AC-like (Elavl4, Caln1, Rbfox3); photoreceptor cell (Prom1); RGC (Tubb3); BC-like (Otx2, Car10, Cabp5). (d) Gene expression trends along pseudotime. (e) Separated gene expression trends along pseudotime.

Subclustering analysis of the AC-like population.

(a) Feature plots of AC-like subtypes showing the expression patterns of Chat (starburst ACs), Dab1 (A17 ACs), and Slc6a9 (nGnG ACs). The AC-like population is outlined in red. (b) Representative immunostaining of EdU and HuC/D on retinal sections. The white arrows refer to tdT+ HuC/D+ cells. (c) Magnified views of the highlighted regions in (b). (d) Percentage of tdT+ EdU+ HuC/D+ cells in overall tdT+ EdU+ cells. n=3 mice, data are presented as mean ± SEM. ns=not significant, ***P < 0.001, by one-way ANOVA with Tukey’s post hoc test. (e) Percentage of tdT+ HuC/D+ cells in overall tdT+ cells. n=3 mice, data are presented as mean ± SEM. ns=not significant, *P < 0.05, by one-way ANOVA with Tukey’s post hoc test.

The newborn neurons showed diverse gene expression profiles linked to different neuronal cell types.

(a) Feature plots showing the additional neuronal markers related to RGC and photoreceptor cells. (b) Dot plots showing different expression of RGC-specific and photoreceptor cells-specific genes (p<0.05).

Preprocessing and filtering of snATAC data.

(a) The UMAP before the removal of contamination cells. (b) Dot plot showing the expression of marker genes of each cluster in unfiltered UMAP. (c) The initial annotation of clusters in the unfiltered UMAP.

Expression pattern of ZO1 in the retina.

(a) Schematic illustration of ZO1 staining experiment. (b) Representative immunostaining of Otx2 and ZO1 on retinal sections from Glast-CreERT2;tdT and Glast-CreERT2;Rbpjflox/flox;tdT mice at 9 months post TAM injection. The white arrows refer to tdT+ Otx2+ cells.