Revised

[version 6; peer review: 1 approved, 3 approved with reservations]

Yoshitoku Watabe

https://orcid.org/0009-0007-4611-8193

https://orcid.org/0009-0001-1242-2097

https://orcid.org/0000-0002-4023-4560

https://orcid.org/0000-0002-8540-7760

https://orcid.org/0000-0002-0212-0909

Yoshitoku Watabe

https://orcid.org/0009-0007-4611-8193

https://orcid.org/0009-0001-1242-2097

https://orcid.org/0000-0002-4023-4560

https://orcid.org/0000-0002-8540-7760

https://orcid.org/0000-0002-0212-0909

PUBLISHED 10 Oct 2026

Author details Author details

OPEN PEER REVIEW

REVIEWER STATUS

Signal transduction via the Notch signaling pathway depends on the expression of receptors regulated at multiple levels. In addition to intracellular machineries such as glycosylation,¹ ubiquitination,² and phosphorylation,³ these regulations include chromatin remodeling and DNA-binding proteins, such as transcription factors and transcriptional repressors, which could regulate the levels of Notch ligands and receptors. Since fine-tuned signal transduction is necessary for development, it is important to clarify the regulatory mechanisms of receptor expression for understanding the regulation of cell differentiation and subsequent tissue development.

The retina is a good model for investigating cellular differentiation since cell fate determination is well documented. The developed retina is composed of multiple cell types, including retinal ganglion cells, amacrine cells, bipolar cells, horizontal cells, rods, cones, and glial cells (Muller glia), all of which originate from retinal progenitor cells (retinal progenitors). In addition, astrocytes and microglia exist during retinal development. Recently, a detailed molecular characterization at single-cell resolution of the developing retina was reported.⁴ In addition, retinal progenitors are also known to express several transcription factors such as Lhx2,⁵ Pax6, ⁶ Rax,⁷ and Vsx2.⁸

Notch signaling involves four Notch receptors that are expressed differently in space. In rat embryonic eyes, Notch1 was expressed on the posterior side of the retina at embryonic days (E) 13.5 and E14.5; Notch2 was expressed in the retinal pigment epithelium throughout E12.5 and E14.5; and Notch3 was expressed throughout the retina from E12.5 to E14.5, with strong expression on the anterior side at E14.5.^9,10 Further analyses on Notch2 using a knock-in mouse suggested that Notch2 was expressed at a slightly detectable level from the E9.5 optic vesicle stage in the retinal neuroepithelium that contained retinal progenitors.¹¹ In addition, deletion of Notch2 in outer ciliary epithelium disrupted ciliary body formation,¹² suggesting that this gene acts in the temporal side of the eye. Indeed, the spatial information in the mouse retina at E13.5 determined upstream regulators involved in Hes1 induction, which acts as a hub for Notch-dependent and -independent inputs in retinal progenitors.¹³ Notably, Notch4 was expressed in arterial endothelial cells rather than retinal progenitors and their descendants in the retina.¹⁴

Notch signaling is also involved in the complex differentiation of retinal progenitors into differentiated cells. Loss-of-function of Notch common co-actors Rbpj¹⁵ and Maml¹⁶ increased Crx-expressing photoreceptor precursor cells and cone/horizontal cell-restricted retinal progenitors, respectively. More specifically, Notch1 has been shown to inhibit the rod fate and maintain pluripotency of retinal progenitors.¹⁷⁻¹⁹ This notion is also supported by a single-cell ATAC-seq and disease-associated 5q14.3 enhancer knockout study.²⁰ In addition, Notch3, as well as Notch1, was shown to be expressed as early as the optic vesicle stage.²¹ Indeed, Notch3 knockout mice showed a reduced number of retinal ganglion cells,²² although zebrafish Notch3 was also essential for Muller glial cells.²³ These results suggest that Notch1 and Notch3 act in retinal progenitors to regulate their differentiation, although it is unclear how retinal progenitor-dominant expressions of Notch1 and Notch3 are regulated during retinal progenitor differentiation.

A murine ophthalmological study examined global epigenetic changes and interacting transcription factors during retinal cell differentiation.²⁴ By chromVAR analysis in the Signac pipeline and CUT&Tag analysis, this study revealed that the epigenetic landscape of cell type-specific enhancers shifted during retinal cell differentiation and identified several transcription factors (e.g., Otx2, Atoh7, Pou4f2, and Isl1) that potentially work in each differentiation stage.²⁴ ChromVAR analysis identified five Atoh7-binding motifs in the Notch1 locus that were specific to naïve retinal progenitors, although CUT&Tag analysis did not result in positive findings for any of these motifs. In contrast, this report also identified chromatin remodeling at retinal progenitor-specific enhancers for key transcription factors, including Vsx2, and showed binding of transcription factors and their binding sites for key transcription factors, including Rax, Lhx2, and Pax6, in naïve and transitional retinal progenitors, although specific genomic regions for these transcription factors are still lacking, as chromVAR was used. Another study also reported chromatin remodeling in the NOTCH1 locus in humans,²⁰ although a comprehensive analysis focusing on possible epigenetic regulation of the four Notch receptor gene expressions in humans, including the cis-regulatory elements (CREs), is still lacking, as this study focused on retinal cell differentiation instead of Notch signaling.

The Notch loci are subject to epigenetic regulation under both normal^25–30 and pathological^31–39 conditions in a context-dependent manner. However, few studies have used genome-wide data to investigate transcription factor expression and the associated transcription factor binding sites underlying Notch receptor gene expression during cell differentiation. Although a single-cell RNA/ATAC-seq dataset from developing retinas, in which differentiation trajectories were well characterized, is available, and the accessibility of the NOTCH1 locus is known to decrease over differentiation,²⁰ how this chromatin remodeling affects NOTCH1 expression is still unknown. To gain insight into chromatin remodeling and possible CREs responsible for regulation of the Notch receptor gene expression, we re-analyzed a publicly available multi-omics dataset of single-cell RNA-seq and single-cell ATAC-seq of human fetal retinas.²⁰

To perform fine-grained annotation of the single-cell RNA-seq and ATAC-seq data, we downloaded an h5ad file,⁴ which includes cell annotations from detailed molecular characteristics analysis of embryonic retinal cells from the CZ CELLxGENE Discover platform (https://cellxgene.cziscience.com/collections/c11009b8-b113-4a99-9890-78b2f9df9d79). The h5ad file was converted into an RDS file and then analyzed in the Seurat pipeline.⁴⁰

A single-cell RNA/ATAC-seq dataset (GSE183684)²⁰ was downloaded from the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/). Data from the three early time points (days 59, 74, and 78) in this dataset were used to balance computational burden and the number of retinal progenitors. The data were processed in the pipelines using Seurat and Signac.⁴¹ Detailed session information for each step is provided in the archived code (See Data availability section). Integration of the single-cell RNA-seq data and the single-cell ATAC-seq data was conducted using the “FindTransferAnchors” function in Seurat. We used the FIMO tool⁴² in the MEME Suite,⁴³ referencing the JASPAR motif database (https://jaspar.elixir.no/) and the human genome assembly hg38 in the University of California, Santa Cruz (UCSC) genome browser (https://genome.ucsc.edu/) to search transcription factor binding motifs. We further compared the identified RAX-binding site with the JASPAR transcription factor binding site track (JASPAR 2026) in the UCSC Genome Browser and with the candidate CREs registered in SCREEN v4 (accessed September 25, 2026).⁴⁴

We started by analyzing the single-cell RNA-seq data⁴ that we will use for annotation in the other single-cell RNA/ATAC-seq data.²⁰ Uniform manifold approximation and projection (UMAP) analysis showed complexly ramified cell clusters ( Figure 1A), and in concordance with the original report,⁴ we identified SOX2 and MKI67-double-positive Retinal progenitors ( Figure 1B-C). The cell number in each cluster is available in Additional Table 1. We performed dot plot analysis of the marker genes (SOX2 and MKI67 for Retinal progenitors; RLBP1 for Muller glia; PAX6 for Retinal ganglion cell/Amacrine cell/Horizontal cell fate; RCVRN, OTX2, PDE6H, NRL, and RHO for Rods and Cones; VSX2 for Bipolar cells; TFAP2A, TFAP2B for Amacrine cells; ONECUT2 for Horizontal cells; POU4F2 and RBPMS for Retinal ganglion cell) according to the original report⁴ to provide the average expression levels and proportion of these marker-expressing cells to guarantee the validity of annotation ( Figure 1D). Importantly, some of the markers were expressed in multiple cell clusters. We then screened the four Notch receptor genes and found that NOTCH1 was expressed in the Retinal progenitors (log2 fold change to background: 1.09, adjusted p-value: 0), Transitory (log2 fold change to background: 1.32, adjusted p-value: 0), and Muller glia (log2 fold change to background: 1.43, adjusted p-value: 0) cell clusters. The other Notch receptor genes were also expressed in the Retinal progenitors, although the expression was sparse, prompting us to focus on NOTCH1. We also examined the expression of RAX, PAX6, LHX2, and VSX2. RAX was enriched in the Retinal progenitors (log2 fold change to background: 0.62, adjusted p-value: 0), Transitory (log2 fold change to background: 0.87, adjusted p-value: 0), Muller glia (log2 fold change to background: 0.72, adjusted p-value: 1.66 x 10⁻¹⁴⁹), Cone precursors (log2 fold change to background: 0.59, adjusted p-value: 7.02 x 10⁻³⁵), Cones (log2 fold change to background: 1.31, adjusted p-value: 0), and Astrocytes (log2 fold change to background: 0.48, adjusted p-value: 5.36 x 10⁻³) cell clusters. PAX6 was also enriched in Retinal progenitors (log2 fold change to background: 0.15, adjusted p-value: 1.84 x 10⁻⁵⁴), Transitory (log2 fold change to background: 0.23, adjusted p-value: 2.34 x 10⁻¹⁶³), Amacrine cells (log2 fold change to background: 1.20, adjusted p-value: 0), Horizontal cells (log2 fold change to background: 1.13, adjusted p-value: 0), Retinal ganglion cells (log2 fold change to background: 0.35, adjusted p-value: 1.15 x 10⁻¹⁶⁰), and Muller glia (log2 fold change to background: 0.18, adjusted p-value: 6.83 x 10⁻³⁹) cell clusters. LHX2 was enriched in Retinal progenitors (log2 fold change to background: 0.95, adjusted p-value: 0), Transitory (log2 fold change to background: 1.43, adjusted p-value: 0), Muller glia (log2 fold change to background: 0.94, adjusted p-value: 1.59 x 10⁻¹⁶³), Astrocytes (log2 fold change to background: 1.15, adjusted p-value: 5.20 x 10⁻²⁰) cell clusters. VSX2 was enriched in Retinal progenitors (log2 fold change to background: 0.42, adjusted p-value: 2.12 x 10⁻¹⁷⁵), Transitory (log2 fold change to background: 1.07, adjusted p-value: 0), Bipolar cells (log2 fold change to background: 2.07, adjusted p-value: 0), Muller glia (log2 fold change to background: 0.60, adjusted p-value: 1.60 x 10⁻⁹¹) cell clusters. All the enriched gene information is available in Additional Table 2. For the following single-cell ATAC-seq analysis, we provide UMAP results highlighting Retinal progenitors ( Figure 1E) and Transitory ( Figure 1F) cell clusters, and feature plots for NOTCH1 ( Figure 1G) and RAX ( Figure 1H). We note that Retinal progenitors and Transitory cell clusters showed similar distributions in the UMAP, in which NOTCH1 and RAX were expressed.

Next, we examined single-cell RNA-seq data from human fetal retinas at days 59, 74, and 78 from GSE183684.²⁰ The cell numbers in each cluster are available in Additional Table 3. After using the “RunHarmony” function⁴⁵ with the Seurat pipeline to reduce batch effects, we annotated the cells according to the earlier single-cell RNA-seq data⁴ using the “FindTransferAnchor” function in Seurat. These data also contained all types of the aforementioned cell clusters ( Figure 2A). Indeed, we observed SOX2/MKI67-double-positive Retinal progenitors ( Figure 2B-C). We also provided a dot plot for characteristic gene expression for each cell cluster ( Figure 2D). As expected, screening of the Notch receptor genes suggested prominent NOTCH1 expression in the Retinal progenitors (log2 fold change to background: 1.26, adjusted p-value: 0), Transitory (log2 fold change to background: 1.03, adjusted p-value: 1.88 x 10⁻²³¹), and Muller glia (log2 fold change to background: 1.45, adjusted p-value: 1.50 x 10⁻¹⁰⁶) cell clusters. In addition, RAX was enriched in Retinal progenitors (log2 fold change to background: 0.64, adjusted p-value: 1.32 x 10⁻¹³¹), Transitory (log2 fold change to background: 0.81, adjusted p-value: 3.88 x 10⁻¹⁵³), Muller glia (log2 fold change to background: 0.38, adjusted p-value: 2.03 x 10⁻¹⁰), Cone precursors (log2 fold change to background: 0.95, adjusted p-value: 6.79 x 10⁻⁸²), Cones (log2 fold change to background: 1.43, adjusted p-value: 3.67 x 10⁻¹⁰⁷), and Rods (log2 fold change to background: 0.96, adjusted p-value: 8.94 x 10⁻¹⁶) cell clusters. PAX6 was enriched in Horizontal cells (log2 fold change to background: 0.84, adjusted p-value: 2.53 x 10⁻¹²²) and Retinal ganglion cells (log2 fold change to background: 0.21, adjusted p-value: 1.98 x 10⁻⁴³) cell clusters but not in Retinal progenitors or Transitory cell clusters, possibly owing to its wide expression across the cell clusters. LHX2 was enriched in Retinal progenitors (log2 fold change to background: 0.87, adjusted p-value: 4.90 x 10⁻¹⁶⁹), Transitory (log2 fold change to background: 1.03, adjusted p-value: 1.62 x 10⁻¹⁵³), Muller glia (log2 fold change to background: 0.21, adjusted p-value: 1.31 x 10⁻³), Amacrine cells (log2 fold change to background: 0.49, adjusted p-value: 8.75 x 10⁻²), and Astrocytes (log2 fold change to background: 1.29, adjusted p-value: 1.16 x 10⁻⁴) cell clusters. VSX2 was enriched in Retinal progenitors (log2 fold change to background: 0.67, adjusted p-value: 3.82 x 10⁻¹⁵²), Transitory (log2 fold change to background: 1.14, adjusted p-value: 6.82 x 10⁻²⁹⁰), Muller glia (log2 fold change to background: 0.75, adjusted p-value: 2.38 x 10⁻⁴⁴), Amacrine cells (log2 fold change to background: 0.41, adjusted p-value: 1.50 x 10⁻⁸), Bipolar cells (log2 fold change to background: 2.64, adjusted p-value: 7.01 x 10⁻¹⁵²) cell clusters. All the enriched gene information is available in Additional Table 4. For the following single-cell ATAC-seq analysis, we provide UMAP results highlighting Retinal progenitors ( Figure 2E) and Transitory ( Figure 2F) cell clusters, and feature plots for NOTCH1 ( Figure 2G) and RAX ( Figure 2H). Similar to the previous data ( Figure 1E-H), feature plot analysis suggested that NOTCH1 and RAX were expressed in Retinal progenitors and Transitory cell clusters.

To examine epigenetic changes underlying NOTCH1 expression dynamics, we examined single-cell ATAC-seq data from human fetal eyes at days 59, 74, and 78 from GSE183684.²⁰ UMAP analysis confirmed that this data contains Retinal progenitors and Transitory cell clusters, along with other differentiated cell clusters, including Cones ( Figure 3A-C). The cell number in each cluster is available in Additional Table 5. We examined Tn5 insertion frequency in the genomic NOTCH1 region, including 100 kb upstream and downstream, with the Signac pipeline ( Figure 3D). To identify the statistically significant Tn5 insertion sites, we used the “Links” function in the Signac pipeline to obtain z-scores and p-values (Additional Table 6). To examine the transcription factor binding sites of RAX, PAX6, LHX2, and VSX2 in those Tn5 insertion sites, we used the FIMO tool and identified potential transcription factor binding sites (Additional Table 7). By comparing the Tn5 insertion sites and the potential transcription factor binding sites, we found a statistically significant Tn5 insertion site in the genomic NOTCH1 region (chr9: 136542126–136542839) containing a transcription factor binding site (chr9: 136542636–136542645) for RAX, but not for PAX6, LHX2, and VSX2. This region is indicated with an arrowhead in the whole genomic NOTCH1 region ( Figure 3D) and the focused region ( Figure 3E). We also performed footprinting analysis and found that RAX-binding is frequently observed in Retinal progenitors, Transitory, and Muller glia cell clusters ( Figure 3F), although this analysis is genome-wide and not specific to the genomic NOTCH1 region. Importantly, the Tn5 insertion event is frequent in Retinal progenitors, Transitory, Amacrine cells, Cone precursors, Cones, Rods, and Muller glia compared with Bipolar cells, Horizontal cells, and Retinal ganglion cells. As RAX is not a classic pioneer factor, in contrast to SOX2 and PAX6, this observation raises the possibility that fine-tuned RAX and NOTCH1 action in Retinal progenitors, Transitory, Amacrine cells, Cone precursors, Cones, Rods, and Muller glia is supported by chromatin remodeling, along with regulated RAX expression.

The involvement of Notch signaling in cell fate choices is well documented, including in Drosophila neurogenesis⁴⁶ and mammalian biliary development.⁴⁷ A computer simulation study suggested that the regulation of Notch receptor expression is necessary for proper signal transduction in the developing tissue.⁴⁸ To the best of our knowledge, few studies have used genome-wide investigations for regulatory mechanisms of Notch receptor gene expression. Although the accessibility of the NOTCH1 locus is known to decrease during retinal cell differentiation,²⁰ how this chromatin remodeling affects NOTCH1 expression, along with candidate CREs, is still unknown. In this study, we hypothesized that chromatin remodeling and transcription factor levels are involved and sought possible examples using a public single-cell RNA/ATAC-seq dataset.²⁰ We used an already processed dataset from another study that intensively examined molecular characteristics of developing retinal cells⁴ to enable reliable annotation of the cells in the single-cell RNA/ATAC-seq dataset.²⁰ We observed prominent NOTCH1 expression both in the reference and query single-cell RNA-seq datasets, although the involvement of other Notch receptors in the developing retina was also reported.^9–14 Therefore, in the single-cell ATAC-seq analysis, we focused on the NOTCH1 locus, including 100 kb upstream and downstream of its coding region. We observed distinct Tn5 insertion frequencies by differentiation status and, importantly, included one RAX-binding site within open chromatin regions. This RAX-binding site is open in Retinal progenitors, Transitory, Amacrine cells, Cone precursors, Cones, Rods, and Muller glia compared with Bipolar cells, Horizontal cells, and Retinal ganglion cells, which is in line with previous studies.^7,49

The chromatin remodeling region (chr9: 136542126–136542839) containing a RAX-binding site (chr9: 136542636–136542645) partly overlaps with a candidate CRE (chr9: 136542638–136542845, Accession #: EH38E3919891) provided by SCREEN v4 (Zhiping Weng Lab, UMass Chan Medical School, https://screen.wenglab.org, Access date: Sep.25, 2026).⁴⁴ However, one limitation is that the score of this RAX-binding site in the JASPAR track of the UCSC Genome Browser is low (226, corresponding to a p-value of 10^−2.26), below the default display threshold of 400. Therefore, the RAX-binding site identified by our integrated analysis of transcription factor expression and binding motifs should be considered a candidate requiring future experimental validation. Another limitation is that the single-cell RNA/ATAC-seq dataset would significantly affect the results. We used the three early time points (days 59, 74, and 78) in a single-cell RNA/ATAC-seq dataset (GSE183684)²⁰ to balance computational burden and the number of retinal progenitors, although performing investigations using the full dataset could lead to different results. Indeed, the Retinal progenitors and Transitory cell clusters are widely distributed and partly overlap in the UMAP results ( Figure 1E-F, Figure 2E-F, and Figure 3B-C). This is mainly because these cell clusters show similar gene expression before terminal differentiation, as shown in the original paper for this annotation,⁴ and, in the query dataset, partly because we used only the three early time points, which may have excluded more differentiated cells. This limitation also supports the need for future experimental validation.

One of the transcription factors linked to NOTCH1 is PAX6. PAX6 haploinsufficiency has been linked to aniridia,⁵⁰ an ophthalmological disease often accompanied by fovea hypoplasia. AAV-mediated PAX6 overexpression in the Small eye (Sey) mouse (a mouse model of aniridia)^51,52 rescued ganglion cell layer thickness, which was preceded by increased Notch1 expression.⁵³ However, the timing of this overexpression (postnatal day 21) mimicked the average age of diagnosis in human patients (22.1 months)⁵⁴ and was much later than the timing of the fetal samples re-analyzed in this study. It is known that PAX6’s DNA-binding affinity is so delicate that disease-causing PAX6 mutations alter rather than ablate this protein-DNA binding⁵⁵ or require additional factors other than alterations in their DNA-binding affinities for their roles in pathogenesis.⁵⁶ We observed PAX6 expression in a wide variety of cell clusters, including Retinal progenitors and Transitory cell clusters, both in the reference and query single-cell RNA-seq datasets. In embryonic stem cells, SOX2 (one of the Retinal progenitors markers)⁴ binds to “hidden” chromatin regions in the nucleosome, and then PAX6 binds to the SOX2-bound “unwrapped” genomic regions to facilitate neural differentiation.⁵⁷ Expression of these pioneer factors⁵⁸ was indeed observed in the Retinal progenitors and Transitory cell clusters in the reference and query single-cell RNA-seq datasets. Given that rax upregulated notch1 transcription,⁷ it is possible that PAX6 and SOX2 unwrap multiple genomic regions before RAX binding to the NOTCH1 locus; however, we were not able to identify PAX6-binding sites in the NOTCH1 locus, although the delicate nature of PAX6 binding might be involved. Although this point needs further investigation, it does not jeopardize the finding that chromatin remodeling occurs at the RAX-binding site in the NOTCH1 locus. As RAX expression is observed in Retinal progenitors, Transitory, Muller glia, Cone precursors, and Cones cell clusters in the two single-cell RNA-seq datasets, it is likely that NOTCH1 expression in Retinal progenitors and Transitory cell clusters is in part regulated at the RAX protein level. However, it is also possible that the chromatin remodeling that we observed in this study could help fine-tune regulation of NOTCH1 expression. In conclusion, we provide a chromatin remodeling region in the NOTCH1 locus that contains a RAX-binding site, raising the possibility of regulation of NOTCH1 expression at the transcription factor protein and chromatin levels.

Y. W.: Data Curation, Formal Analysis, Software, Visualization, Writing – Original Draft Preparation.

S.K.: Writing – Original Draft Preparation.

T.N.: Writing – Original Draft Preparation.

S.T.: Supervision, Writing – Review & Editing.

M.Y.: Conceptualization, Formal Analysis, Funding Acquisition, Methodology, Project Administration, Software, Writing – Original Draft Preparation.

Grant information

This study was supported by the JSPS KAKENHI Grant-in-Aid for Early-Career Scientists (grant no. JP23K14429) to M.Y.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Article Versions (6)

Copyright

© 2026 Watabe Y et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Version 5

Version 4

Version 3

Version 2

Version 1

Version 6

VERSION 6 PUBLISHED 06 Jan 2025

Comment