https://orcid.org/0009-0001-4981-1950
This is an uncorrected proof.
Figures
Abstract
Müller glia arise from late-stage retinal progenitor cells (RPCs) as a distinct lineage that diverges from neurogenic trajectories. Here, we identify the maternally imprinted gene Plagl1 as a key transcriptional regulator of gliogenesis in the murine retina. Plagl1 is expressed during the RPC-to-glia transition and is dynamically regulated in Müller glia following injury. To define its developmental role, we analyzed Plagl1⁺/⁻pat null mutant retinas at postnatal day 7 (P7), when central retinal gliogenesis is complete. In the absence of Plagl1, Sox9 ⁺ glial/precursor cells were displaced and proliferated ectopically, with structural dysmorphologies, reactive gliosis, and impaired visual processing persisting into later postnatal stages. Bulk RNA-seq and ATAC-seq revealed widespread reductions in chromatin accessibility and transcriptional dysregulation affecting epigenetic modifiers, translational machinery, fate-specifying transcription factors, cell cycle regulators, and signaling pathways. Single-cell pseudobulk analysis showed that Plagl1 loss disrupts chromatin, transcriptional, and translational programs specifically within Sox9 ⁺ cells, encompassing Müller glia and precursor populations, pinpointing these cells as the source of defects in Plagl1⁺/⁻pat retinas. Notch signaling was elevated in Plagl1-deficient glia, and genetic activation at P14 displaced Sox9 ⁺ glial cells, without inducing proliferation. Similarly, conditional deletion of Plagl1 in postnatal Müller glia at P14 disrupted positioning and not cell cycle exit, confirming a cell-autonomous requirement for Müller glia positioning that is independent of proliferation control. Since these conditional manipulations could only be performed at P14 at the earliest, they reveal Plagl1’s later functions in postmitotic glia and complement, rather than mirror, the earlier P7 mixed RPC/glial null phenotype. Together these findings establish Plagl1 as a critical regulator of the late-stage RPC to Müller glia transition, acting through coordinated control of chromatin accessibility and gene expression programs to ensure timely cell cycle exit. This function aligns with Plagl1’s broader tumor suppressor role in stabilizing postmitotic, differentiated cell states across tissues.
Author summary
As the only glial cells in the retina, Müller glia play an essential role in maintaining tissue homeostasis and supporting retinal integrity and function. During development, Müller glia are derived from multipotent RPCs, which also give rise to the six types of retinal neurons, all of which differentiate in a defined temporal order. Müller glia are among the last retinal cells to differentiate, transitioning from cycling RPCs into post-mitotic glial cells. Here, we show that the RPC-to-Müller glial cell transition is orchestrated by the maternally imprinted gene Plagl1. Plagl1 encodes a transcription factor that is initially expressed in RPCs and later persists in retinal Müller glia. In the absence of Plagl1, Müller glia initiate glial-specific marker expression but fail to form a mature monolayer and continue to proliferate ectopically. Mechanistically, loss of Plagl1 compacts open chromatin, induces widespread changes in gene expression, and elevates several signaling pathways including Notch, which, when activated, disrupts the Müller glial monolayer. Thus, Plagl1 guides the RPC-to-Müller glial cell transition, providing new insights into this unique developmental process.
Citation: Touahri Y, Pak A, David LA, Hanna J, Liu H, Xiao Y, et al. (2026) Plagl1 regulates the retinal progenitor cell to Müller glial cell transition. PLoS Genet 22(3): e1012020. https://doi.org/10.1371/journal.pgen.1012020
Editor: Fengwei Yu, National University of Singapore, SINGAPORE
Received: February 21, 2025; Accepted: January 1, 2026; Published: March 18, 2026
Copyright: © 2026 Touahri 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 author and source are credited.
Data Availability: Data and Code Availability • Bulk RNAseq and ATACseq data associated with this manuscript were deposited at GEO and can be accessed at: RNA-seq data: GEO: GSE173942 and ATAC-seq data: GEO: GSE173853 • scRNAseq data was deposited at GEO: GSE228191 • Uncropped Western blot images were deposited in Mendeley Data, V1, doi: 10.17632/4xk8c2zhxt.1 • This paper does not report original code. • All numerical data that underlies graphs and summary statistics are reported in S12 Table.
Funding: This work was supported by a Natural Sciences and Engineering Research Council of Canada Discovery Grant (RGPIN-2017-03649 and RGPIN-2024-04345), both to CS, and Canadian Institutes of Health Research (CIHR) Operating Grants to CS (MOP-142338), CS and YT (PJT-180243), CS and IA (PJT – 400973). This work was also supported by a CIHR Project Grant to CJG (MOP-165973). YT was supported by a U of Calgary Eyes High Fellowship. LAD was supported by a University of Toronto Department of Laboratory Medicine and Pathobiology fellowship, CIHR Frederick Banting and Charles Best Masters award, and University of Toronto Vision Science Research Program (VSRP) award. JH was supported by a Canada Graduate Scholarship – Doctoral (CGS-D)/CIHR award, VSRP Scholarship, Ontario Graduate Scholarship and R.O. Torrance Bursary. HL was supported by a Natural Sciences and Engineering Research Council (NSERC) of Canada summer studentship. NT and LA were supported by studentships from the ACHRI/CIHR Training Grant at the University of Calgary. JZ was supported by a University of Toronto Department of Biochemistry Undergraduate Research Opportunity Program (UROP) summer studentship. MH was supported by an Alberta Innovates Health Solution (AIHS) summer studentship (MH). RD was supported by a CIHR Canada Hope Fellowship. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: CS holds the William Dixon Chair in Ophthalmology Research. IA holds a Tier 1 Canada Research Chair in Brain Repair and Regeneration.
Introduction
The retina is comprised of six neuronal cell types and a specialized glial population known as Müller glia, all of which originate from a common multipotent pool of retinal progenitor cells (RPCs) [1–5]. In mice, retinal differentiation unfolds in a defined temporal sequence between embryonic day (E) 11.5 and postnatal day (P) 11 [6–9]. Ganglion cells, horizontal cells and cone photoreceptors are primarily generated during embryonic development, while amacrine cells, bipolar cells, rod photoreceptors and Müller glia continue to emerge postnatally. Temporal identity transcription factors (TFs) such as Ikzf1/Ikzf4, Pou2f1/Pou2f2, Casz1 and Nfia/b/x, are expressed in RPCs and guide early versus late cell fate decisions [10–14]. However, the gene regulatory networks (GRNs) that direct Müller glia differentiation and maintain their identity remain incompletely defined.
RPCs are proliferative neuroepithelial cells that undergo interkinetic nuclear migration (INM), a cell cycle phase-dependent process in which nuclei translocate apically during G2/M-phase and basally during S-phase [15]. Upon differentiation, RPCs exit the cell cycle and halt INM. Single-cell multiomic analyses have stratified RPCs into pre-neurogenic neuroepithelial cells, early and late-stage RPCs and fate-restricted neurogenic RPCs [16,17]. Neurogenic RPCs generate the six principal neuronal types, whereas Müller glia are thought to emerge directly from late-stage RPCs [16]. This model is supported by clonal analyses in zebrafish and time-lapse imaging of dissociated rat RPCs which show that Müller glia are typically the final cell type produced within clones [18,19]. While neuronal output appears stochastic, with RPCs adopting neuronal fates in ratios that reflect cell-type abundance [18,19], blastomere transplantation experiments in zebrafish reveal that each RPC generates a single Müller glial cell per clone, regardless of clone size, suggesting a deterministic mode of glial specification [20]. However, conventional lineage-tracing methods rely on observable mitoses and cannot resolve division-independent transitions, leaving open the possibility that glial commitment may occur without intervening cell divisions.
Complementary single-cell transcriptomic and epigenomic analyses reveal distinct gliogenic trajectories that diverge from neurogenic lineages [16,17]. Several TFs are co-expressed in late-stage RPCs and Müller glia, including Nfia/b/x, Lhx2, Sox2 and Sox9 [14,21,22], suggesting that the GRNs governing late-stage RPC identity and gliogenesis are closely linked. In contrast, neurogenic RPCs express proneural factors such as Ascl1 and Hes6, which are absent from the Müller glial program [21]. These divergent transcriptional profiles underscore a molecular bifurcation between neurogenic and gliogenic trajectories.
Upon terminal differentiation, Müller glia acquire essential functions for retinal homeostasis, maintaining structural integrity, guiding light transmission to photoreceptors, and providing neurotrophic support to retinal neurons [23,24]. In cold-blooded vertebrates, Müller glia also function as stem cell-like cells capable of regenerating retinal neurons following injury. In zebrafish, retinal damage triggers Müller glia to delaminate, reposition their nuclei [25,26], de-differentiate into proliferative progenitor cells, and re-differentiate into retinal neurons that integrate into damaged neural circuits [27]. In contrast, mammalian Müller glia exhibit limited regenerative capacity. In rodents, injury induces reactive gliosis, with only transient, proliferation observed within the first two days post-injury [28–32]. In mice, nuclear displacement varies by injury type, with light damage causing greater disruption than chemical insults [33]. Bulk and single-cell RNA-seq and ATAC-seq profiling of injured retinas of fish and mice has identified three Müller glial states: resting, reactive, and restoration to resting, the latter unique to mice [28]. While reactive Müller glia upregulate a conserved, pro-inflammatory, rapid response across species, slow-response programs diverge: zebrafish activate cell cycle and neurogenic competence genes, whereas mice favor reversion to quiescence and ribosomal biogenesis [28].
Imprinted genes, which are mono-allelically expressed based on parent-of-origin, are regulated by selective methylation and silencing of either the maternal or paternal allele. Despite their functional diversity, including roles in transcription, signaling and cell cycle regulation, many imprinted genes converge on a shared role in maintaining cellular quiescence in mature somatic or stem cell populations across tissues, including lung, hematopoietic system, skeletal muscle, and brain [34–48]. A meta-analysis of imprinted gene expression identified co-regulated subnetworks, suggesting evolutionary pressure toward coordinated control of growth and differentiation [34,49]. Plagl1, a maternally imprinted gene encoding a zinc finger transcription factor, is a member of one such network [34,49]. Numerous studies support the role of PLAGL1 as a negative regulator of cell division. PLAGL1 is located on human chromosome 6q24-25, a locus associated with growth suppression, with intrauterine grown restriction observed when the expressed paternal allele is duplicated [50,51]. PLAGL1 also functions as a bona fide tumor suppressor gene, with loss of PLAGL1 expression observed in breast, ovarian, prostate, pituitary, gastric, non-Hodgkin’s lymphoma and colorectal cancers [51–56]. Finally, Plagl1 induces cell cycle arrest when expressed in epithelial, cortical and RPCs [57,58]. These properties position Plagl1 as a candidate regulator of fate transitions in the retina, where the balance between neurogenesis and gliogenesis depends on tightly controlled exit from the cell cycle.
During retinal development, Plagl1 is expressed in RPCs and coordinates neuronal differentiation, with Plagl1 loss leading to ectopic proliferation and the overproduction of rod photoreceptors and amacrine cells in mice [58,59], consistent with its growth suppressive properties across tissues [52]. Here we show that in the early postnatal retina, Plagl1 expression persists in RPCs and is newly initiated in Müller glia. By P7, when differentiation is complete in the wild-type central retina, Plagl1 deficiency leads to ectopic proliferation accompanied by widespread changes in gene expression and chromatin accessibility. Together, these findings establish Plagl1 as a critical regulator of the late-stage RPC-to-Müller glia transition, ensuring timely cell-cycle exit and stabilizing Müller glial identity through coordinated transcriptional and epigenetic control.
Results
Plagl1 is expressed in Müller glia and dynamically regulated in response to retinal damage
Plagl1 is expressed in both early- and late-stage RPCs during embryonic and early postnatal development [58]. To assess whether Plagl1 expression persists in postmitotic retinal cells, we performed RNAscope in situ hybridization on P7 retinas, focusing on the central retina where cellular differentiation is complete. Plagl1 transcripts were detected in the inner nuclear layer (INL) and ganglion cell layer (GCL), overlapping with Sox9 expression (Fig 1A), a marker of RPCs, post-mitotic precursors and Müller glia in the INL and astrocytes in the GCL [60]. To further investigate Plagl1 expression in mature glial populations, we analyzed an RNA-seq dataset derived from GFP-enriched Müller glia sorted from 2-month-old GLASTCreER;Sun1-sGFP mice [28]. Plagl1 and Sox9 were co-expressed in GFP+ Müller glia alongside additional late-stage RPC-glial markers, whereas transcripts associated with other retinal cell types were confined to GFP-negative fractions (Fig 1B). Immunohistochemical analysis confirmed robust co-expression of Plagl1 and Sox9 at the protein level in P7 retinas; 91.8 ± 0.8% of Plagl1+ cells co-expressed Sox9, and 78.9 ± 2.2% of Sox9+ cells were Plagl1+ (Fig 1C). These findings demonstrate that Plagl1 transcription and translation persist in differentiated retinal layers and largely overlap with Sox9+ populations, consistent with expression in Müller glia, RPCs, committed precursors and/or astrocytes.
(A) RNAscope in situ hybridization (ISH) with Plagl1 (green) and Sox9 (red) riboprobes on P7 retina. Scale bar = 100 µm. Right side panels are high magnifications of boxed areas. Scale bars = 25 µm. (B) RNAseq analysis of P60 Müller glia (GFP+ cells sorted from GLASTCreER;Sun1-sGFP mice) extracted from accession GSE135406, showing Plagl1 expression along with Müller glial markers (Sox9, Vim, Sox2, Glul, Rlbp), bipolar (Vsx1, Vsx2) and photoreceptor (Crx, Otx2) markers in GFP+ relative to GFP- cells. (C) Immunostaining with Plagl (green) and Sox9 (red) antibodies on P7 retina. Blue is DAPI nuclear stain. Scale bars = 25 µm. (D) tSNE-dimensionality reduction to represent retinal cell clustering following NMDA treatment (data extracted from [28]). Relative Plagl1 expression in astrocytes = 0.108, resting Müller glia = 0.236 activated Müller glia = 0.308. (E) Heat map showing Plagl1 transcript distribution in each of the clusters in panel D. (F) Temporal profiling of Plagl1 expression in Müller glia after 3, 6, 12, or 36 hr (NMDA damage) or 4, 16, or 24 hr (light damage). RNAseq data extracted from GSE135406. (G) Immunolabeling of GFAP, Rho, Arr3, and Sox9 on adult wild-type retinas at 3 days post-PBS or -MNU treatment. Blue is DAPI nuclear stain. Scale bars = 50 µm. (H) Sox9+ cell counts at 3 days post-MNU treatment. N = 3 control, N = 5 Plagl1+/-pat. Mean values and error bars representing s.e.m. were plotted. p-values were calculated with an unpaired t-test. (I) RNAscope ISH for Plagl1 and Sox9 on adult wild-type retinas at 3 days post-PBS or -MNU treatment. Blue is DAPI nuclear stain. Scale bars = 50 µm. (J) Gfap and Plagl1 expression quantified by qPCR in retinas at 3-, 6-, 12-, 24- and 72-hr post-treatment with MNU or PBS. N = 4 for all time points. Mean values and error bars representing the standard error of the mean (s.e.m.) were plotted. p-values were calculated with an unpaired t-test comparing individual time points to PBS control. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium.
Müller glia exhibit dynamic transcriptional responses following retinal injury, characterized by the upregulation of gliogenic markers such as GFAP [28–32]. To evaluate whether Plagl1 expression is modulated by injury, we analyzed transcriptomic data from enriched Müller glia isolated from two-month-old mice subjected to N-methyl-D-aspartate (NMDA) or light damage (LD) [28]. In this single-cell (sc) RNA-sequencing dataset, Plagl1 was expressed at the highest levels in activated followed by resting Müller glia (Fig 1D and 1E). Complementary bulk RNA-seq data collected at 3-, 6-, 12-, and 36-hours following NMDA exposure, and 4-, 16-, and 24-hours following LD revealed a rapid induction of Plagl1 transcripts within 3–4 hours of damage, followed by a decline below baseline levels by 12–16 hours (Fig 1F). These temporal dynamics suggest that Plagl1 functions as an early injury-responsive gene in Müller glia.
To define the temporal dynamics of Plagl1 expression following photoreceptor degeneration, two-month-old wild-type mice were treated with N-methyl-N-nitrosourea (MNU), which induces selective photoreceptor loss [61]. Retinal tissues were analyzed at 2-, 3-, 7-, and 21-days post-injection. Reactive gliosis was evident at all time points, marked by GFAP upregulation and disorganization of Sox9 ⁺ cells, as shown by immunostaining and RNAScope (Figs 1G, 1I, and S1A–S1D). Importantly, Sox9 ⁺ cell numbers remained unchanged, indicating that their disorganization did not arise from aberrant proliferation (Fig 1H).
MNU-induced photoreceptor degeneration progressed rapidly: by day 3, photoreceptor degeneration was already pronounced, and by days 7 and 21, the ONL was largely absent, as indicated by loss of Rhodopsin (Rho) protein and Rcvrn mRNA (rods), together with loss of Arr3 protein (cones) (Figs 1G, 1I, S1A and S1B). At day 3, RNAscope revealed ectopic Plagl1 and Sox9 transcripts in regions above the degenerating ONL that lacked DAPI+ nuclei, likely reflecting non-specific trapping of RNA within the debris-rich tissue and not true cellular expression (Fig 1I). Consistent with this interpretation, such apical accumulation was absent at 7- and 21- days post-injury, when the retinal architecture had stabilized (S1A and S1B Fig). Within the INL, Plagl1 expression was downregulated by day 3 post-injury, as confirmed by RNAscope and qPCR (Figs 1I-1J and S1A-S1D). Notably, a transient upregulation of Plagl1 was detected as early as 3 hours after MNU exposure, preceding GFAP induction, and remained elevated for approximately 12 hours before declining to baseline (Figs 1I-1J, S1C, and S1D).
Together, these findings demonstrate that Plagl1 is expressed in Müller glia and undergoes dynamic regulation following retinal injury, with transient early induction and sustained downregulation coinciding with the shift from resting to reactive states.
Plagl1 is required for timely cell-cycle exit and proper organization of Sox9 ⁺ Müller glia and glial precursors
Given its expression in late-stage RPCs and Müller glia, we hypothesized that Plagl1 regulates the transition from a progenitor to glial fate. To test this, we examined Plagl1+/-pat null mutant retinas, which lack Plagl1 expression due to maternal allele silencing and paternal allele mutation [34,49,58]. Since Plagl1 functions as a tumor suppressor gene that promotes cell cycle exit across tissues [51–56], including the embryonic retina [58], we first assessed proliferation at P7, when differentiation is complete in the central retina. A 30-minute BrdU pulse followed by Sox9 immunostaining showed that Sox9+ cells formed a compact monolayer of BrdU-negative cells in the center of wild-type retinas, comprising postmitotic committed precursors and Müller glia (Fig 2A). In contrast, some Plagl1+/-pat retinas displayed dispersed Sox9+ cells, particularly within dysmorphic regions of the central retina, where ONL rosettes formed (Fig 2A). Such rosettes are associated with defects in adherens junctions at the outer limiting membrane (OLM), where Müller glial endfeet normally anchor photoreceptor inner segments to maintain epithelial integrity [62,63]. Both dysmorphic and non-dysmorphic regions contained scattered ectopic BrdU+ cells, 73.9 ± 6.3% of which co-expressed Sox9, indicating that they are Müller glia and/or late-stage RPCs that failed to exit the cell cycle (Fig 2B). Ectopic proliferation was not observed at P14 or P21, suggesting a transient defect restricted to the early postnatal period. Thus, loss of Plagl1 delays cell cycle exit and disrupts the spatial organization of Sox9+ glial and precursor cells, likely reflecting aberrant proliferation during the final stages of Müller glial maturation.
(A) Co-immunostaining of Sox9 and BrdU on P7 wild-type and Plagl1+/-pat retinal sections injected with BrdU for 30 min prior to harvest (A). Blue is DAPI nuclear stain. Images in second row are high magnifications of the boxed areas in the upper panels. Scale bars: 100µm. (B) Graphs show quantification of BrdU+ cells, and the percentage of BrdU+ cells that are Sox9+ in P7 wild-type and Plagl1+/-pat retinal sections injected with BrdU 30 min prior to dissection at P7. wild-type: N = 9, Plagl1+/-pat N = 8. Mean values and error bars representing s.e.m. were plotted. p-values calculated with one way ANOVA and post-hoc Tukey test. (C) BrdU birthdating (P7 BrdU to P14 dissection) in vivo, followed by BrdU co-immunostaining with Sox9 on P14 wild-type and Plagl1+/-pat retinal sections (C). Blue is DAPI counterstain. Scale bars: 75µm. (D) Quantification of BrdU birthdating (P7 BrdU to P14 dissection), showing the percentage of BrdU+ cells that are Sox9+ in P14 wild-type and Plagl1+/-pat retinas in vivo. wild-type: N = 5, Plagl1+/-pat N = 5. Mean values and error bars representing s.e.m. were plotted. p-values calculated with unpaired t-test. (E) Co-immunostaining for Sox9 and Ki67 in P7 wild-type retinas. Blue is DAPI nuclear stain. Scale bars: 200µm. (F) Co-immunostaining for Sox9 and Ki67 in P7 Plagl1+/-pat retinas. Blue is DAPI nuclear stain. Scale bars: 100µm. (G) Quantification of Ki67+ cells/field. wild-type: N = 3, Plagl1+/-pat N = 3. Mean values and error bars representing s.e.m. were plotted. p-values calculated with unpaired t-test.
To determine whether the progeny of these ectopically proliferating cells persist, we performed BrdU birthdating by labeling dividing cells at P7 and examining retinas at P14. In wild-type central retinas, BrdU⁺ cells were absent, consistent with the cessation of central RPC proliferation by P7 (Fig 2C). In contrast, BrdU⁺ cells were readily detected in P14 Plagl1+/-pat central retinas, even outside of rosette-like structures (Fig 2C). Among these ectopic BrdU⁺ cells, 43.2 ± 1.6% expressed Sox9 (Fig 2D), indicating that the ectopically dividing glial precursors persist in the absence of Plagl1. To confirm that BrdU⁺ cells reflected active proliferation rather than passive BrdU uptake during apoptosis, we performed Sox9-Ki67 co-immunolabelling. In wild-type retinas, Sox9+ cells were present throughout the central-to-peripheral axis, whereas Ki67+ cells were confined to the peripheral ciliary margin (CM) at P7 (Fig 2E). In Plagl1+/-pat retinas, however, ectopic Ki67+ cells appeared in the central retina, often within rosettes, some of which co-expressed Sox9 (Fig 2E).
To assess whether cell death contributed to these abnormalities, we immunostained for cleaved caspase 3 (cc3). Apoptosis is a normal feature of embryonic and early postnatal retinal development [64], as confirmed by the presence of cc3+ cells in the P5 wild-type retina (S2A Fig). However, no cc3+ cells were observed in wild-type or mutant retinas at P7 or P21, despite the presence of ectopic cell proliferation at P7 (S2A Fig). This contrasts with embryonic Plagl1+/-pat retinas, in which apoptosis is reduced [58]. Together, these findings indicate that, in the absence of Plagl1, a subset of Sox9 ⁺ cells fail to exit the cell cycle by P7, and become disorganized, likely as a secondary consequence of sustained proliferation rather than altered cell death.
Plagl1 is required to maintain retinal integrity and for proper visual function
To assess the long-term consequences of Plagl1 loss, we collected Plagl1+/-pat animals at juvenile (P14, P21, P30) and early adult (P45, P60) stages. At P21, we performed optical coherence tomography (OCT) imaging to obtain a non-invasive, high-resolution assessment of retinal morphology. Infrared en face images were used to capture fundus landmarks and to align scans across animals, ensuring that wild-type and mutant retinas were compared at similar eccentricities and along matched axes (Fig 3A). In wild-type retinas, cross-sectional OCT imaging revealed sharply defined laminar architecture, with clear contrast between hyperreflective synaptic layers and hyporeflective nuclear layers. (Fig 3A). In contrast, Plagl1+/–pat retinas exhibited focal hyporeflectivity within the inner plexiform layer (IPL), consistent with a disruption of the laminar architecture (Fig 3A). These changes suggest structural disorganization involving both nuclear and synaptic compartments.
(A) Infrared en face fundus images (left) and corresponding OCT-derived cross-sectional scans (right) from P21 wild-type (N = 5) and Plagl1+/–pat (N = 4) retinas. Green horizontal lines on the fundus images indicate the location of the cross-sectional OCT scans. Scale bars: 200 µm. (B) DAPI staining of P21 wild-type and Plagl1+/-pat retinas, showing retinal ectopias (asterisks) and rosettes (arrowheads). Right side panels are high magnification images of boxed areas. Scale bars: 25µm (C) Quantification of rosettes and ectopias in wild-type and Plagl1+/-pat retinas at P7, P14, P21 and P60. N = 5 for all stages. (D-F) Immunostaining of in P21 wild-type and Plagl1+/-pat retinas for Rhodopsin and Arr3 (D), ZO1 (E), Iba1 and CD11b (F). Arrowheads mark ectopias and rosettes. Blue is DAPI counterstain. Scale bars: 25µm (G) Staining for Isolectin B4 in P30 wild-type and Plagl1+/-pat flatmounted retinas. Arrowheads show nodes and tortuous vessels. Right images are high magnifications of the boxed area in left images. Scale bars: 25µm (H) Immunostaining for vimentin (Vim) in P14 wild-type and Plagl1+/-pat retinal sections. Arrow -heads point to ectopic Vim expression in Plagl1+/-pat retinas. Blue is DAPI counterstain. Scale bars: 25µm (I) Immunostaining for GFAP in P14, P21 and P45 wild-type and Plagl1+/-pat retinal sections. Arrowheads point to ectopic GFAP expression in Plagl1+/-pat retinas. Blue is DAPI counterstain. Scale bars: 25µm (J) Immunostaining for pERK in P14 wild-type and Plagl1+/-pat retinal sections. Arrowheads point to pERK upregulation in Plagl1+/-pat retinas. Blue is DAPI counterstain. Scale bars: 25µm (K) Western blotting for pERK in P14 wild-type and Plagl1+/-pat retinas. N = 6 wild-type and N = 6 Plagl1+/-pat retinas. Violin plots display the data distribution, with superimposed mean values and error bars indicating the s.e.m..p-values calculated with unpaired t-test. (L) Immunostaining for Glul staining in P14 wild-type and Plagl1+/-pat retinal sections. Arrowheads point to Glul gaps in Plagl1+/-pat retinas. Blue is DAPI counterstain. Scale bars: 25µm (M) Western blotting for Glul in P14 wild-type and Plagl1+/-pat retinas showing decrease of Glul protein levels. N = 6 wild-type and N = 6 Plagl1+/-pat retinas. Violin plots display the data distribution, with superimposed mean values and error bars indicating the s.e.m. p-values calculated with unpaired t-test. (N,O) Full field ERG recordings of P30 wild-type and Plagl1+/-pat mice under scotopic (N) and photopic (O) conditions. N = 6 (wild-type), N = 5 (Plagl1+/-pat). Plots show a- and b-wave amplitudes and implicit times. Statistics: Kruskall-Wallis. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; OLM, outer limiting membrane.
Histological sections from Plagl1+/–pat retinas at P7, P14, P21, and P60 revealed additional dysmorphologies, including ectopias, defined as protrusions of ONL nuclei into the subretinal space, and rosettes, which were focal ONL dysplasias that also deformed the adjacent INL (Figs 3B-3C and S2B). Rod (Rho+) and cone (Arr3+, S-opsin+, M-opsin+) photoreceptors were the most disrupted, with their outer segments internalized into the central lumens of retinal rosettes at P21 (Figs 3D and S2C). In contrast, amacrine (Pax6+, syntaxin+, calretinin+), bipolar (Vsx2+, PKC+), horizontal (calbindin+), and ganglion (Pax6+, Brn3a+) cells were generally retained in appropriate laminar positions, although local disruptions were evident near rosettes (S2C Fig). Rlbp1+ Müller glial processes were also abnormal in P21 Plagl1+/–pat retinas, as was the OLM, where Müller glial processes terminate, marked by ZO-1 immunostaining of adherens junctions (Figs 3E and S2C). Plagl1+/-pat retinas displayed additional spontaneous features of an injury response, including infiltration of CD11b+ macrophages and Iba1+ microglia (Fig 3F) and abnormal node-like structures and tortuosity of the Isolectin-B4+ vasculature (Fig 3G).
The morphological abnormalities observed in Plagl1+/–pat retinas resembled phenotypes reported in mutant mouse models with underlying Müller glial dysfunction [23,65,66]. To investigate whether Plagl1 loss induces reactive gliosis, we examined markers associated with stress-induced Müller glial activation, including the intermediate filament proteins glial fibrillary acidic protein (GFAP) and Vimentin (Vim). Vimentin labeled Müller glial cell bodies, processes, and endfeet in both P14 wild-type and Plagl1+/–pat retinas, however expression was markedly elevated in mutants (Fig 3H). In wild-type retinas, GFAP expression was restricted to astrocytes lining the GCL, whereas in Plagl1+/–pat retinas, GFAP was also detected in Müller glial processes at P14, P21, and P45, particularly in near rosette-like structures (Fig 3I). To further assess gliotic activation, we examined phosphorylation of ERK at T202/Y204 (pERK), a marker of cellular stress. pERK levels were elevated in P14 Plagl1+/–pat retinas, as confirmed by immunostaining and Western blot analysis (Fig 3J and 3K). Additionally, expression of glutamate-ammonia ligase (Glul, or glutamine synthetase), a key Müller glial enzyme typically downregulated during gliosis, was reduced in P14 mutant retinas by both assays (Fig 3L and 3M). Together, these findings indicate that Plagl1 loss triggers a reactive gliosis response in Müller glia, marked by cytoskeletal remodeling, stress signaling, and metabolic dysregulation.
To evaluate functional consequences of Plagl1 loss, we recorded full-field electroretinograms (ERGs) in P30 juvenile mice, after P25 when full visual acuity is normally acquired. Under both scotopic (Fig 3N) and photopic (Fig 3O) conditions, a- and b-wave amplitudes were comparable between wild-type and Plagl1+/–pat animals, indicating that gross retinal circuitry remained intact. However, implicit times for both a- and b-waves were significantly prolonged in Plagl1+/–pat mice under both adaptation states, suggesting delayed activation of rod- and cone-driven phototransduction and impaired inner retinal signal modulation (Fig 3N and 3O). These functional delays, in the context of preserved waveform amplitudes, point to a biochemical signal processing defect rather than a global synaptic failure. Together, these findings demonstrate that Plagl1 is required for maintaining retinal structural integrity, regulating Müller glial homeostasis, and ensuring timely transmission of visual signals.
Loss of Plagl1 thus disrupts laminar organization, induces reactive gliosis, and impairs the kinetics of retinal signal processing, highlighting its essential role in coordinating retinal development and function. Of note, the disrupted Sox9 ⁺ monolayer likely reflects a cell-autonomous effect of Plagl1 loss in Müller glia, where its expression is highest, rather than a secondary consequence of retinal dysmorphology. Moreover, the subtle ERG effects likely stem from localized structural defects rather than widespread retinal disruption, as also observed in Crb1 [62] and Dicer [67] mutant models.
Plagl1 regulates a large network of retinal genes controlling transcription, chromatin, translation, proliferation, and a glial fate
Plagl1 can function as a transcriptional activator or repressor depending on cellular context and target genes [52]. To investigate how Plagl1 regulates the transition from late-stage RPCs to Müller glia, we performed bulk RNA-seq on P7 wild-type (N = 4) and Plagl1+/-pat (N = 8) retinas, a stage when mutant cells exhibit ectopic proliferation. Differential expression analysis identified 2552 genes significantly altered in Plagl1+/-pat retinas, including 1410 upregulated and 1142 downregulated transcripts (adjusted p value< 0.05; fold change < -1.2 or >1.2; Fig 4A and S1 Table). Transcript coverage across the Plagl1 locus confirmed the absence of reads mapping to exons 5 and 6 in Plagl1+/-pat retinas (Fig 4B), consistent with the targeted deletion [49] and sustained maternal allele silencing previously observed in other cell types [68]. Thus, Plagl1 remains mono-allelically expressed in the postnatal retina, in contrast to some IGN genes that become biallelically expressed in adult somatic cells, such as Dlk1 [46] and Igf2 [45].
(A) Volcano plot of DEGs from bulk RNAseq analysis of P7 wild-type and Plagl1+/-pat retinas. (B) RNAseq read coverage of Plagl1 genomic locus in wild-type and Plagl1+/-pat showing absence of coverage of exon 5 in Plagl1+/-pat indicating that all the sequenced transcripts are of paternal origin. (C,D) Gene ontology terms elevated in P7 Plagl1+/-p retinas associated with Biological Processes (C) and KEGG pathway analysis (D). (E) Venn diagrams showing Müller glia-specific genes that are differentially expressed, followed by Müller glia specific genes that are up and down-regulated in Plagl1+/-pat retinas. (F) padj values for Müller glia specific genes among the DEGs in P7 Plagl1+/-pat retinas. (G) Mean FPKM values for select DEGs in P7 Plagl1+/-pat retinas were plotted with s.e.m.. padj values were computed using DESeq2.
Gene ontology (GO) analysis revealed enrichment of biological processes related to “covalent chromatin modification”, “positive regulation of neuronal differentiation”, “stem cell population maintenance” and “regulation of translation” (Fig 4C and S2 Table). Related terms were also enriched in Molecular Function and Cellular Component GO categories (S3A and S3B Fig). KEGG pathway analysis identified Notch signaling, implicated in both RPC proliferation [69–71] and Müller glial differentiation [69–73], as significantly enriched, along with AMPK signaling, which inhibits protein translation [74] (Fig 4D).
To assess whether Plagl1 loss affects genes enriched in Müller glia, we compared the Plagl1+/–pat DEGs to a Müller glia transcriptome generated from GLASTCreER;Sun1-sGFP mice [28]. Of the 2635 glial-enriched genes, 372 were differentially expressed, including 275 upregulated and 97 downregulated transcripts (Fig 4E). Genes significantly altered in Plagl1+/–pat retinas (padj<0.05) clustered into four major categories: transcription factors, translational regulators, chromatin modifiers/transcriptional repressors, and Notch signaling components, reflecting broad disruption of developmental and regulatory programs (Fig 4F). Statistically significant genes from annotated categories included TFs from the Nfi [14], Sox [73] Lhx [75], and Tead [76] families [75], as well as components of the Notch (Notch1, Notch2, Hes1, Hes5, Dll1, Jag2) and Shh (Smo, Gli2, Gli3) pathways (Fig 4G). While several of these factors are expressed in late RPCs and/or Müller glia, others are active in other retinal lineages, suggesting widespread transcriptional deregulation. As cell-type attribution is not possible with bulk RNA-seq, these findings reflect changes in gene expression across the whole retinal tissue rather than specific cellular compartments.
Consistent with widespread transcriptional changes, several chromatin modifiers and transcriptional repressors (Arid1a, Kdm4a, Smarca4, Atn1, Rcor2, Ncor2) were elevated in Plagl1+/-pat retinas (Figs 4F and S3C). In contrast, ribosomal genes (e.g., Rpl14, Rps21, Rpl37a, Rpl7, Rps14, Rps7) were among the most downregulated, while Eif4ebp2, a key negative regulator of protein translation, was upregulated (Figs 4F and S3D). qPCR validation confirmed increased expression of late RPC/Müller glia markers (Sox9, Lhx2, Hes1) and decreased expression of ribosomal genes (Rpl13, Rpl26) in P7 Plagl1+/-pat retinas (S3E Fig).
Finally, because Sox9+ cells proliferate ectopically in Plagl1+/-pat retinas, to further investigate the transcriptional basis, we compiled a curated list of 616 cell cycle-associated genes and assessed their expression in our RNA-seq dataset (S4A Fig and S3 Table). Of these, 90 genes were differentially expressed (14.6%) in Plagl1+/-pat retinas, including 50 down-regulated and 40 up-regulated (S4 Fig). Downregulated genes included known cell cycle progression inhibitors such as Cdkn2d, while upregulated genes included positive regulators such as Cdk6, Cdk11b, and Pak4. Of these, 64 deregulated genes (27 upregulated, 37 downregulated) are expressed in Müller glia, suggesting cell type-specific relevance (S4B Fig). We acknowledge that some canonical cell cycle drivers, including Ccne2 and Ccng1, were downregulated despite the presence of proliferating cells. This likely reflects the complexity of cell cycle regulation, as well as the cellular heterogeneity of the retina at P7, in which most cells are postmitotic. The mixed directionality of cell cycle gene expression may therefore reflect both cell type-specific regulation and dilution of signal in bulk tissue.
In summary, transcriptomic changes in Plagl1+/–pat retinas reflect dysregulation of key pathways governing chromatin remodeling, protein translation, and cell cycle control, collectively supporting a model in which Plagl1 orchestrates the coordinated exit from proliferation and initiation of glial differentiation.
Plagl1 loss results in a global reduction in chromatin accessibility
“Covalent chromatin modification” and “histone modification” were the two top GO Biological Process terms associated with DEGs in P7 Plagl1+/-pat retinas (Fig 4C), suggesting potential epigenetic dysregulation. To determine whether these transcriptomic changes were accompanied by alterations to chromatin accessibility, we performed ATAC-seq on P7 wild-type (N = 3) and Plagl1+/-pat (N = 3) retinas to identify nucleosome-sparse (open) regions of genomic DNA. Globally, we identified 833 differentially accessible regions (DARs) in Plagl1+/-pat retinas, of which 98% (814/833) showed decreased accessibility (Fig 5A), indicating widespread chromatin compaction. For downstream analyses, we focused on the 311 DARs located within ±5000 bp of transcription start sites (TSSs) that could be confidently assigned to genes. These DARs were predominantly centered around the TSS and enriched in promoter and intronic regions, with fewer located in intergenic, downstream, exonic, or 3′UTR regions (Fig 5B and 5C). This distribution reflects both the expected promoter bias of the ± 5 kb window and broader regulatory disruption across multiple genomic compartments.
(A) ATACseq MAplot showing differentially accessible chromatin regions in P7 Plagl1+/-pat vs wild-type retinas. (B) Distribution of the 311/833 gene-assigned DARs identified around the TSS in P7 Plagl1+/-pat retinas. (C) Distribution of the 311/833 gene-assigned DARs in promoter regions, introns, intergenic regions, immediate downstream regions, exons, 3’UTRs and 5’UTRs in P7 Plagl1+/-pat retinas. (D) Biological process GO terms associated with the 311 gene-assigned DARs. (E) Gene regulatory network associated with the P7 retina, with associated changes in Plagl1+/-pat retinas. Red circled genes are down-regulated and blue circled genes are upregulated.
Of the 311 gene-assigned DARs located within ±5 kb of TSSs, nearly all (308/311) exhibited decreased accessibility in Plagl1+/-pat retinas. Among these, 15 overlapped with down-regulated DEGs and 36 with up-regulated DEGs in mutants (S4 Table). Although the overall overlap between DARs and DEGs was modest, a few genes known to be expressed in late-stage RPCs and early Müller glia, including Lhx2, were differentially expressed and associated with DARs in Plagl1+/–pat retinas. This limited concordance between chromatin accessibility and gene expression has been reported in other systems, supporting the notion that changes in chromatin state do not always translate directly into transcriptional output [77]. To further explore the functional relevance of these DARs, we performed GO term enrichment analysis on the gene-assigned regions. The top terms included “negative regulation of cell cycle,” “mitotic cell cycle phase transition,” and “regulation of translation” (Fig 5D). Thus, both bulk RNA-seq and ATAC-seq analyses converge on the same biological processes, cell cycle regulation and protein translation, as disrupted in Plagl1+/–pat retinas.
To investigate how Plagl1 regulates the biological processes disrupted in mutant retinas, we constructed a GRN by integrating differentially expressed genes (DEGs) from bulk RNA-seq, differentially accessible regions (DARs) from ATAC-seq, and a published Plagl1 ChIP-seq dataset [68]. This analysis revealed Plagl1 as a central hub TF whose loss leads to increased expression of 39 DEGs identified as direct Plagl1 targets (S5 Table; blue nodes, and Fig 5E). Within the GRN, Plagl1 primarily functions as a transcriptional repressor, targeting genes involved in Müller glial activation and late-stage RPC identity, including Nfix, Lhx2, and Vsx2 (Fig 5E). In addition, Plagl1 represses key downstream effectors of multiple signaling pathways, including Notch (Hes1, Hes5), Hedgehog (Gli2, Gli3), Wnt (Tcf4, Tcf7l2), and Yap (Tead1), suggesting broad regulatory influence over gliogenic and proliferative programs. Taken together, these data support a model in which late-stage RPCs in Plagl1+/–pat retinas initiate gliogenic gene expression and begin transitioning toward a Müller glial fate, but fail to properly exit the cell cycle.
Single-cell transcriptomics maps Plagl1 expression to glial and precursor states
Bulk ‘omic’ analyses are ideal for gene discovery but lack the cellular resolution required to detect transcriptional changes within specific retinal cell types. To identify cell type-specific molecular alterations in Plagl1+/-pat retinas, we performed single-cell RNA-sequencing (scRNA-seq) on P7 retinas from eight pups (cs1-cs8), including 5 wild-type and 3 Plagl1+/-pat samples. To reduce ambient RNA contamination, particularly from photoreceptors prone to lysis, we used a dead cell removal kit prior to library preparation and applied CellBender for computational correction. Quality control (QC) thresholds excluded low-quality cells and doublets, and QC metrics remained consistent before and after correction (S5A–S5C Fig).
A total of 8893, 8131, 7610, 17747, 19685, 22614, 23261 and 15615 cells passed QC in cs1-to-cs8, respectively, with an average of 5650 transcript reads per cell across all eight samples. Since biological sex is difficult to determine at P7, we assessed the expression of sex-specific markers, including Xist for female and Eif2s3y for male pups, revealing that all Plagl1+/-pat samples were male, while the wild-type group included two male and three female retinas (S5D and S5E Fig). Transcript coverage analysis confirmed that Plagl1 exon 5 was not transcribed in the three Plagl1+/-pat retinas (S5F Fig), indicating that all Plagl1 transcripts detected in Plagl1+/-pat samples originated from the mutant paternal allele (S5G Fig).
Unsupervised clustering using Seurat and UMAP visualization identified 41 retinal cell clusters (Fig 6A). Canonical markers defined four Müller glial clusters (7, 11, 23, 37) and two precursor populations (16, 20), all expressing Plagl1 together with the glial markers Sox9 and Lhx2 (Fig 6B and 6C). Clusters 16 and 20 also expressed neuronal commitment markers Neurog2 and Gadd45g, consistent with their identity as fate-committed precursors (S6A Fig). The near absence of cell cycle marker (Mki67, Top2a, Ube2c) expression in clusters 16 and 20 further indicated that these cells are postmitotic and undergoing early lineage specification (S7A Fig).
(A) UMAP dimensional reduction plot showing clustering of P7 retinal cells. (B) Dot plot showing expression levels of genes used to assign cluster identities in P7 Plagl1+/-pat vs wild-type retinas. (C) Feature plots showing the distribution of Plagl1, Sox9 and Lhx2 in each cluster in P7 Plagl1+/-pat and wild-type retinas. (D) Mean transcript counts per cell for Plagl1 and Sox9 in each cluster in P7 Plagl1+/-pat vs wild-type retinas. p-values calculated with unpaired t-test for each cluster. Only Plagl1 transcript counts were different between wild-type and mutants in the clusters with p-values shown. (E) Proportion of sequenced cells per cluster in in P7 Plagl1+/-pat vs wild-type retinas. p-values calculated with unpaired t-test for each cluster, but none were significant. (F) Overlay of UMAP plots for P7 Plagl1+/-pat vs wild-type retinas. (G) PCA plot comparing the transcriptomes of all cells in P7 Plagl1+/-pat vs wild-type retinas. (H) PCA plot comparing the transcriptomes of Sox9 + cells only in P7 Plagl1+/-pat vs wild-type retinas. (I) Gene set enrichment analysis showing PANTHER Protein class terms that are enriched in P7 Plagl1+/-pat Müller glial clusters after pseudobulk analysis. (J) Gene set enrichment analysis showing PANTHER Protein class terms that are enriched in P7 Plagl1+/-pat Sox9+ cells after pseudobulk analysis. (K) Summary of log2-fold changes in gene expression among the top DEGs in Sox9 + cells in a pseudobulk analysis of P7 Plagl1+/-pat retinas.
Photoreceptor markers (Neurod1, Nrl, Nr2e3) were enriched in clusters 0–4, 6, and 9, which lacked cone markers (Opn1mw, Opn1sw, Arr3) and were therefore designated as rod photoreceptors (Figs 6B and S6B). These rod markers were also detected in cluster 16, which co-expressed Plagl1, Sox9, and Sox2, consistent with its designation as a committed precursor population (S6B Fig). In contrast, Müller glia and other differentiated cell types showed low photoreceptor gene expression, regardless of CellBender correction (S6B and S6D Fig). However, the broad distribution of Nr2e3 and Nrl transcripts across clusters and genotypes suggested ambient RNA contamination or transcriptional overlap (S6C Fig), limiting interpretation of photoreceptor gene enrichment in Plagl1 mutants. Nonetheless, rod photoreceptor cluster assignments remained robust based on marker abundance (Fig 6A and 6B).
To refine annotation of Plagl1-expressing clusters, we examined mean transcript levels of Plagl1 alongside the glial markers Sox9 and Sox2 across all clusters (Fig 6D and S6 Table). Plagl1 expression was highest in Müller glia, lower in committed precursors and minimal in astrocytes and other differentiated cell types. Since Plagl1 is maternally imprinted, both wild-type and mutant animals express only a single allele, the wild-type or mutant copy, respectively. Nonetheless, mutants exhibited reduced Plagl1 RNA in cluster 7 Müller glia, cluster 16 precursors, and cluster 34 astrocytes, suggestive of a cis-acting defect in transcription and/or RNA stability of the mutant allele in certain cellular contexts (Fig 6D and S7 Table). Sox9 was broadly expressed across glial lineages, whereas Sox2 showed a more restricted pattern, marking Müller glia, committed precursors, astrocytes, and a single amacrine population (S7B Fig). These patterns confirm that Plagl1 is co-expressed with established glial markers, supporting its potential role in regulating glial differentiation.
To assess genotype-dependent changes in retinal cell composition, we quantified the proportion of cells assigned to each cluster (Fig 6E and S8 Table). No statistically significant differences in cell representation were identified, indicating that major retinal cell populations are similarly represented in Plagl1+/–pat retinas (Fig 6E and S9 Table). Accordingly, overlayed UMAPs further showed near-complete superimposition of wild-type and mutant cells (Fig 6F), confirming that transcriptional differences between genotypes are not driven by gross changes in cell-type abundance or clustering structure.
At P7, a small population of proliferating RPCs is expected in the peripheral retina. However, we did not resolve a discrete RPC cluster, likely due to limitations of dimensionality reduction, which can obscure transcriptional gradients and compress transitional progenitor states into adjacent mature cell-type clusters. As a result, RPCs that share gene expression profiles with Müller glia, including Sox9, Lhx2 and Plagl1 (Fig 6C), may have been misclassified. To investigate this possibility, we examined the distribution of canonical cell cycle markers (Mki67, Ube2c, Top2a). A subset of cells within clusters 11 and 23 displayed proliferative signatures in both genotypes (S7A and S7B Fig), suggesting that RPCs are embedded within these Müller glia clusters. At this resolution, however, proliferation marker expression did not differ significantly between genotypes (S7A and S7B Fig). We also detected cell-cycle gene expression in clusters 35 and 36, annotated as endothelial cells, suggesting they are proliferative (Fig 7A and 7B).
(A) Dot plot showing over-representation of Notch signaling genes in Plagl1-expressing Müller glial clusters in P7 wild-type and Plagl1+/-pat retinas. (B) Mean transcript counts per cell for Notch signaling genes in all clusters in P7 wild-type and Plagl1+/-pat retinas. p-values calculated with unpaired t-test for each cluster, but none were significant. (C) RNAscope probe labelling of Hes1, Hes5, and Sox9 on P7 retinal cross sections from wild-type and Plagl1+/-pat containing rosettes and showing Hes1 and Hes5 up-regulation is colocalized with Sox9 staining. Scale bars: 10µm. (D) Sox9 expression in P0 or P21 retinal explants cultured for 5 DIV and then an additional 2 DIV after treatment with DAPT or DMSO. Created in BioRender. Schuurmans, C. (2025) https://BioRender.com/3eaeonm. Scale bars: 25µm. Sox9+ cells were quantified (N = 6 for each treatment). Mean values and error bars representing s.e.m. were plotted. p-values calculated with unpaired t-test. (E) Schematic of iCre and mCherry AAVs injected into the vitreous of ICNflox/flox (floxed STOP) mice. Created in BioRender. Schuurman, C. (2025). https://www.biorender.com/zwr8ryy (F) Sox9 expression in P11 Rosa26+/+ and ICNflox/flox retinas co-injected with AAV2/8-GFAP-iCre and AAV2/8-GFAP-mCherry at P11 and harvested after 10 days (P21). Scale bars: 25µm. (G) Quantification of the ratio of mCherry+Sox9+ cells in the INL versus ONL in P21 Rosa26+/+ and ICNflox/flox retinas at 10 days post-transduction. Mean values and error bars representing s.e.m. were plotted. p-values calculated with unpaired t-test. (H) Ki67 expression in Rosa26+/+ and ICNflox/flox retinas after transduction with mCherry expressing vector carrying iCre. Scale bars: 10µm.
Together, these analyses establish a high-resolution cellular map of P7 wild-type and Plagl1+/–pat retinas, reveal the glial- and precursor-enriched expression of Plagl1, and demonstrate that Plagl1 haploinsufficiency does not grossly alter overall cell-type composition. Although histological analyses clearly show ectopic proliferation in Plagl1⁺/⁻pat retinas, this phenotype is difficult to resolve by scRNA-seq because proliferating RPCs are rare at P7, transcriptionally similar to Müller glia, and readily compressed into adjacent glial clusters during dimensionality reduction.
Cell type-specific transcriptional differences in Plagl1+/-pat Müller glia and Sox9+ cells
Having established that Plagl1 haploinsufficiency does not overtly alter retinal cell-type composition, we next asked whether more subtle, cell type-specific transcriptional changes were detectable. We first performed principal component analysis (PCA) to evaluate genotype-dependent variation across the entire dataset. Partial segregation along PC2 was observed: three of four wild-type replicates clustered separately from the three Plagl1 ⁺ / ⁻ pat samples, while one wild-type sample diverged from both groups (Fig 6G). Because the dataset contains only male mutants but a mixture of male and female wild-type samples, we assessed whether sex or sequencing depth contributed to this pattern. PCA of wild-type samples alone revealed separation of male and female retinas along PC1, indicating that sex contributes to global transcriptional variance, while read-depth differences also accounted for a modest proportion of sample-level variation. However, neither factor aligned with the genotype-associated axis (PC2), demonstrating that sex- and depth-related variance do not explain the transcriptional divergence between wild-type and mutant retinas.
To determine whether genotype-dependent differences were concentrated within specific retinal populations, we repeated PCA using only Sox9 ⁺ cells, comprising Müller glia, fate-committed precursors, and astrocytes. In this subset, a clear segregation between genotypes emerged along PC2 (Fig 6H), indicating that Plagl1 loss-of-function induces a transcriptional shift specifically within glial and precursor lineages. To ensure that this divergence was suitable for differential expression analysis, we assessed data quality and dispersion in pseudobulk data from Müller glia and Sox9 ⁺ cells. Quarter-root mean deviance plots showed expected variability at low expression levels, while empirical Bayes-moderated values closely followed the fitted trend line, indicating stable variance modeling. Similarly, biological coefficient of variation (BCV) plots demonstrated that dispersion estimates were consistent across genes and appropriately trended (S5H, S5I Fig). Together, these metrics support the robustness of the pseudobulk datasets for downstream differential expression analysis.
Pseudobulk differential expression analysis was performed on the four Müller glial clusters and on Sox9 ⁺ cells. This revealed 1,421 differentially expressed genes (DEGs; padj < 0.05) in Müller glia and 165 DEGs in Sox9 ⁺ cells. Gene set enrichment analysis using the Panther classification system revealed consistent enrichment of Protein Class terms related to chromatin-binding and chromatin-regulatory proteins, gene-specific transcriptional regulators, and translational machinery in both Müller glial clusters and Sox9 ⁺ cells (S10 and S11 Tables and Fig 6I and 6J). Within the Sox9 ⁺ population specifically, the enrichment reflected widespread deregulation of genes involved in chromatin regulation, including histone demethylases (Kdm5c, Kdm5d, Kdm6a), nucleosome-binding proteins (Hmgn2, Hist1h1e), and chromatin remodelers (Chd6, Smarcd1, Rcor2), as well as numerous factors linked to RNA processing and translation, such as DEAD-box helicases (Ddx3x, Ddx3y), translation initiation factors (Eif2s3x, Eif2s3y), and multiple ribosomal proteins (Rps28, Rpl38, Rpl22l1, Rps29, Rps27) (S11 Table). Several transcriptional regulators central to glial identity and progenitor competence were also deregulated, including Plagl1, Sox9, and Id1 (Fig 6K and S11 Table).
Together, these findings indicate that Plagl1 haploinsufficiency induces a coordinated reorganization of chromatin-associated, transcriptional and translational programs specifically within Sox9 ⁺ cells, including Müller glia and precursor populations, providing a mechanistic basis for the altered cellular behaviors observed in Plagl1⁺/⁻pat retinas. Notably, these results parallel our earlier bulk RNA-seq analysis of whole retinas, which revealed differential expression of genes associated with “covalent chromatin modification” and “regulation of translation” (Fig 4C). While bulk RNA-seq could not assign these changes to specific cell types, the scRNA-seq pseudobulk analysis confirms that Müller glia and Sox9 ⁺ cells are key contributors to these transcriptional shifts.
Notch signaling is elevated in Plagl1+/-pat Müller glia and is required to sustain a glial identity and sufficient to induce glial delamination
Several pathways were upregulated in conventional Plagl1+/-pat null mutants at P7 that could contribute to the aberrant developmental transition of late RPCs into Müller glia. We first investigated Shh signaling, which induces Müller glial proliferation in chick [78], and can stimulate the proliferation of mammalian Müller glia cultured in vitro [79]. In the scRNA-seq dataset, multiple Shh pathway genes, including Gli3, Smo and Ccnd1, were enriched in Plagl1-expressing Müller glial clusters 7 and 11 in both genotypes (S8A and S8B Fig). RNAscope confirmed Smo expression in INL Müller glia and its upregulation in Plagl1⁺/⁻pat retinas, consistent with bulk RNA-seq (Figs 4G, S8A, and S8B). To test whether elevated Shh signaling alters Müller glial behaviour, we activated the pathway using a conditional Smo (SmoM2) gain-of-function allele carrying an activating W539L point mutation [80]. Following 4-OHT induction between P14-P17, neither Müller glial positioning nor proliferation was affected in P21 Slc1a3-CreERT; SmoM2; Rosa-zsGreen retinas, which resembled Slc1a3-CreERT;Rosa-zsGreen control retinas (S8C Fig). Hyperactive Shh signaling is therefore not sufficient to alter the fate or organization of Sox9+ mature Müller glia (S8D Fig).
Notch signaling, a pro-proliferative pathway in RPCs [69–71] and later required for Müller glial differentiation [69,72-73–], also emerged as a candidate driver of the Plagl1⁺/⁻pat retinal phenotype. Bulk RNA-seq revealed increased expression of Notch pathway components in Plagl1⁺/⁻pat retinas (Fig 4G), prompting closer examination in the single-cell dataset. Dotplot analysis confirmed co-expression of Notch1, Notch2, Hes1, and Hes5 with Plagl1 in Müller glia clusters 7, 11, and 37, with lower expression in cluster 23, in both genotypes (Fig 7A). Although mean transcript levels suggested shifts in Hes1 and Hes5 expression in specific clusters, particularly endothelial cells and cluster 23 Müller glia, these differences were not statistically significant (Figs 7B, S7A, and S7B, and S6 and S7 Tables). However, to directly assess Notch signaling levels in Plagl1⁺/⁻pat Müller glia, RNAscope was performed, which confirmed elevated Hes1 and Hes5 transcripts co-localizing with Sox9+ cells in P7 Plagl1⁺/⁻pat retinas (Fig 7C). Together, these data indicate that Notch signaling is increased in Plagl1⁺/⁻pat retinas at the tissue level, consistent with bulk RNA-seq and RNAscope validation, even though single-cell transcript counts lacked sufficient power to resolve significant changes within individual Müller glial clusters.
To determine the functional role of Notch signaling during the late RPC to Müller glial transition, we first inhibited the pathway in P0 retinal explants using DAPT, a γ-secretase inhibitor. DAPT treatment of P0 explants for 2 DIV nearly abolished Sox9 expression, a Müller glial marker, demonstrating that Notch signaling is required not only to induce but also to maintain Müller glial identity at early postnatal stages (Fig 7D). In contrast, DAPT treatment of P21 retinal explants had no effect on Sox9+ cell number, indicating that Notch signaling is not required to maintain a Müller glial fate in mature retinas (Fig 7D).
We next tested whether hyperactive Notch signaling is sufficient to alter Müller glial behaviour using a conditional Notch intracellular domain (NICD) gain-of-function transgenic line (ICNflox/flox) (Fig 7E). To induce NICD expression in Müller glia, P11 Rosa26+/+ (wild-type) or ICNflox/flox animals were injected with two adeno-associated viruses (AAV2/8) carrying a GFAP promoter to drive iCre and/or mCherry (control) expression (Fig 7E). Notably, while the GFAP mini-promoter can exhibit leaky neuronal expression when paired with neurogenic TFs, ectopic expression is minimal when coupled to Cre recombinase or fluorescent reporters [81,82]. In Rosa26+/+ retinas, 14 days-post transduction, mCherry+ Müller glia formed an INL monolayer (Fig 7I and 7F). Conversely, in ICN flox/flox retinas, mCherry+Sox9+ transduced cells were disorganized in the INL and 20% of these cells translocated into the ONL (Fig 7F). mCherry⁺Sox9 ⁺ cells lacked Ki67 expression in both control and ICNflox/flox retinas, despite Ki67 labeling in the ciliary margin, which served as an internal positive control (Fig 7G). These findings suggest that Sox9 ⁺ cells do not ectopically proliferate following Notch activation.
Thus, elevated Notch signaling partially recapitulates the Plagl1 loss-of-function phenotype, inducing Müller glial translocation without promoting proliferation. These findings support a model in which Notch activation is sufficient to disrupt Müller glial positioning. Although mCherry expression was occasionally detected in the ONL, suggesting rare ectopic ICN activation in neurons, this was infrequent and does not account for the robust and reproducible Müller glial displacement observed across replicates.
Conditional Plagl1 knock-out disrupts Müller glial architecture in the postnatal retina
Since conventional Plagl1+/–pat mutants represent a global knockout, the Müller glial phenotypes observed could reflect secondary effects from earlier developmental disruptions, including ectopic proliferation, or non-cell-autonomous effects. To directly assess the cell-intrinsic role of Plagl1 in postnatal Müller glia, we generated a conditional knockout using the Slc1a3-CreERT driver and a floxed Plagl1 allele in which exons 5 and 6 were flanked by loxP sites (Fig 8 and 8C) [83]. The Slc1a3-CreERT driver line has previously been validated for glial specificity when activated by tamoxifen at P12 [84]. Since Müller glia only ectopically proliferate in conventional Plagl1+/–pat mutants prior to P7, we first tested the specificity of Slc1a3-CreERT;Rosa-zsGreen reporter mice following 4-OHT administration at P5, P7, or P14, with retinas analyzed at P21 (Fig 8A and 8B). zsGreen expression was restricted to Müller glia only when 4-OHT was administered at P14; earlier injections resulted in widespread labeling of cells in the ONL, consistent with Cre activity in late-stage RPCs differentiating into rod photoreceptors (Fig 8B).
(A) Establishment of a conditional transgenic approach to activate gene expression in perinatal/postnatal Müller glia using Slc1a3-CreERT driver mice and a Rosa-zsGreen reporter line. (B) 4-OHT treatment of Slc1a3-CreERT;Rosa-zsGreen mice at P5, P7 and P14, followed by analysis of the distribution of zsGreen+ cells at P21. Schematics created in BioRender. Schuurmans, C. (2025). https://BioRender.com/n6p34rb. Scale bars: 50µm. (C) Targeting strategy to generate a floxed allele of Plagl1 (from [83]), and crosses to Slc1a3-CreERT to generate a Müller glia specific Plagl1-cKO. Schematics created in BioRender. Schuurmans, C. (2025). https://BioRender.com/gjcg2ar (D) Tamoxifen injections were administered at P14 every second day and animals were assessed at P21. Sox9 and zsGreen immunostaining in P21 Slc1a3-CreERT;Rosa-zsGreen control and Slc1a3-CreERT;Plagl1 floxed retinas (i.e., Plagl1-cKOs). Schematics created in BioRender. Schuurmans, C. (2025). https://BioRender.com/n6p34rb. Scale bars: 25µm. (E) Total zsGreen+ (ZsG+) and Sox9+ cells were quantified per field. Mean values and error bars representing s.e.m. were plotted. p-values calculated with unpaired t-test.
To restrict recombination to Müller glia, we administered 4-OHT daily from P14–P17 and analyzed retinas at P21. In control Slc1a3-CreERT;Rosa-zsGreen mice, zsGreen co-localized with Sox9 in a monolayer of Müller glia within the INL (Fig 8D). In contrast, Slc1a3-CreERT;Plagl1+/fl-pat (Plagl1-cKO) retinas showed disruption of the Sox9+ monolayer, indicative of altered nuclear positioning or structural disorganization. As expected, no BrdU-labeled cells were detected in wild-type or Plagl1-cKO retinas [58], consistent with our previous observation that Plagl1+/–pat mutants do not proliferate ectopically after E18.5 explants are cultured for 8 days in vitro [58]. These findings demonstrate that Plagl1 is required cell-autonomously to maintain Müller glial architecture in the postnatal retina. The conditional knockout recapitulates key structural features of the global mutant phenotype, validating our interpretation that Plagl1 directly regulates Müller glial organization and cell cycle exit.
Discussion
Plagl1 is expressed in RPCs during embryogenesis, where it restricts cell number by limiting rod and amacrine cell production [58]. Here, we show that at perinatal stages, Plagl1 expression becomes enriched in Müller glia, consistent with its inclusion in the adult Müller glial transcriptome [85,86]. Using Plagl1+/-pat retinas, we demonstrate that Plagl1 is required for the proper transition from late-stage RPCs to Müller glia. Loss of Plagl1 disrupts this process, leading to ectopic proliferation, nuclear mislocalization, and reactive gliosis of Müller glia, coupled with gross structural disorganization and impaired visual function. Although the visual deficits may partly reflect increased production of rods and amacrine cells in Plagl1 ⁺ / ⁻ pat retinas, Müller glia also influence retinal extracellular resistivity [87], which can alter ERG a-wave timing, while not generating it, hence having no effect on its amplitude. Gliosis can also impair K⁺ siphoning and therefore delay extracellular K⁺ dynamics and influence ON bipolar cell depolarization timing [88]. Finally, Müller cell hyperactivity can also delay b-waves by altering field potentials as well as by modifying synaptic timing at the photoreceptor to ON bipolar cell synapse [23]. Thus, the physiological defects observed in Plagl1 mutants likely arise from a combination of altered neuronal output and disrupted Müller glial homeostasis.
The Plagl1 mutant phenotype reveals a distinct failure in the final stage of retinal gliogenesis. Although Müller glia are specified, as late-stage RPCs undergo gliogenesis, they fail to exit the cell cycle, exhibiting elevated Notch signaling, and displaying abnormal lamination. Notably, the Plagl1 mutant phenotype parallels that of Chd4 cKOs [89], with both models showing persistent RPC proliferation and chromatin remodeling, suggesting that Plagl1 and Chd4 may operate within a shared regulatory axis controlling late-stage gliogenesis. Comparison with other gliogenic mutants highlights the temporal specificity of Plagl1 function. Nfia/b/x triple mutants disrupt an earlier transition, preventing RPCs from activating gliogenic programs and resulting in excess rods and reduced Müller glia and bipolar cells [14]. Lhx2 cKOs reveal an even more proximal block since RPCs are prematurely depleted and fail to acquire gliogenic competence due to rapid downregulation of Notch pathway components (Notch1, Dll1, Dll3) and gliogenic transcription factors (Hes1, Hes5, Sox8, Rax) [22]. Together, these models delineate a temporal and mechanistic hierarchy of retinal lineage termination: Lhx2 establishes gliogenic competence, NFI factors initiate glial fate specification, Chd4 regulates epigenetic transitions, and Plagl1 enforces cell cycle exit and structural maturation.
Chromatin accessibility profiling in Plagl1+/–pat retinas revealed widespread compaction, with differentially accessible regions (DARs) enriched near promoters as well as in intronic and intergenic compartments. These changes suggest distinct regulatory mechanisms, since promoter-proximal DARs are more directly linked to transcriptional initiation and often reflect changes in core regulatory machinery or TF binding. In contrast, DARs located in intronic and intergenic regions may correspond to enhancers, silencers, or architectural elements that modulate gene expression through chromatin looping, splicing regulation, or long-range interactions. The global reduction in accessibility across both proximal and distal elements indicates extensive chromatin remodeling, consistent with a role for Plagl1 in coordinating local and global regulatory programs during retinal maturation. Transcriptomic analyses further revealed widespread dysregulation of genes encoding TFs, chromatin modifiers, translational regulators, and components of Notch and Shh signaling pathways, highlighting Plagl1 as a central regulator of gene expression dynamics in the developing retina. Future studies will be needed to identify direct transcriptional targets of Plagl1, which will help to map Plagl1-dependent regulatory circuits.
Among the affected pathways, Notch signaling emerged as a key Plagl1-responsive axis, correlating with increased expression of cell cycle genes and known roles in promoting RPC proliferation [69–71] and Müller glia-derived progenitor cell formation in regenerative species [90–93]. Functional assays revealed that Notch activation in P11 Müller glia recapitulates nuclear translocation without inducing proliferation, while Notch inhibition impairs early but not late glial fate specification, consistent with its role in restricting neurogenic competence during reprogramming [94]. These findings underscore Plagl1’s influence on chromatin accessibility and transcriptional control of gliogenic and proliferative programs.
Plagl1’s role in enforcing cell cycle exit aligns with its function as a maternally imprinted tumor suppressor gene within the imprinted gene network (IGN), many members of which negatively regulate cell growth [34,49]. Our data establish Plagl1 as a key regulator of Müller glial quiescence in vivo, a function not observed in mouse embryonic fibroblasts [68], suggesting cell type-specificity. The role of Plagl1 in promoting cell cycle exit may be restricted to somatic cells with latent stem cell potential, such as Müller glia, or to late-stage precursor populations. Other IGN genes with similar roles include Dlk1 [46] and Cdkn1c [47], which maintain adult neural stem cell quiescence. Additionally, the role of Plagl1 in promoting cell cycle exit in the perinatal retina aligns with its tumor suppressor role. In humans, PLAGL1 localizes to a chromosome 6q24-25, a region linked to growth inhibition, and is frequently inactivated in multiple cancers [51,53–56]. Accordingly, misexpression of Plagl1 in epithelial cell lines induces cell cycle arrest and apoptosis [50,51].
Prior studies suggest that Plagl1 functions as a temporal identity factor that biases RPCs toward late-born retinal fates, including Müller glia. In mouse E18.5 retinal explants, Plagl1 overexpression promotes bipolar and Müller glial differentiation while suppressing a rod fate [58], and in Xenopus, targeting the two dorsal blastomeres fate-mapped to give rise to the retina, enhances Müller glial specification and inhibits RGC formation by modulating cell cycle exit timing, similarly promoting a late temporal identity [59]. These findings contrast with in vivo electroporation at P0, where Plagl1 fails to induce gliogenesis in wild-type retinas and only partially rescues p27Kip1 expression in Lhx2-deficient RPCs [95]. This discrepancy may reflect differences in RPC competence windows, experimental context, or dosage sensitivity. Together, these data support a model in which Plagl1 modulates, but does not instruct, Müller glial fate, acting within a broader transcriptional network shaped by developmental timing and lineage context.
In conclusion, Plagl1 is a critical regulator of mammalian Müller glia development, one of the few genes that, when mutated initiates both glial translocation and a proliferative response, at least in the early postnatal window. This study reveals that mammalian Müller glia may be prepared to rapidly exit quiescence, with strategies that increase both protein translation and Notch signaling being potentially beneficial to prepare these glial cells to initiate repair. Given that current strategies aimed at inducing Müller glia to regenerate require a stimulating injury to activate these glial cells [96], the identification of a gene that can instead be precisely targeted presents a new path forward for glial regeneration.
Materials and methods
Ethics statement
All animal experiments were approved by the University of Calgary Animal Care Committee (ACC) (AC11–0053) and later by the Sunnybrook Research Institute ACC (AUP 16–606) in compliance with the Guidelines of the Canadian Council of Animal Care and conformed to the ARVO statement for the Use of Animals in Ophthalmic and Vision Research.
Materials availability
Plasmids generated in this study will be available upon request. Transgenic mouse lines generated in this study are not deposited in a central repository. As we are limited in the number of stock animals that we can maintain, the transfer of animals is possible only with reasonable compensation for processing and shipping and a completed Materials Transfer Agreement if there is a potential for commercial application.
Experimental models and subject details
None of our experimental animals had been previously used for other procedures. All animals were healthy.
Mice.
Animal sources and maintenance. Most transgenic lines were purchased from Jackson Laboratory, ME, USA unless indicated. Plagl1+/-pat animals were considered null mutants due to maternal imprinting of the wild-type allele, and were generated by crossing Plagl1+/-mat males (received from Laurent Journot) [49] with C57/Bl6 wild-type females (JAX#000664). The mutant allele corresponds to the targeted deletion described in Varrault et al. (2006), which removes the translation start site located in exon 5 and the entire zinc-finger domain encoded in exon 6, extending to the 3′ Ava I site located 601 bp into the 1963 bp coding sequence of exon 6. This eliminates approximately 30.6% of the coding sequence, including all C2H2 zinc-finger motifs required for DNA binding and transcriptional activity (i.e., functionally null). Plagl1fl/+ males (received from Pengpeng Bi) [83] were crossed with Slc1a3-CreERT females (JAX:012586) to generate Plagl1 cKOs. To activate Slc1a3-CreERT, 100 μg of 4-hydroxy-tamoxifen (4-OHT) was administered to P14 pups for 3 consecutive days and retinas were collected 7 days after the last injection. A Rosa-ZsGreen reporter (Jackson Lab: 007906) was used for lineage tracing. B6;129-Gt(ROSA)26Sortm(Notch1)Dam mice (JAX:008159) [97] were backcrossed to C57BL/6J for >15 generations. The resulting B6.ICN1fl/+ strain was intercrossed to generate B6.ICN1fl/fl mice, abbreviated ICN1fl/fl.Gt(ROSA)26Sortm1(Smo/EYFP)Amc/J (JAX:005130) males were crossed with Slc1a3-CreERT females as described above. CD1 mice (#022; Charles River Laboratories, Senneville, QC, Canada) were used for MNU experiments. Male and female pups were pooled for all experiments as very few animals were double and triple transgenics.
Method details
Electroretinogram (ERG).
Animals were dark-adapted for 1 hr prior to ERG recordings. ERGs were recorded under dim red light. Stimulation and acquisition were performed with an Espion E2 system (Diagnosys LLC) (flash duration 10 µs, bandpass filtering 0.3Hz-300Hz), as described [98,99]. Briefly, scotopic intensity responses were measured using increasing light flash pulses of increasing incremental strengths (-5.22 to 2.86 log cds/m2), followed by photopic intensity responses (30 cd/m2 background light) with flash strengths spanning -1.6 to 2.9 log cds/m2.
Optical coherence tomography (OCT).
Mice were anesthetized with 2% isoflurane. Pupils were dilated using tropicamide topical drops (Mydriacyl 1%; Alcon Canada Inc). Eyes were imaged using OCT performed with a confocal scanning laser ophthalmoscope (cSLO) mounted with a 25-diopter lens (Heidelberg Engineering, Germany). Normal saline was applied to the cornea every 2 minutes to keep it hydrated. OCT was performed on a 30° × 20° area, acquiring a total of 31 scans at an average of 20 frames per scan. Infrared cSLO scans (IR-cSLO) were also acquired (λ = 815 nm) with an average of 30 frames per scan to assess the retinal fundus. Retinal scans were exported from the HEYEX software (Heidelberg Engineering, Germany) as tiff images and processed using ImageJ software (NIH, USA).
Tissue processing.
Retinal tissues were drop fixed in 4% paraformaldehyde (PFA)/1X phosphate buffered saline (PBS) at 4°C overnight for dissected retinas or eyes or for 3 hr for retinal explants. Tissues were then rinsed 3 x 10 min in 1X PBS, cryopreserved in 20% sucrose/1X PBS overnight at 4°C, and embedded and frozen in O.C.T (Tissue-Tek, Sakura Finetek U.S.A. Inc., Torrance, CA). Cryosections were cut at 10μm on a Leica CM3050s cryostat (Leica Biosystems, Buffalo Grove, IL, USA) and collected on Fisherbrand Superfrost Plus slides (Thermo Fisher Scientific, Markham, ON).
Immunohistochemistry.
Cryosections were blocked for 1 hr at room temperature in blocking solution: 10% normal goat serum in PBST (1X PBS/0.1% Triton X-100). Primary antibodies were diluted in blocking solution and incubated at 4ºC overnight. Primary antibodies included: rabbit anti-Arr3 (1/500, Millipore #AB15282), rat anti-BrdU (1/500, Serotec #OBT0030S), mouse anti-Brn3a (1/500, Chemicon #5945), mouse anti-calbindin (1/1000, Sigma #9848), rabbit anti-calretinin (1/2000, Swant #76699/4), rabbit anti-Caspase3 (1/500, Abcam #ab2302), rat anti-CD11b (M1/70, 1/500, Abcam #ab8878), rabbit anti-GFAP (1/500, Sigma #G9269), rabbit anti-Glul (1/500, Abcam #73593), rabbit anti-Iba1 (1/500, Wako Chemicals USA #019–19741), rat anti-mKi67 (1/250, Invitrogen 14-5698-82), rabbit anti-M-opsin (1/250, Millipore #5405), rabbit anti-Pax6 (1/500, Convance #PRB-278P), rabbit anti-pERK (Map kinase p44/42 phospho-Thr202/Tyr204) (1/500, Cell signaling #4370), mouse anti-rhodopsin (1/500, Chemicon #MAB5356), mouse anti-Rlbp1 (1/500, Abcam #15051), rabbit anti-S-opsin (1/250, Millipore #5407), rabbit anti-Sox9 (1/500, Millipore #AB5535), mouse anti-syntaxin (1/500, Sigma #S 0664), mouse anti-vimentin (1/500, Sigma #V5255), and mouse anti-ZO1-1A12 (1/100, ThermoFisher Scientific #33–9100). Slides were washed 3 x 10 min in PBST and then incubated in secondary antibodies conjugated to Alexa-568, Alexa-488 or Alexa-647 diluted at 1/500 in PBST for 1 hr at room-temperature, followed by 3 x 10 min washes in PBST. Sections were coverslipped in Aqua-Poly/Mount (Polysciences #18606).
Isolectin staining.
Eyes were dissected and retinas were flatmounted or processed for sectioning as described. Wholemount retinas were fixed in 4% PFA/1X PBS for 2 hours. Flatmounted retinas or cryosections were fixed in ice cold 70% ethanol for 30 min. Fixed tissue was then incubated in Isolectin B4 (1/250, Sigma #L2140) in PBST overnight at 4°C, washed 3 X 10 min in PBST. Wholemount retinas were mounted on microscope slides and mounted tissues or sections were coverslipped in Aqua-Poly/Mount.
BrdU labeling.
To label S-phase cells, animals were injected intraperitoneally with 100 g/kg body weight BrdU (Sigma, Oakville ON) either 30 min before sacrifice for proliferation assays or at P7 for birthdating assays. Tissues were processed for anti-BrdU staining as described above except for the addition of a pre-treatment step with 2N HCl for 15 min at 37°C prior to blocking.
RNAscope.
Double fluorescent in situ hybridizations were performed using a RNAscope Multiplex Fluorescent Detection Kit v2 (ACD; #323110) according to the manufacturer’s directions. ACD probes used included: Mm-Sox9 (C2: 401051-C2), Mm-Hes1 (#417701), Mm-Hes5 (400991 -C2), Mm-Plagl1 (C1: 462941), Smo (C2:318411), and Ccnd1 (C2:442671). Opal 570 (Akoya; #FP1488001KT; 1:1500) was applied for channel 1 and Opal 520 (Akoya; #FP1487001KT; 1:1500) was applied for channel 2. Retinal sections were counterstained with DAPI and mounted in Aquapolymount as described.
Intravitreal injection.
Mice were anesthetized with 2% isoflurane and their body temperature was kept at 37°C using a heated blanket, with saline applied to keep the eye moist. A 30G needle was used to puncture the sclera at the margin and cornea level, then a Hamilton syringe was inserted through the sclera into the vitreous body. We then injected into the vitreous 1.2 μl containing 3.5x10^13 GC of pAAV2/8-GFAP-iCre and 3.5x10^13 GC of pAAV2/8-GFAP-mCherry [100,101], diluted in 0.1 Fast Green dye.
4-OHT administration.
4-OHT was first dissolved in Ethanol at 20 mg/ml, sonicated for 15 min, and diluted to 10mg/ml in peanut oil. Excess ethanol was evaporated for 30 min using a speed vacuum concentrator. 20mg/kg of 4-OHT was administered intra-peritoneally for 3 consecutive days using a 1ml syringe and 23G needle.
Retinal explants.
Retinas were dissected and the RPE was removed from P0 eyes in ice-cold PBS. Retinas were flattened on 0.25 μm Nucleopore membranes (Whatman # 110409) in 6 well plates and cultured for 5 days in vitro (DIV) at 37°C, 5% CO2 in retinal explant medium. (50% Dulbecco’s Modified Eagle Medium (DMEM, Wisent; #319–005-CL), 25% Hanks’ Balanced Salt Solution (HBSS, Gibco; #24020–117), 25% heat inactivated horse serum (ref), 200 μM L-Glutamine (Wisent; #609–065-EL), 0.6 mM HEPES (Wisent; # 330–050-EL), 1% Pen/Strep (Wisent; # 450–201-EL). DAPT (565770, Calbiochem Milipore Sigma) or equivalent volume of DMSO were added to the retinal explant media at 10μM final concentration for 2 days. Explants were harvested after 5 DIV.
Western blotting.
Retinas were collected from adult and postnatal pups at the indicated stages, lysed in RIPA buffer with protease (1x protease inhibitor complete, 1 mM PMSF) and phosphatase (50 mM NaF, 1 mM NaOV) inhibitors, and 10 µg of lysate was run on 10% SDS-PAGE gels for Western blot analysis as described previously (Ma et al., 2007). Primary antibodies included pERK (Map kinase p44/42 phospho-Thr202/Tyr204; Cell Signaling #4370), ERK (Map kinase p44/42; Cell Signaling #9102), Glul (Abcam, #ab73593, 1/10.000) and Actin (Abcam; #ab8227, 1/10.000). Densitometries were calculated using ImageJ. The average values of normalized expression levels were plotted from N = 6 per genotype.
RT2 real-time quantitative reverse transcription-PCR.
Total RNA was extracted from dissected retinas using Trizol RNA Isolation Reagent (Thermo Fisher Scientific; #15596–026), following the manufacturer’s instructions. For complimentary DNA (cDNA) synthesis, 0.5 µg of total RNA was converted using an RT2 first strand kit (Qiagen, #330401) following the manufacturer’s instructions. qPCR reactions were performed using a CFX384 cycler (Biorad Laboratories, Canada) using an RT2 SYBR Green PCR Master Mix (Qiagen; #330500) following the manufacturer’s instructions. The following RT2 qPCR primers were used: Sox9 (PPM05134D), Lhx2 (PPM31533A), Hes1 (PPM5647A), Hes5 (PPM31391A), Plagl1 (PPM37555A), Gfap (PPM04716A), Recoverin (PPM31906A), Rpl13 (PPM40286A), Rpl26 (PPM24941A) and three reference genes for normalization: Gapdh (PPM02946E), B2m (PPM03562A), and Hrpt (PPM03559F). The ΔΔCt method was used to determine relative gene expression using the Biorad CFX manager software.
RNA extraction and bulk RNA-seq.
Total RNA was extracted from dissected retinas using Trizol RNA Isolation Reagent (Thermo Fisher Scientific; #15596–026), following the manufacturer’s instructions except the overnight incubation step was reduced to 30 min at -80oC. RNA quality was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). For RNA-Seq 8 Plagl1+/-pat and 4 wild-type littermates were analyzed. 500 ng RNA per sample was sequenced on Illumina NextSeq500 Platform. 75 base single-end sequence reads were generated on NEBNext Ultra II Directional RNA Library Prep Kit for Illumina. Basecalling and demultiplexing were done using IIllumina CASAVA 1.9 pipeline. Reads were mapped to mouse genome (Ensembl, GRCm38) using hisat2 version 2.1.0 with a mapping rate close to or over 98% in each sample.
Bulk RNA-seq data analyses.
DEGs were detected using DESeq2 1.24.0 Bioconductor package as described in the package vignette [102]. Genes with adjusted p-values (Benjamini-Hochberg adjustment for multiple comparisons) [103] less than 0.05 (5% chance of gene being a false positive) and over 1.5-fold change in either direction were selected as DEGs. Over-represented GO terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) [104] pathways were detected using clusterProfiler 3.12.0 Bioconductor package [105]. Cell cycle-related genes were selected from the Gene Ontology (GO) category “GO0007049: cell cycle” and from a functional classification of additional cell cycle genes based on GO categories and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways using DAVID.
ATAC-seq.
ATAC-seq was performed using 75,000 cells as input material for the library preparation with Active motif kit (Catalog No. 53150) following the manufacturer’s instructions. 75 bp paired-end sequencing was performed on Illumina NEXTSEQ500. Reads were mapped to the reference genome (Ensembl, GRCm38) using Bowtie2 short read aligner version 2.3.5.1. Areas of open chromatin were predicted using Macs2 v.2.2.7.1 based on alignment files filtered of mitochondrial reads. Differential enrichment was detected using DiffBind Bioconductor package. Data was normalized using Trimmed Mean of Medians (TMM) normalization native to EdgeR Bioconductor package. In total, differential binding tests were applied to 56771 peaks. The peak had to be present in at least 2 samples within the experimental group to be included in the analysis. False discovery rate (FDR) below 0.05 was a threshold to consider the peak differentially bound. No peaks were below 0.05 FDR threshold. DiffBind results were annotated with nearest promoter, gene symbol, entrez_id and gene description using ChIPpeakAnno Bioconductor package. Term enrichment analysis was conducted using 2 different approaches: 1) over-representation analysis (ORA) conducted using geneSCF software and 2) Gene set enrichment analysis (GSEA). Genes with TSS located within 5000 bp of DARs were used in ORA. The input for GSEA was the list of all genes within 5000 bp from the nearest TSS sorted by the Fold Change. GSEA enrichment analysis was conducted using ClusterProfiler v.3.18.1 Bioconductor package [106].
Gene regulatory network (GRN) inference.
ATAC-seq and RNA-seq were integrated by overlapping the lists of significantly changed genes and differentially accessible regions (DARs). Genes or DARS with adjusted p-values below 0.05 were considered significant. The list of DARs was limited to 311 peaks located within 5000 bp from the nearest transcriptional start site (TSS) of gene assigned DARs. The GRN was inferred following our published approach [107]. Prior knowledge networks (PKN) among differentially expressed genes (DEGs) were assembled from MetaCore (GeneGo Inc. [108]), TRRUST [109], RegNetwork [110] and Plagl1 ChIP-seq data [68]. The Pearson correlation coefficient for each pair of TFs and gene was computed over the entire scRNA-seq dataset. The intersection between PKN and Pearson correlation coefficient above 90 percentile was maintained. Then, interactions were further removed if the chromatin regions of target genes were not open in all wild-type ATAC-seq replicates. Cytoscape v.3 was used to visualize GRNs [111], in which differentially up-regulated and down-regulated genes in the wild-type bulk RNA-seq data with respect to the Plagl1+/-pat bulk RNA-seq data were indicated as red and blue nodes, respectively.
scRNA-seq method.
P7 wild-type (N = 5) and Plagl1+/-pat (N = 3) retinas were dissected in cold PBS (311–010-CL, Wisent), enucleated, and cut into 2 mm pieces. Retinal tissues were dissociated into single cell suspensions using a Papain dissociation system (LK003150, Worthington Biochemical Corp) according to the manufacturer’s instructions and then resuspended in PBS with 0.04% BSA. Dead cells were eliminated using a Dead Cell Removal Kit (130-090-101, Miltenyi Biotec) with MS columns and a gentleMACS Octo Dissociator according to the manufacturer’s instructions. Cell number and viability were determined using a Countess II FL Automated Cell Counter (ThermoFisher Scientific). Cells were diluted in 1X PBS with 0.04% bovine serum albumin (BSA) to 1X 10^6 cells/ml and cellular concentration and viability were accessed again, with only samples with 90% viability and higher used. Live cells were processed using the Chromium Next GEM Single Cell 5’ Reagent Kit v2 (10X genomics, Cat # 1000263). For each reaction, 16500 cells were loaded onto GEM Chip K for an expected recovery of 10,000 cells. Gel Beads-in-emulsion were generated using the Chromium Controller followed by cDNA generation and amplification (13 cycles) as per manufacturer’s instructions. For each sample, 50 ng of cDNA was used for library generation. Equal molar of each library for all samples was pooled and sequenced at an expected depth of 44,000 reads/cell using the Illumina NovaSeq S4 flow cell system (The Centre for Applied Genomics, Hospital for Sick Children).
scRNA-seq data analysis.
scRNA-seq analysis of wild-type (N = 5) and Plagl1+/-pat (N = 3) retinal samples was performed using the Seurat v.4.0.1 R package [112]. Cells that were of low quality or represented doublets were excluded by filtering out cells with greater than 5000 and fewer than 1000 detected genes, cells with greater than 30000 RNA counts, or cells with mitochondrial RNA percent larger than 10. For each library, comparisons were made to denoised data of raw 10x matrices, which were denoised with CellBender to remove ambient RNA and empty droplets while preserving true cell barcodes [113]. The data was then transformed by the SCTransform function while regressing out the variance due to mitochondrial RNAs and cell cycle genes. Clustering was performed by the RunPCA, FindNeighbors and FindClusters functions using the first 30 principal components. The 2D projection of the clustering was carried out by the RunUMAP function. The annotation of cell type to each cluster was performed by using the marker set described in Fig 6B. The RNA reads mapped to the first coding-domain-containing exon of the Plagl1 gene (ENSMUSE00000666670) were visualized using the Integrative Genomics Viewer (IGV) [114]. Differential gene expression analysis between wild-type and Plagl1+/-pat for each cluster was carried out using the FindMarkers function with min.pct being 0.1 and the adjusted p-value cutoff of 0.05. The mean transcript count was defined as the ratio of total UMI counts to the number of detected genes in each cell. Gene ontology enrichment analysis was performed with an open-source online tool “Powered by Panther” (Gene Ontology Resource).
Pseudobulk gene expression analysis.
For each Müller glia cluster separately, or Sox9 + cells as a whole, raw counts were aggregated per donor by summing UMI counts across cells from the same donor within the subset, Genes with CPM > 1 in ≥50% of samples (within the subset) were kept. Libraries were TMM-normalized, and a design matrix with condition (WT vs KO) was fit using the quasi-likelihood (QL) GLM pipeline of edgeR QC (19910308): estimateDisp(…, robust = TRUE), glmQLFit, and glmQLFTest for the WT vs KO contrast. P-values were Benjamini–Hochberg adjusted and genes with padj < 0.05 were defined as differentially expressed genes (DEGs).
Principal component analysis (PCA).
Two complementary PCA analyses were performed; 1). PCA was run on SCTransform residuals using the variable features (top 3,000) across all retained cells (WT and KO combined). 2). For Müller glia and Sox9 ⁺ subsets separately, we used the pseudobulk matrices described above. The top 2,000 most variable genes across samples within each subset were selected, mean-centered and scaled features, and ran PCA to visualize between-sample structure.
Image acquisition and processing.
Images were captured on a Leica DMRXA2 optical microscope (Leica Microsystems Canada Inc., Richmond Hill, Ontario, Canada) using LasX software. Confocal images were acquired using a Leica SP8 spectral confocal microscope (Leica Microsystems CMS GmbH). The images were prepared using Adobe Photoshop CC 2017 (Adobe Systems Inc., San Jose, CA, USA). Schematics were generated with a license from BioRender. For Plagl1+/-pat retinal transverse sections, all areas and sections containing rosettes and/or ectopia were imaged, the equivalent areas/ sections number were imaged for the wild-type littermate and used as controls.
Statistical analysis.
Quantification was performed on a minimum of three biological replicates (N = 3) for each, and a minimum of three sections from each eye. N numbers (biological replicates), statistical tests and p values for each count are indicated in the figures and figure legends. Graphs and statistics were generated using GraphPad Prism Software 8 (GraphPad Inc., La Jolla, CA). All data expressed as mean value ± standard error of the mean (s.e.m.). Asterisks show level of statistical significance, with actual p-values in the figures.
Supporting information
S1 Fig. Plagl1 expression is down-regulated in Müller glia following injury.
Related to Fig 1. (A,B) RNAscope staining for Plagl1, Sox9, and immunofluorescence for GFAP, Rho and Arrestin3 in P7 wild-type retinal section of mice treated with PBS and MNU at 7- (A) and 21- (B) days post injection. (C,D) Quantification of Plagl1, GFAP and Recoverin expression by qRT-PCR in wild-type retinas treated with PBS and MNU and collected after 7-days (C) and 21-days (D) post injection. N = 3 for all time points. p-values calculated with unpaired t-test. GCL, ganglion cell layer; INL inner nuclear layer; ONL, outer nuclear layer. Scale bars: 50 µm.
https://doi.org/10.1371/journal.pgen.1012020.s001
(TIF)
S2 Fig. Plagl1 is required to sustain retinal integrity.
Related to Fig 3. (A) Immunostaining for cleaved caspase 3 (cc3) in P5 and P21 wild-type retinas, and cc3 and BrdU co-immunostaining in P7 wild-type and Plagl1+/-pat retinas. Scale bars: 25µm. (B) DAPI nuclear staining of P7, P14, and P60 wild-type and Plagl1+/-pat retina sections, showing retinal ectopias (asterisks) and rosettes (arrowheads). Right side panels are high magnifications of boxed areas. (C) Immunofluorescence for retinal cell specific markers S-opsin (cones), Pax6 (amacrine and ganglion cells), Calretinin (amacrine cells), Calbindin (horizontal cells), Brn3a (ganglion cells), M-opsin (cones), Syntaxin (amacrine cells), PKC and Vsx2 (bipolar cells), and Rlbp1 (Müller glia) with DAPI nuclear counter-staining of P21 wild-type and Plagl1+/-pat retinal sections, showing ectopia (asterisks) and rosettes (arrowheads) contain all retinal cell types. At least 3 Plagl1+/-pat mice of 2 different litters and their wild-type littermates were analyzed. gcl, ganglion cell layer; inl, inner nuclear layer; onl, outer nuclear layer. Scale bars: 50µm.
https://doi.org/10.1371/journal.pgen.1012020.s002
(TIF)
S3 Fig. Analyses of DEGs and translation in P7 Plagl1+m/-p retinas.
Related to Fig 4. (A,B) GO analysis of molecular function and cellular compartment associated with P7 Plagl1+m/-p retinal DEGs. (C,D) Mean FPKM values for selected chromatin and transcriptional regulators (C) and genes related to protein translation (D) that were differentially expressed in P7 Plagl1+/-pat retinas were plotted with s.e.m. padj values were computed using DESeq2. (E) qPCR showing upregulated Sox9, Lhx2, and Hes1 transcripts and downregulated Rpl13 and Rpl26 transcripts in P7 Plagl1+/-pat retinas. p-values calculated with unpaired t-test.
https://doi.org/10.1371/journal.pgen.1012020.s003
(TIF)
S4 Fig. Analyses of cell cycle-related genes amongst the DEGs in P7 Plagl1+m/-p retinas.
Related to Fig 4. (A) Heatmap representing cell cycle genes differentially expressed in wild-type vs Plagl1+/-pat retinas. (B) FPKM values for selected cell cycle regulators. p-values calculated with unpaired t-test.
https://doi.org/10.1371/journal.pgen.1012020.s004
(TIF)
S5 Fig. Quality control tests for scRNA-seq dataset.
Related to Fig 6. (A-C) Quality control tests for scRNA-seq data collected from 8 retinal samples, showing nFeature_RNA, or the number of genes detected in each sample (A), nCount_RNA, or the total number of unique molecular identifiers (UMIs) detected within a cell (B) and percental of mitochondrial gene expression (C). Data shown in plots A–C were analyzed after CellBender-based ambient RNA correction. (D,E) Genotyping for biological sex by examining average gene reads for Xist (female, D) and Eif2s3y (male, E). (F) Integrative Genomics Viewer (IGV) analysis of read counts for introns/exons in the Plagl1 genomic locus from P7 Plagl1+/-pat vs wild-type retinal samples. (G). Average Plagl1 gene reads per cell in each sample, and averaged for wild-type and Plagl1+/-pat retinas. (H) Quarter-root mean deviance and biological coefficient of variation (BCV) plots for Müller glia pseudobulk data. Deviance plots show expected variability at low expression levels, and BCV plots confirm consistent dispersion estimates across genes. (I) Equivalent deviance and BCV plots for M Sox9 ⁺ cell pseudobulk data. These metrics support robust variance modeling and validate the suitability of both datasets for differential expression analysis using edgeR
https://doi.org/10.1371/journal.pgen.1012020.s005
(TIF)
S6 Fig. Precursor and photoreceptor gene enrichment in single cell transcriptomic data.
Related to Fig 6. (A,B) Feature plots of neuronal precursor (Neurog2, Gadd45g) (A) and rod photoreceptor (Neurod1, Nrl, Nr2e3) (B) markers in the scRNA-seq dataset. (C) Mean transcript counts/cell for Nr2e3 and Nrl in wild-type and Plagl1+/-pat retinas in each cluster. p-values calculated with unpaired t-test for each cluster, but none were significant. (D) Feature plots of rod photoreceptor (Neurod1, Nrl, Nr2e3) markers in the scRNA-seq dataset after CellBender correction.
https://doi.org/10.1371/journal.pgen.1012020.s006
(TIF)
S7 Fig. Cell cycle gene enrichment in single cell transcriptomic data.
Related to Fig 6. (A) Feature plots of cell cycle markers (Mki67, Top2a, Ube2c) and rod photoreceptor in the scRNA-seq dataset. (B) Mean transcript counts/cell for Mki67, Top2a, and Ube2c in wild-type and Plagl1+/-pat retinas in each cluster. p-values calculated with unpaired t-test for each cluster, but none were significant.
https://doi.org/10.1371/journal.pgen.1012020.s007
(TIF)
S8 Fig. Shh pathway analysis.
Related to Fig 7. (A) Dot plot showing over-representation of Shh signaling genes in Müller glial clusters in P7 wild-type and Plagl1+/-pat retinas. (B) RNAscope probe labelling of Plagl1 and Smo or Plagl1 and Ccnd1 on P7 retinal cross sections from wild-type and Plagl1+ /-pat retinas, showing Smo up-regulation. (C) Targeting strategy to generate a conditional gain-of-function of constitutively active Smo-M2 [80]. Created in BioRender. Schuurmans, C. (2025) https://BioRender.com/fdq40gz. (D) Sox9 and zsGreen expression in Müller glia in control and Slc1a3-CreER;SmoM2 floxed retinas after 4-OHT injections from P14. gcl, ganglion cell layer; inl, inner nuclear layer; onl, outer nuclear layer. Scale bars: 25µm.
https://doi.org/10.1371/journal.pgen.1012020.s008
(TIF)
S1 Table. Related to Fig 4. FPKM values for all gene expression in P7 wild-type and Plagl1+/-pat retinas from bulk RNA-seq.
https://doi.org/10.1371/journal.pgen.1012020.s009
(XLSX)
S2 Table. Related to Fig 4. List of differentially expressed genes (DEGs) in P7 wild-type and Plagl1+/-pat retinas from bulk RNA-seq.
https://doi.org/10.1371/journal.pgen.1012020.s010
(XLSX)
S3 Table. Related to Fig 4. Proliferation-related gene expression in P7 wild-type and Plagl1+/-pat retinas from bulk RNA-seq.
Comparisons were made to a list of 616 genes involved in cell cycle regulation, including genes in the GO category “GO0007049: cell cycle” and a functional classification of additional cell cycle genes based on GO categories and KEGG pathways using DAVID.
https://doi.org/10.1371/journal.pgen.1012020.s011
(XLSX)
S4 Table. Related to Fig 5.
Overlap between up-regulated and down-regulated DEGs and DARs (from ATAC-seq data) with decreased accessibility in P7 Plagl1+/-pat retinas.
https://doi.org/10.1371/journal.pgen.1012020.s012
(XLSX)
S5 Table. Related to Fig 5. Gene regulatory network analysis identifying predicted upstream regulators of Plagl1 and Plagl1 target genes.
https://doi.org/10.1371/journal.pgen.1012020.s013
(XLSX)
S6 Table. Related to Fig 6.
Mean transcript counts per cluster, showing a sum of all clusters. The mean transcript count for select genes was defined as the ratio of total UMI counts to the number of detected genes in each cell.
https://doi.org/10.1371/journal.pgen.1012020.s014
(XLSX)
S7 Table. Related to Fig 6. Mean transcript counts per cluster, showing individual values for each sequenced sample.
The mean transcript count for select genes was defined as the ratio of total UMI counts to the number of detected genes in each cell. Individual p-values comparing wild-type and mutant samples calculated with an unpaired t-test are shown.
https://doi.org/10.1371/journal.pgen.1012020.s015
(XLSX)
S8 Table. Related to Fig 6. Mean cell counts per cluster, showing a sum of all clusters.
https://doi.org/10.1371/journal.pgen.1012020.s016
(XLSX)
S9 Table. Related to Fig 6.
Mean cell counts per cluster, showing individual values for each sequenced sample. Individual p-values comparing wild-type and mutant samples calculated with an unpaired t-test are shown.
https://doi.org/10.1371/journal.pgen.1012020.s017
(XLSX)
S10 Table. Related to Fig 6.
DEGs and Panther GO analysis of pseudobulk Müller glia data. Clusters 7, 11, 23 and 37 were combined.
https://doi.org/10.1371/journal.pgen.1012020.s018
(XLSX)
S11 Table. Related to Fig 6.
DEGs and Panther GO analysis of pseudobulk Sox9+ cells. All cells expressing Sox9, regardless of cluster number, were combined.
https://doi.org/10.1371/journal.pgen.1012020.s019
(XLSX)
S12 Table. Numerical data for all graphs and summary statistics.
https://doi.org/10.1371/journal.pgen.1012020.s020
(XLSX)
Acknowledgments
We thank Valerie Wallace for critical comments on an earlier version of this manuscript. We thank Natalia Klenin for technical assistance. The authors acknowledge the support of the Sunnybrook Research Institute (SRI) Genomics Core Facility, the SRI Histology Core Facility (Petia Stefanova), and the Sunnybrook Centre for Cytometry and Scanning Microscopy (Kevin Conway).
References
- 1. Wetts R, Fraser SE. Multipotent precursors can give rise to all major cell types of the frog retina. Science. 1988;239(4844):1142–5. pmid:2449732
- 2. Holt CE, Bertsch TW, Ellis HM, Harris WA. Cellular determination in the Xenopus retina is independent of lineage and birth date. Neuron. 1988;1(1):15–26. pmid:3272153
- 3. Jensen AM, Raff MC. Continuous observation of multipotential retinal progenitor cells in clonal density culture. Dev Biol. 1997;188(2):267–79. pmid:9268574
- 4. Turner DL, Cepko CL. A common progenitor for neurons and glia persists in rat retina late in development. Nature. 1987;328(6126):131–6. pmid:3600789
- 5. Turner DL, Snyder EY, Cepko CL. Lineage-independent determination of cell type in the embryonic mouse retina. Neuron. 1990;4(6):833–45. pmid:2163263
- 6. Young RW. Cell differentiation in the retina of the mouse. Anat Rec. 1985;212(2):199–205. pmid:3842042
- 7. Stiemke MM, Hollyfield JG. Cell birthdays in Xenopus laevis retina. Differentiation. 1995;58(3):189–93. pmid:7713326
- 8. Cepko CL, Austin CP, Yang X, Alexiades M, Ezzeddine D. Cell fate determination in the vertebrate retina. Proc Natl Acad Sci U S A. 1996;93(2):589–95. pmid:8570600
- 9. Alexiades MR, Cepko CL. Subsets of retinal progenitors display temporally regulated and distinct biases in the fates of their progeny. Development. 1997;124(6):1119–31. pmid:9102299
- 10. Elliott J, Jolicoeur C, Ramamurthy V, Cayouette M. Ikaros confers early temporal competence to mouse retinal progenitor cells. Neuron. 2008;60(1):26–39. pmid:18940586
- 11. Mattar P, Ericson J, Blackshaw S, Cayouette M. A conserved regulatory logic controls temporal identity in mouse neural progenitors. Neuron. 2015;85(3):497–504. pmid:25654255
- 12. Javed A, Mattar P, Lu S, Kruczek K, Kloc M, Gonzalez-Cordero A, et al. Pou2f1 and Pou2f2 cooperate to control the timing of cone photoreceptor production in the developing mouse retina. Development. 2020;147(18):dev188730. pmid:32878923
- 13. Javed A, Santos-França PL, Mattar P, Cui A, Kassem F, Cayouette M. Ikaros family proteins redundantly regulate temporal patterning in the developing mouse retina. Development. 2023;150(2):dev200436. pmid:36537580
- 14. Clark BS, Stein-O’Brien GL, Shiau F, Cannon GH, Davis-Marcisak E, Sherman T, et al. Single-Cell RNA-Seq Analysis of Retinal Development Identifies NFI Factors as Regulating Mitotic Exit and Late-Born Cell Specification. Neuron. 2019;102(6):1111-1126.e5. pmid:31128945
- 15. Baye LM, Link BA. Nuclear migration during retinal development. Brain Res. 2008;1192:29–36. pmid:17560964
- 16. Lyu P, Hoang T, Santiago CP, Thomas ED, Timms AE, Appel H, et al. Gene regulatory networks controlling temporal patterning, neurogenesis, and cell-fate specification in mammalian retina. Cell Rep. 2021;37(7):109994. pmid:34788628
- 17. Shiau F, Ruzycki PA, Clark BS. A single-cell guide to retinal development: Cell fate decisions of multipotent retinal progenitors in scRNA-seq. Dev Biol. 2021;478:41–58. pmid:34146533
- 18. Gomes FLAF, Zhang G, Carbonell F, Correa JA, Harris WA, Simons BD, et al. Reconstruction of rat retinal progenitor cell lineages in vitro reveals a surprising degree of stochasticity in cell fate decisions. Development. 2011;138(2):227–35. pmid:21148186
- 19. He J, Zhang G, Almeida AD, Cayouette M, Simons BD, Harris WA. How variable clones build an invariant retina. Neuron. 2012;75(5):786–98. pmid:22958820
- 20. Rulands S, Iglesias-Gonzalez AB, Boije H. Deterministic fate assignment of Müller glia cells in the zebrafish retina suggests a clonal backbone during development. Eur J Neurosci. 2018;48(12):3597–605. pmid:30408243
- 21. Zhang X, Leavey P, Appel H, Makrides N, Blackshaw S. Molecular mechanisms controlling vertebrate retinal patterning, neurogenesis, and cell fate specification. Trends Genet. 2023;39(10):736–57. pmid:37423870
- 22. de Melo J, Clark BS, Blackshaw S. Multiple intrinsic factors act in concert with Lhx2 to direct retinal gliogenesis. Sci Rep. 2016;6:32757. pmid:27605455
- 23. Bringmann A, Pannicke T, Grosche J, Francke M, Wiedemann P, Skatchkov SN, et al. Müller cells in the healthy and diseased retina. Prog Retin Eye Res. 2006;25(4):397–424. pmid:16839797
- 24. Franze K, Grosche J, Skatchkov SN, Schinkinger S, Foja C, Schild D, et al. Muller cells are living optical fibers in the vertebrate retina. Proc Natl Acad Sci U S A. 2007;104(20):8287–92. pmid:17485670
- 25. Nagashima M, Barthel LK, Raymond PA. A self-renewing division of zebrafish Müller glial cells generates neuronal progenitors that require N-cadherin to regenerate retinal neurons. Development. 2013;140(22):4510–21. pmid:24154521
- 26. Lahne M, Hyde DR. Interkinetic Nuclear Migration in the Regenerating Retina. Adv Exp Med Biol. 2016;854:587–93. pmid:26427463
- 27. Gallina D, Todd L, Fischer AJ. A comparative analysis of Müller glia-mediated regeneration in the vertebrate retina. Exp Eye Res. 2014;123:121–30. pmid:23851023
- 28. Hoang T, Wang J, Boyd P, Wang F, Santiago C, Jiang L, et al. Gene regulatory networks controlling vertebrate retinal regeneration. Science. 2020;370(6519):eabb8598. pmid:33004674
- 29. Dyer MA, Cepko CL. Control of Müller glial cell proliferation and activation following retinal injury. Nat Neurosci. 2000;3(9):873–80. pmid:10966617
- 30. Takeda M, Takamiya A, Jiao J-W, Cho K-S, Trevino SG, Matsuda T, et al. alpha-Aminoadipate induces progenitor cell properties of Müller glia in adult mice. Invest Ophthalmol Vis Sci. 2008;49(3):1142–50. pmid:18326742
- 31. Ooto S, Akagi T, Kageyama R, Akita J, Mandai M, Honda Y, et al. Potential for neural regeneration after neurotoxic injury in the adult mammalian retina. Proc Natl Acad Sci U S A. 2004;101(37):13654–9. pmid:15353594
- 32. Liu B, Hunter DJ, Rooker S, Chan A, Paulus YM, Leucht P, et al. Wnt signaling promotes Müller cell proliferation and survival after injury. Invest Ophthalmol Vis Sci. 2013;54(1):444–53. pmid:23154457
- 33. Pavlou M, Probst M, Blasdel N, Prieve AR, Reh TA. The impact of timing and injury mode on induced neurogenesis in the adult mammalian retina. Stem Cell Reports. 2024;19(2):239–53. pmid:38278154
- 34. Al Adhami H, Evano B, Le Digarcher A, Gueydan C, Dubois E, Parrinello H, et al. A systems-level approach to parental genomic imprinting: the imprinted gene network includes extracellular matrix genes and regulates cell cycle exit and differentiation. Genome Res. 2015;25(3):353–67. pmid:25614607
- 35. Berg JS, Lin KK, Sonnet C, Boles NC, Weksberg DC, Nguyen H, et al. Imprinted genes that regulate early mammalian growth are coexpressed in somatic stem cells. PLoS One. 2011;6(10):e26410. pmid:22039481
- 36. Yan X, Himburg HA, Pohl K, Quarmyne M, Tran E, Zhang Y, et al. Deletion of the Imprinted Gene Grb10 Promotes Hematopoietic Stem Cell Self-Renewal and Regeneration. Cell Rep. 2016;17(6):1584–94. pmid:27806297
- 37. Venkatraman A, He XC, Thorvaldsen JL, Sugimura R, Perry JM, Tao F, et al. Maternal imprinting at the H19-Igf2 locus maintains adult haematopoietic stem cell quiescence. Nature. 2013;500(7462):345–9. pmid:23863936
- 38. Zou P, Yoshihara H, Hosokawa K, Tai I, Shinmyozu K, Tsukahara F, et al. p57(Kip2) and p27(Kip1) cooperate to maintain hematopoietic stem cell quiescence through interactions with Hsc70. Cell Stem Cell. 2011;9(3):247–61. pmid:21885020
- 39. Matsumoto A, Takeishi S, Kanie T, Susaki E, Onoyama I, Tateishi Y, et al. p57 is required for quiescence and maintenance of adult hematopoietic stem cells. Cell Stem Cell. 2011;9(3):262–71. pmid:21885021
- 40. Asai T, Liu Y, Di Giandomenico S, Bae N, Ndiaye-Lobry D, Deblasio A, et al. Necdin, a p53 target gene, regulates the quiescence and response to genotoxic stress of hematopoietic stem/progenitor cells. Blood. 2012;120(8):1601–12. pmid:22776820
- 41. Kubota Y, Osawa M, Jakt LM, Yoshikawa K, Nishikawa S-I. Necdin restricts proliferation of hematopoietic stem cells during hematopoietic regeneration. Blood. 2009;114(20):4383–92. pmid:19770359
- 42. Mirshekar-Syahkal B, Haak E, Kimber GM, van Leusden K, Harvey K, O’Rourke J, et al. Dlk1 is a negative regulator of emerging hematopoietic stem and progenitor cells. Haematologica. 2013;98(2):163–71. pmid:22801971
- 43. Bracko O, Singer T, Aigner S, Knobloch M, Winner B, Ray J, et al. Gene expression profiling of neural stem cells and their neuronal progeny reveals IGF2 as a regulator of adult hippocampal neurogenesis. J Neurosci. 2012;32(10):3376–87. pmid:22399759
- 44. Mairet-Coello G, Tury A, Van Buskirk E, Robinson K, Genestine M, DiCicco-Bloom E. p57(KIP2) regulates radial glia and intermediate precursor cell cycle dynamics and lower layer neurogenesis in developing cerebral cortex. Development. 2012;139(3):475–87. pmid:22223678
- 45. Ferrón SR, Radford EJ, Domingo-Muelas A, Kleine I, Ramme A, Gray D, et al. Differential genomic imprinting regulates paracrine and autocrine roles of IGF2 in mouse adult neurogenesis. Nat Commun. 2015;6:8265. pmid:26369386
- 46. Ferrón SR, Charalambous M, Radford E, McEwen K, Wildner H, Hind E, et al. Postnatal loss of Dlk1 imprinting in stem cells and niche astrocytes regulates neurogenesis. Nature. 2011;475(7356):381–5. pmid:21776083
- 47. Furutachi S, Matsumoto A, Nakayama KI, Gotoh Y. p57 controls adult neural stem cell quiescence and modulates the pace of lifelong neurogenesis. EMBO J. 2013;32(7):970–81. pmid:23481253
- 48. Andersen DC, Laborda J, Baladron V, Kassem M, Sheikh SP, Jensen CH. Dual role of delta-like 1 homolog (DLK1) in skeletal muscle development and adult muscle regeneration. Development. 2013;140(18):3743–53. pmid:23946446
- 49. Varrault A, Gueydan C, Delalbre A, Bellmann A, Houssami S, Aknin C, et al. Zac1 regulates an imprinted gene network critically involved in the control of embryonic growth. Dev Cell. 2006;11(5):711–22. pmid:17084362
- 50. Spengler D, Villalba M, Hoffmann A, Pantaloni C, Houssami S, Bockaert J, et al. Regulation of apoptosis and cell cycle arrest by Zac1, a novel zinc finger protein expressed in the pituitary gland and the brain. EMBO J. 1997;16(10):2814–25. pmid:9184226
- 51. Bilanges B, Varrault A, Mazumdar A, Pantaloni C, Hoffmann A, Bockaert J, et al. Alternative splicing of the imprinted candidate tumor suppressor gene ZAC regulates its antiproliferative and DNA binding activities. Oncogene. 2001;20(10):1246–53. pmid:11313869
- 52. Abdollahi A. LOT1 (ZAC1/PLAGL1) and its family members: mechanisms and functions. J Cell Physiol. 2007;210(1):16–25. pmid:17063461
- 53. Pagotto U, Arzberger T, Theodoropoulou M, Grübler Y, Pantaloni C, Saeger W, et al. The expression of the antiproliferative gene ZAC is lost or highly reduced in nonfunctioning pituitary adenomas. Cancer Res. 2000;60(24):6794–9. pmid:11156367
- 54. Koy S, Hauses M, Appelt H, Friedrich K, Schackert HK, Eckelt U. Loss of expression of ZAC/LOT1 in squamous cell carcinomas of head and neck. Head Neck. 2004;26(4):338–44. pmid:15054737
- 55. Abdollahi A, Gruver BN, Patriotis C, Hamilton TC. Identification of epidermal growth factor-responsive genes in normal rat ovarian surface epithelial cells. Biochem Biophys Res Commun. 2003;307(1):188–97. pmid:12849999
- 56. Abdollahi A, Godwin AK, Miller PD, Getts LA, Schultz DC, Taguchi T, et al. Identification of a gene containing zinc-finger motifs based on lost expression in malignantly transformed rat ovarian surface epithelial cells. Cancer Res. 1997;57(10):2029–34. pmid:9158001
- 57. Adnani L, Langevin LM, Gautier E, Dixit R, Parsons K, Li S, et al. Zac1 Regulates the Differentiation and Migration of Neocortical Neurons via Pac1. J Neurosci. 2015;35(39):13430–47. pmid:26424889
- 58. Ma L, Cantrup R, Varrault A, Colak D, Klenin N, Götz M, et al. Zac1 functions through TGFbetaII to negatively regulate cell number in the developing retina. Neural Dev. 2007;2:11. pmid:17559664
- 59. Ma L, Hocking JC, Hehr CL, Schuurmans C, McFarlane S. Zac1 promotes a Müller glial cell fate and interferes with retinal ganglion cell differentiation in Xenopus retina. Dev Dyn. 2007;236(1):192–202. pmid:17072860
- 60. Wang X, Fan J, Zhang M, Ni Y, Xu G. Upregulation of SOX9 in Glial (Müller) cells in retinal light damage of rats. Neurosci Lett. 2013;556:140–5. pmid:24120428
- 61. Chen YY, Liu SL, Hu DP, Xing YQ, Shen Y. N -methyl- N -nitrosourea-induced retinal degeneration in mice. Experimental Eye Research. 2014;121:102–13. pmid:24509257
- 62. van de Pavert SA, Kantardzhieva A, Malysheva A, Meuleman J, Versteeg I, Levelt C, et al. Crumbs homologue 1 is required for maintenance of photoreceptor cell polarization and adhesion during light exposure. J Cell Sci. 2004;117(Pt 18):4169–77. pmid:15316081
- 63. Fu X, Sun H, Klein WH, Mu X. Beta-catenin is essential for lamination but not neurogenesis in mouse retinal development. Dev Biol. 2006;299(2):424–37. pmid:16959241
- 64. Cusato K, Bosco A, Rozental R, Guimarães CA, Reese BE, Linden R, et al. Gap junctions mediate bystander cell death in developing retina. J Neurosci. 2003;23(16):6413–22. pmid:12878681
- 65. García-García D, Locker M, Perron M. Update on Müller glia regenerative potential for retinal repair. Curr Opin Genet Dev. 2020;64:52–9. pmid:32619816
- 66. Tworig JM, Feller MB. Müller Glia in Retinal Development: From Specification to Circuit Integration. Front Neural Circuits. 2022;15:815923. pmid:35185477
- 67. Wohl SG, Jorstad NL, Levine EM, Reh TA. Müller glial microRNAs are required for the maintenance of glial homeostasis and retinal architecture. Nat Commun. 2017;8(1):1603. pmid:29150673
- 68. Varrault A, Dantec C, Le Digarcher A, Chotard L, Bilanges B, Parrinello H, et al. Identification of Plagl1/Zac1 binding sites and target genes establishes its role in the regulation of extracellular matrix genes and the imprinted gene network. Nucleic Acids Res. 2017;45(18):10466–80. pmid:28985358
- 69. Furukawa T, Mukherjee S, Bao ZZ, Morrow EM, Cepko CL. rax, Hes1, and notch1 promote the formation of Müller glia by postnatal retinal progenitor cells. Neuron. 2000;26(2):383–94. pmid:10839357
- 70. Jadhav AP, Cho S-H, Cepko CL. Notch activity permits retinal cells to progress through multiple progenitor states and acquire a stem cell property. Proc Natl Acad Sci U S A. 2006;103(50):18998–9003. pmid:17148603
- 71. Dorsky RI, Rapaport DH, Harris WA. Xotch inhibits cell differentiation in the Xenopus retina. Neuron. 1995;14(3):487–96. pmid:7695895
- 72. Hojo M, Ohtsuka T, Hashimoto N, Gradwohl G, Guillemot F, Kageyama R. Glial cell fate specification modulated by the bHLH gene Hes5 in mouse retina. Development. 2000;127(12):2515–22. pmid:10821751
- 73. Muto A, Iida A, Satoh S, Watanabe S. The group E Sox genes Sox8 and Sox9 are regulated by Notch signaling and are required for Müller glial cell development in mouse retina. Exp Eye Res. 2009;89(4):549–58. pmid:19490914
- 74. Bolster DR, Crozier SJ, Kimball SR, Jefferson LS. AMP-activated protein kinase suppresses protein synthesis in rat skeletal muscle through down-regulated mammalian target of rapamycin (mTOR) signaling. J Biol Chem. 2002;277(27):23977–80. pmid:11997383
- 75. Honnell V, Norrie JL, Patel AG, Ramirez C, Zhang J, Lai Y-H, et al. Identification of a modular super-enhancer in murine retinal development. Nat Commun. 2022;13(1):253. pmid:35017532
- 76. Hamon A, Masson C, Bitard J, Gieser L, Roger JE, Perron M. Retinal Degeneration Triggers the Activation of YAP/TEAD in Reactive Müller Cells. Invest Ophthalmol Vis Sci. 2017;58(4):1941–53. pmid:28384715
- 77. Han S, Okawa S, Wilkinson GA, Ghazale H, Adnani L, Dixit R, et al. Proneural genes define ground-state rules to regulate neurogenic patterning and cortical folding. Neuron. 2021;109(18):2847-2863.e11. pmid:34407390
- 78. Todd L, Fischer AJ. Hedgehog signaling stimulates the formation of proliferating Müller glia-derived progenitor cells in the chick retina. Development. 2015;142(15):2610–22. pmid:26116667
- 79. Gu D, Wang S, Zhang S, Zhang P, Zhou G. Directed transdifferentiation of Müller glial cells to photoreceptors using the sonic hedgehog signaling pathway agonist purmorphamine. Mol Med Rep. 2017;16(6):7993–8002. pmid:28983586
- 80. Jeong J, Mao J, Tenzen T, Kottmann AH, McMahon AP. Hedgehog signaling in the neural crest cells regulates the patterning and growth of facial primordia. Genes Dev. 2004;18(8):937–51. pmid:15107405
- 81. Wang LL, Serrano C, Zhong X, Ma S, Zou Y, Zhang CL. Revisiting astrocyte to neuron conversion with lineage tracing in vivo. Cell. 2021;184(21):5465–81 e16. Epub 2021/09/29. pmid:34582787
- 82. Xie Y, Zhou J, Wang LL, Zhang CL, Chen B. New AAV tools fail to detect Neurod1-mediated neuronal conversion of Müller glia and astrocytes in vivo. eBioMedicine. 2023;90.
- 83. Shang R, Zhang H, Bi P. Generation of mouse conditional knockout alleles in one step using the i-GONAD method. Genome Res. 2021;31(1):121–30. pmid:33328166
- 84. Ueki Y, Wilken MS, Cox KE, Chipman L, Jorstad N, Sternhagen K, et al. Transgenic expression of the proneural transcription factor Ascl1 in Müller glia stimulates retinal regeneration in young mice. Proc Natl Acad Sci U S A. 2015;112(44):13717–22. pmid:26483457
- 85. Roesch K, Jadhav AP, Trimarchi JM, Stadler MB, Roska B, Sun BB, et al. The transcriptome of retinal Müller glial cells. J Comp Neurol. 2008;509(2):225–38. pmid:18465787
- 86. de Melo J, Clark BS, Blackshaw S. Multiple intrinsic factors act in concert with Lhx2 to direct retinal gliogenesis. Sci Rep. 2016;6:32757. pmid:27605455
- 87. Dmitriev AV, Dmitriev AA, Linsenmeier RA. K(+)-dependent Müller cell-generated components of the electroretinogram. Vis Neurosci. 2021;38:E010. pmid:34294176
- 88. Francke M, Faude F, Pannicke T, Bringmann A, Eckstein P, Reichelt W, et al. Electrophysiology of rabbit Müller (glial) cells in experimental retinal detachment and PVR. Invest Ophthalmol Vis Sci. 2001;42(5):1072–9. pmid:11274088
- 89. Shah S, Medisetti S, Fernandes JAL, Mattar P. Chd4 remodels chromatin to control retinal cell type specification and lineage termination. Development. 2025;152(20):dev204697. pmid:40905097
- 90. Das AV, Mallya KB, Zhao X, Ahmad F, Bhattacharya S, Thoreson WB, et al. Neural stem cell properties of Müller glia in the mammalian retina: regulation by Notch and Wnt signaling. Dev Biol. 2006;299(1):283–302. pmid:16949068
- 91. Del Debbio CB, Balasubramanian S, Parameswaran S, Chaudhuri A, Qiu F, Ahmad I. Notch and Wnt signaling mediated rod photoreceptor regeneration by Müller cells in adult mammalian retina. PLoS One. 2010;5(8):e12425. pmid:20865053
- 92. Ghai K, Zelinka C, Fischer AJ. Notch signaling influences neuroprotective and proliferative properties of mature Müller glia. J Neurosci. 2010;30(8):3101–12. pmid:20181607
- 93. Hayes S, Nelson BR, Buckingham B, Reh TA. Notch signaling regulates regeneration in the avian retina. Dev Biol. 2007;312(1):300–11. pmid:18028900
- 94. Le N, Vu T-D, Palazzo I, Pulya R, Kim Y, Blackshaw S, et al. Robust reprogramming of glia into neurons by inhibition of Notch signaling and nuclear factor I (NFI) factors in adult mammalian retina. Sci Adv. 2024;10(28):eadn2091. pmid:38996013
- 95. de Melo J, Zibetti C, Clark BS, Hwang W, Miranda-Angulo AL, Qian J, et al. Lhx2 is an essential factor for retinal gliogenesis and notch signaling. J Neurosci. 2016;36(8):2391–405. pmid:26911688
- 96. Todd L, Jenkins W, Finkbeiner C, Hooper MJ, Donaldson PC, Pavlou M, et al. Reprogramming Müller glia to regenerate ganglion-like cells in adult mouse retina with developmental transcription factors. Sci Adv. 2022;8(47):eabq7219. pmid:36417510
- 97. Murtaugh LC, Stanger BZ, Kwan KM, Melton DA. Notch signaling controls multiple steps of pancreatic differentiation. Proc Natl Acad Sci U S A. 2003;100(25):14920–5. pmid:14657333
- 98. Pinilla I, Lund RD, Sauvé Y. Contribution of rod and cone pathways to the dark-adapted electroretinogram (ERG) b-wave following retinal degeneration in RCS rats. Vision Res. 2004;44(21):2467–74. pmid:15358082
- 99. Cantrup R, Dixit R, Palmesino E, Bonfield S, Shaker T, Tachibana N, et al. Cell-type specific roles for PTEN in establishing a functional retinal architecture. PLoS One. 2012;7(3):e32795. pmid:22403711
- 100. Touahri Y, Dixit R, Kofoed RH, Miloska K, Park E, Raeisossadati R, et al. Focused ultrasound as a novel strategy for noninvasive gene delivery to retinal Müller glia. Theranostics. 2020;10(7):2982–99. pmid:32194850
- 101. Ghazale H, Park E, Vasan L, Mester J, Saleh F, Trevisiol A, et al. Ascl1 phospho-site mutations enhance neuronal conversion of adult cortical astrocytes in vivo. Front Neurosci. 2022;16:917071. pmid:36061596
- 102. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. pmid:25516281
- 103. Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B: Statistical Methodology. 1995;57(1):289–300.
- 104. Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 2000;28(1):27–30. pmid:10592173
- 105. Yu G, Wang L-G, Han Y, He Q-Y. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16(5):284–7. pmid:22455463
- 106. Wu T, Hu E, Xu S, Chen M, Guo P, Dai Z, et al. ClusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innovation (Camb). 2021;2(3):100141. pmid:34557778
- 107. Okawa S, Angarica VE, Lemischka I, Moore K, Del Sol A. A differential network analysis approach for lineage specifier prediction in stem cell subpopulations. NPJ Syst Biol Appl. 2015;1:15012. pmid:28725462
- 108. Nikolsky Y, Ekins S, Nikolskaya T, Bugrim A. A novel method for generation of signature networks as biomarkers from complex high throughput data. Toxicol Lett. 2005;158(1):20–9. pmid:15871913
- 109. Han H, Cho J-W, Lee S, Yun A, Kim H, Bae D, et al. TRRUST v2: an expanded reference database of human and mouse transcriptional regulatory interactions. Nucleic Acids Res. 2018;46(D1):D380–6. pmid:29087512
- 110. Liu Z-P, Wu C, Miao H, Wu H. RegNetwork: an integrated database of transcriptional and post-transcriptional regulatory networks in human and mouse. Database (Oxford). 2015;2015:bav095. pmid:26424082
- 111. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13(11):2498–504. pmid:14597658
- 112. Hao Y, Hao S, Andersen-Nissen E, Mauck WM 3rd, Zheng S, Butler A, et al. Integrated analysis of multimodal single-cell data. Cell. 2021;184(13):3573-3587.e29. pmid:34062119
- 113. Fleming SJ, Chaffin MD, Arduini A, Akkad A-D, Banks E, Marioni JC, et al. Unsupervised removal of systematic background noise from droplet-based single-cell experiments using CellBender. Nat Methods. 2023;20(9):1323–35. pmid:37550580
- 114. Robinson JT, Thorvaldsdóttir H, Winckler W, Guttman M, Lander ES, Getz G, et al. Integrative genomics viewer. Nat Biotechnol. 2011;29(1):24–6. pmid:21221095