Overexpression of Meis factors in late-stage retinal progenitors yields complex effects on temporal patterning and neurogenesis

Abstract

The vertebrate retina serves as a model for studying neurogenesis and cell fate specification, with retinal progenitor cells following a tightly regulated temporal sequence to generate distinct cell types. Meis1 and Meis2 are transcription factors implicated in early retinal development, but their role in late-stage RPCs remains poorly understood. Here, we investigate whether Meis1 and Meis2 overexpression in postnatal mouse RPCs can alter temporal identity and induce early-born cell types. Using electroporation and single-cell RNA sequencing, we find that while these factors modestly upregulate early-stage gene regulatory network components, they do not repress late-stage transcription factors or induce early-born retinal cells. Meis1 overexpression reduces proliferation and inhibits neurogenesis, whereas Meis2 overexpression accelerates neurogenic progression without altering fate commitment. Our findings suggest that overexpression of Meis1 and Meis2 modulate largely non-overlapping aspects of temporal identity and neurogenic competence but are insufficient to fully reprogram late-stage progenitors. These results have implications for regenerative strategies aimed at reprogramming retinal cells for therapeutic purposes.

Introduction

The vertebrate retina, as a lateral extension of the embryonic neural tube, provides a simplified yet integral model for studying neurogenesis and cell fate specification in the central nervous system (CNS)¹. It consists of six major neuronal and one glial cell type, all of which arise from retinal progenitor cells (RPCs) in a temporally regulated sequence. Birthdating studies have shown that retinal ganglion cells are generated first, followed by horizontal cells, GABAergic amacrine cells, cone photoreceptors, non-GABAergic amacrine cells, rod photoreceptors, bipolar cells, and finally Müller glia^2,3. Clonal analyses of isolated RPCs demonstrate that these changes in developmental potential are largely intrinsic (Gomes et al., 2011; Cayouette et al., 2003), driven by dynamic transcription factor expression that governs temporal competence^4,5. While several key transcription factors, including Ikzf1/4, Nfia/b/x, Foxp1, and Caszl, have been identified as regulators of temporal identity^6,7,8,9, much remains unknown about how these factors coordinate developmental transitions.

Recent single-cell RNA sequencing (scRNA-Seq) and chromatin accessibility (scATAC-Seq) studies have identified additional candidate regulators of temporal patterning in the developing retina^7,10,11. Early-stage TFs exhibit high activity in retinal neuroepithelial cells prior to neurogenesis and gradually decline as early-born cell types, such as ganglion cells and cone photoreceptors, are generated. In contrast, the late-stage GRN becomes increasingly active from embryonic day (E) 17 in mice (gestational weeks 11–13 in humans), persisting into postmitotic Müller glia, the final retinal cell type to emerge^1,10. These GRNs are composed of interdependent TFs that reinforce their own expression while repressing factors from the opposing network, yet the functional roles of many predicted regulators remain unexplored., most of which have not been systematically studied for their ability to regulate temporal patterning.

Meis1 and Meis2 are transcription factors predicted to contribute prominently to the early-stage GRN¹⁰. Both genes are selectively expressed in early-stage RPCs and are predicted to positively regulate transcription of multiple early-stage transcription factors. Genetic studies have demonstrated that Meis factors are essential for many different aspects of early retinal development. In zebrafish, meis1 is essential for both dorso-ventral and naso-temporal patterning of the retina prior to the onset of neurogenesis¹², while Meis1 deficiency in mice disrupts neuroretinal formation and delays the onset of neurogenesis^13,14. Combined loss of Meis1/2 completely disrupts neuroepithelial proliferation and neurogenesis, arresting progenitors at an undifferentiated state¹⁵. The functions of Meis1/2 during later stages of neurogenesis have been less extensively investigated. Loss of function of Meis1 following the onset of retinal neurogenesis inhibits expression of Foxn4, which is necessary for generation of early-born horizontal and amacrine cells¹⁶. Meis2 is transiently expressed in developing cone photoreceptors and non-GABAergic amacrine cells and persists in a subset of mature GABAergic amacrine cells¹⁷. However, whether ectopic Meis1/2 expression in late-stage RPCs can reprogram temporal identity and induce early-born cell types remains unknown. This question is particularly relevant given recent efforts to induce neurogenesis from mature Müller glia following overexpression of neurogenic bHLH factors or Oct4, inhibition of Notch signaling, or loss of function of the late-stage GRN components Nfia/b/x^{18,19,20,21,22}. While these strategies have successfully converted Müller glia into interneurons, they have largely failed to produce photoreceptors.

In this study, we use electroporation and scRNA-Seq to investigate whether MEIS1, MEIS2, or their combined overexpression can induce murine postnatal RPCs to adopt an early-stage temporal identity and generate early-born cell types, with the use of human orthologues for overexpression allowed us to test the evolutionary conservation of the role for MEIS factors. While overexpression of these factors modestly increased expression of early-stage GRN components, it did not suppress late-stage TFs. MEIS1 overexpression reduced proliferation and inhibited neurogenesis, whereas Meis2 accelerated neurogenic progression. However, neither factor induced ectopic formation of early-born retinal cells. These findings suggest that MEIS1/2 can modulate aspects of temporal identity and neurogenic competence in an evolutionarily conserved manner but are insufficient to fully reprogram late-stage progenitors or drive robust neurogenesis.

Results

Meis1 and Meis2 expression in retinal progenitor cells

To assess the temporal dynamics of Meis1 and Meis2 expression in primary and neurogenic retinal progenitor cells (RPCs), we analyzed previously published single-cell RNA sequencing (scRNA-Seq) datasets from developing mouse and human retina^7,23 (Fig. 1a, b; S1). Consistent with prior studies¹⁴, both genes were highly expressed in early-stage primary RPCs, peaking at embryonic day (E) 12 in mice, coinciding with the onset of retinal neurogenesis²⁴. A similar pattern was observed in fate-restricted neurogenic RPCs, with expression also peaking at E12. In the human retina, MEIS1 levels declined progressively from gestational week (GW) 8 onward, whereas MEIS2 exhibited an opposing trend, peaking at GW17 (Fig. S2a, b; S3a, b).

Previously, our group identified evolutionarily conserved gene regulatory networks (GRNs) controlling temporal patterning in mouse and human retinal progenitors, including predicted targets of Meis1 and Meis2¹⁰. In mice, genes predicted to be activated by these factors were enriched in early-stage progenitors, whereas genes predicted to be repressed by Meis1/2 were upregulated at later stages, mirroring the dynamic expression of these TFs (Fig. 1a, b; S1a, b). In the human retina, genes predicted to be repressed by MEIS1/2 exhibited similar late-stage upregulation, but the activation of predicted target genes in early RPCs was less pronounced (Fig. S2a, b; S3a, b).

MEIS1 and MEIS2 overexpression in late-stage mouse retinal progenitor cells

To investigate whether MEIS1 and MEIS2 influence RPC temporal identity and neurogenesis, we electroporated P0 mouse retinas ex vivo with pCAGIG-based plasmids expressing full-length human MEIS1, MEIS2, or a 50:50 mixture of both. Control retinas were electroporated with empty pCAGIG vectors (Fig. 2a). Retinas were cultured until P2 or P5, after which GFP-positive cells were isolated via fluorescence-activated cell sorting (FACS) (Fig. S4a, b). scRNA-Seq analysis was performed on duplicate samples for each condition, pooling multiple retinas (n = 12–16 per condition). All cells included in the analysis expressed at least one copy of GFP mRNA (Fig. S4c–e).

Cell types were identified via UMAP clustering and marker gene expression (Fig. 2b, c; Fig. S5, Supplemental Table 1). The composition of electroporated cell populations was consistent with previous studies^7,10, with primary and neurogenic RPCs and amacrine cells predominant at P2, while rods and bipolar cells became more abundant at P5 (Fig. 2d). Minimal numbers of Müller glia were detected, and rare GFP-positive cones, starburst amacrine, and horizontal cells were likely contaminants. Small numbers of GFP-negative microglia were also removed before analysis.

MEIS1/2 overexpression did not alter cell type composition at either P2 or P5 (Fig. 2d), indicating that these factors do not independently induce early-born cell types such as retinal ganglion cells, cones, horizontal cells, or GABAergic amacrine cells. They also neither promoted nor inhibited neurogenesis nor altered the maintenance of primary or neurogenic RPC pools. However, potential effects on temporal patterning or gene expression required further investigation.

MEIS1 and MEIS2 regulate expression of genes specific to early-stage retinal progenitors

To assess whether MEIS1 and MEIS2 regulate their own expression, we examined their auto- and cross-regulation. Overexpression of MEIS1 activated endogenous Meis1 but had no effect on Meis2, while MEIS2 induced Meis2 expression but did not affect Meis1 (Fig. 3a). A comprehensive analysis of gene expression changes induced by MEIS overexpression (log2 fold change > 0.5, p < 0.00001) revealed that both MEIS1 and MEIS2 upregulated the early-stage RPC marker Hmga1 (Fig. 3b, c; Fig. S6a, Supplemental Table 2). MEIS2 specifically upregulated early-stage factors Pbx1, E2f1, and Foxp4, as well as the late-stage factor Nfib, while MEIS1 upregulated Hes1 and Hes5, key Notch pathway regulators. MEIS1 downregulated the neurogenic bHLH factors Neurod1 and Neurod4, the latter of which was also downregulated by MEIS2 (Fig. 3b, c). Notably, co-expression of both factors reversed these effects, increasing Neurod1 and Neurod4 expression (Fig. S6a, Supplemental Table 2). Considerable divergence was seen between MEIS1 and MEIS2-regulated genes, with only 35 genes differentially regulated by both MEIS1 and MEIS2, as opposed to 82 by MEIS1 alone and 699 by MEIS2 alone (Fig. 3d).

Gene ontology (GO) analysis of differentially expressed genes revealed distinct functional categories. MEIS1 targets were enriched for glycolysis, gliogenesis, neurogenesis, eye development, and apoptosis (Fig. 3e), whereas MEIS2 targets included glycolysis, gliogenesis, neurogenesis, apoptosis, cell differentiation, and mRNA transcription (Fig. 3f). Despite substantial functional overlap, the specific genes regulated by each factor showed limited overlap (Fig. S6b).

Fig. 1

Expression of Meis1/2 and expression of genes predicted to be directly upregulated or downregulated by Meis factors in mouse retina. (A) Primary retinal progenitor cells and (B) neurogenic retinal progenitors were analyzed from the dataset published in Clark et al. 2019. Violin expression plots for Meis1 and Meis2. Violin scoring plots for predicted upregulated and downregulated GRNs from Lyu et al. 2021.

Fig. 2

Generating scRNA-Seq datasets for wild type murine retinas overexpressing either control pCAGIG or MEIS1/2 constructs. (A) Schematic for ectopic MEIS1/2 overexpression and isolation of GFP-positive electroporated cells from P0 retinal explants. (B) Uniform manifold approximation (UMAP) and projection for the scRNA-Seq dataset created in A. (C) DotPlot for the cell types present at P2 and P5 within the scRNA-Seq dataset and their associated marker genes. (D) Proportion of each cell type present in each of the treatment conditions.

Fig. 3

Changes in gene expression induced by MEIS1/2 overexpression. (A) Violin plots showing how each treatment affects the expression of endogenous Meis1 or Meis2 across the full dataset. (B) Volcano plot generated from differentially-expressed genes between the P2 and P5 control treated samples and the P2 and P5 samples overexpressing either MEIS1 or (C) MEIS2 for every cell within those datasets. (D) Venn diagram showcasing the overlap in the genes expressed at a 0.5 Log2 Fold change between P2 and P5 MEIS1 and MEIS2 overexpressing samples. (E) Curated GO Terms selected from generated list from the 117 genes with a 0.5 Log2 Fold increase with MEIS1 overexpression. (F) Curated GO Terms selected from the 734 genes with a 0.5 Log2-fold increase with MEIS2 overexpression. P-values are indicated by * number. * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001.

Finally, we tested whether MEIS factor overexpression led to induction of genes specific to individual postmitotic retinal cell types in either primary or neurogenic RPCs (Fig. S7, S8). We observed that MEIS2, both individually or in combination with MEIS1, activated expression of cone-specific genes at both primary and neurogenic RPCs at P2 and P5, and likewise activated amacrine cell-specific genes at P5. No other significant changes in cell-specific gene expression were observed.

Effects of overexpression of MEIS1 and MEIS2 on retinal progenitor cell Temporal identity and differentiation

To determine whether MEIS1 and MEIS2 influence RPC temporal identity, we calculated a temporal index based on enrichment for early- and late-stage markers¹⁰. Genes previously predicted to be activated by MEIS factors were upregulated in all conditions at P2, except for MEIS1 in neurogenic RPCs (Fig. 4a, b; Fig. S9). Surprisingly, neither MEIS1 nor MEIS2 repressed predicted targets; instead, these genes exhibited modest but statistically significant upregulation, particularly at P2 (Fig. 4a, b; Fig. S9).

Analysis of early- and late-stage marker genes revealed that MEIS1 strongly induced early-stage markers in both primary and neurogenic RPCs, while MEIS2 showed this effect only at P5. Co-expression of MEIS1 and MEIS2 enhanced early-stage identity across all conditions. However, none of the tested constructs repressed late-stage markers; instead, MEIS1 and MEIS1 + MEIS2 unexpectedly upregulated these genes (Fig. 4b-c, Fig. S9). In primary RPCs, MEIS2 but not MEIS1 repressed Müller glia-specific genes at P2 while inducing amacrine/horizontal and cone markers (Fig. S7). Further analysis showed that MEIS2 upregulated amacrine/horizontal and cone-specific genes at P2 and P5 in neurogenic RPCs, with additional activation of bipolar-specific genes at both time points (Fig. S8).

Finally, early-stage RPCs exhibit higher proliferation rates and symmetric self-renewing divisions²⁵. To examine whether MEIS overexpression affected RPC proliferation, we analyzed cell cycle phase markers (Fig. S12). At P2, primary RPCs were predominantly in G1, with a smaller G2 population. At P5, the proportion of G1 cells increased as expected. MEIS1 increased G2-phase cells at P2 but reduced them at P5, whereas MEIS2 had no effect (Fig. S12A). To go along with the cell cycle density, we observed that MEIS1 treatment increased the expression of Pcna and Mki67 at P2 (Fig. S12B). In neurogenic RPCs, all MEIS overexpression conditions increased the fraction of S/G2 transition cells at P2, with MEIS2 maintaining this effect at P5 (Fig. S12C). We did not observe notable changes in the expression of Pcna or Mki67 with any treatment in the neurogenic RPC population (Fig S12D). To directly assess proliferation, we performed EdU labeling at P3, P5, and P7, analyzing GFP + EdU + cells at P8 (Fig. 5a, b). MEIS1 significantly reduced EdU incorporation, whereas MEIS2 had no effect (Fig. 5c). We investigated whether apoptosis might be playing a role in the reduction in the observed numbers of GFP+/EdU + cells at P8 following MEIS1 overexpression. By analyzing expression of proapoptotic genes in primary and neurogenic RPCs, we observed that MEIS1 overexpression increased the aggregate proapoptotic score for primary and neurogenic RPCs at P5 relative to GFP treated controls (Fig. S13), potentially accounting for the reduced number of GFP+/EdU + positive cells observed following MEIS1 treatment at P8 (Fig. 5c).

To assess whether MEIS1/2 overexpression altered RPC differentiation timing, we performed pseudotime analysis using AUCell values for early-stage markers. At P2, MEIS2 accelerated differentiation in the youngest primary RPCs, with both MEIS1 and MEIS2 exerting this effect at P5. In neurogenic RPCs, MEIS2 accelerated differentiation at both P2 and P5, whereas MEIS1 had no effect. No treatment affected the oldest progenitor populations (Fig. 6; Fig. S14).

Fig. 4

MEIS overexpression’s impact on predicted target genes and progenitor cell temporal identity. (A) AUCell R package was used to generate composite scores for genes predicted to be upregulated by MEIS1 and/or MEIS2 during retinal development. Split into separate violin plots for primary and neurogenic retinal progenitor cells. (B) Scoring violin plots in RPCs for genes expressed in primary RPCs during early developmental stages. (C) Scoring violin plots for genes expressed in neurogenic RPCs during early developmental stages. P-values are indicated by * number. ns = not significant, * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001.

Fig. 5

Impact of MEIS overexpression on retinal explant proliferation. (A) Schematic showcasing the experimental design to analyze effects of MEIS1 and/or MEIS2 overexpression on RPC proliferation. (B) Representative explants stained for EdU incorporation, GFP to represent electroporated cells, and DAPI for nuclei. (C) Box plots to show the percentage of electroporated cells that are positive for EdU incorporation. P-values are indicated by * number. ns = not significant and * p < 0.05. Scale bar = 50 μm.

Fig. 6

Monocle generated pseudotime trajectories for late born cell types present within the dataset. (A) Using AUCell, a “youngest” primary RPC or (B) neurogenic RPC was identified for each treatment dataset based on expression of known marker genes. This cell was used to generate maturation trajectories for each treatment dataset.

Discussion

This study systematically examines the effects of ectopically expressing MEIS1 and MEIS2 in late-stage retinal progenitors, both individually and in combination. In mice, Meis1 and Meis2 are primarily expressed in early-stage retinal progenitor cells (RPCs), whereas in humans, MEIS2 exhibits a distinct expression pattern, with increased levels in late-stage primary and neurogenic RPCs. Despite differences in expression dynamics, genes predicted to be repressed by MEIS1/2 are consistently upregulated in late-stage RPCs in both species, whereas genes predicted to be activated by these factors show only modest early-stage enrichment. This discrepancy may reflect the divergent expression pattern of MEIS2 between mouse and human.

Although Meis1/2 have been implicated in promoting early-stage temporal identity, their overexpression in postnatal RPCs did not alter cell fate specification. However, significant changes in gene expression were observed, particularly in genes known to regulate retinal temporal patterning and neurogenesis. MEIS1 selectively upregulated Meis1 but had no effect on Meis2, whereas MEIS2 activated Meis2 but did not regulate Meis1. Both MEIS1 and MEIS2 robustly induced the early-stage marker Hmga1, but otherwise activated largely non-overlapping gene sets. For instance, MEIS1 specifically activated the Notch effectors Hes1 and Hes5, while MEIS2 induced genes associated with both early (Pbx1, Foxp4) and late-stage (Nfib) temporal identity. Despite these differences, genes regulated by both factors were functionally enriched for neurogenesis, gliogenesis, apoptosis, and metabolic processes, such as glycolysis. The strong induction of Hk1, Eno1b, and Eno1 following overexpression of MEIS1 and/or MEIS2 suggests a potential role for these factors in regulating metabolic states in early-stage RPCs.

While neither MEIS1 nor MEIS2 overexpression leads to changes in cell fate, they selectively modulated gene expression in ways that provide insight into their developmental functions. Both genes upregulated a subset of early-stage RPC-specific genes but failed to repress late-stage markers, with both of factors these likely accounting for their lack of effect on retinal cell fate specification. Surprisingly, MEIS2 even activated late-stage genes such as Nfib. In addition, MEIS2 repressed Müller glia-specific genes while upregulating genes associated with cones, amacrine cells, and bipolar interneurons—cell types that endogenously express Meis2 during differentiation. These findings suggest that MEIS2 can activate lineage-specific transcriptional programs without fully driving cell fate commitment. Notably, MEIS2, but not MEIS1, accelerated the differentiation of postnatally generated neurons, further supporting its role in promoting neuronal maturation. Single-cell RNA sequencing suggested that MEIS1 promotes primary RPC proliferation while MEIS2 enhances neurogenic RPC proliferation. However, histological analysis revealed a reduction in EdU-labeled cells at P8, suggesting that overexpression of these factors does not sustain proliferation and may instead lead to cell cycle exit or cell loss.

Taken together, these findings demonstrate that MEIS1 and MEIS2 overexpression can activate a subset of early-stage RPC genes and promote expression of select early-born neuronal markers across mammalian species, yet they are insufficient to fully reprogram temporal identity or alter the fate of late-born retinal cell types. This is despite their known necessity for initiating neurogenesis and generating early-born cell types^13,14,15,16. Several factors may account for these limitations. First, Meis1 and Meis2 function within a complex gene regulatory network governing early-stage temporal identity¹⁰. The absence of essential cofactors or intrinsic partners normally present in early-stage RPCs may constrain their ability to reprogram late-stage progenitors. Second, the fact that the ex vivo electroporation protocol used here excludes extraretinal cell types, as well as retinal pigment epithelium, may result in the lack of critical extrinsic signals required for efficient fate reprogramming. Third, both MEIS1 and MEIS2 upregulate proapoptotic factors, potentially leading to selective loss of cells that undergo more efficient temporal reprogramming.

This study highlights the challenges of using individual transcription factors to reprogram temporal identity in retinal progenitors. These findings have implications for regenerative strategies aimed at converting pluripotent stem cells or endogenous retinal glia into photoreceptors and other neuron types lost in blinding diseases. Understanding the broader regulatory networks governing temporal patterning may be key to overcoming these limitations and enhancing the success of cell reprogramming approaches.

Methods

Mice

The use of animals for these studies was conducted using protocols approved by the Johns Hopkins Animal Care and Use Committee, in compliance with ARRIVE guidelines, and were performed in accordance with the protocol approved by the Institutional Animal Care and Use Committee (IACUC) at the Johns Hopkins School of Medicine under protocol number MO22M22.Timed pregnant mice were ordered from Charles River Laboratories for the retinal explant electroporation experiments ending in scRNA-seq or histological analysis.

Retinal explant culture, electroporation, and edu labeling

Retinas were dissected from E19 - P0 embryos from timed pregnant mice, age verified by crown-rump length. Retinas were then electroporated with pCAGIG overexpression constructs at 100 mg in 70 µl of pBS to create 4 treatment groups: pCAGIG empty vector (control, labeled as GFP treated in figures), pCAGIG-MEIS1 vector, pCAGIG-MEIS2 vector, or a 50/50 mixture of pCAGIG-MEIS1 and pCAGIG MEIS2 vectors. The protein sequence of the MEIS1 isoform used in corresponds to GenBank Accession Number NP_002389, while that of the MEIS2 isoform to NP_002390.

Dissected retinas were electroporated using these plasmid concoctions using Model ECM 830 electroporator (BTX-Harvard Apparatus) with this paradigm: 5 square pulses, 80 V and 50 ms. Retinas were electroporated in batches of 5. Electroporated retinas were then flattened using a series of radial cuts, and mounted on 0.2 μm Nuclepore Track-Etch membranes on 2 ml DMEM F12, 10% FBS with 0.1% streptomycin/puromycin in 12-well plates. The explants were incubated at 37 degrees Celsius, 5% CO2 for 2 or 5 days for single cell RNA sequencing experiments. On the third day, 50% of the explant culture media was replaced with new media. Explants used for histological assays were cultured for 8 days. On the third and sixth day, 50% of the explant culture media was replaced with new media. On the third, fifth, and seventh day explants were pulsed with EdU (5 mg/ml in PBS) added directly to the culture media.

Immunohistochemistry and imaging

P8 retinal explants were fixed in 4% paraformaldehyde (Electron Microscopy Sciences, no. 15710) for 2 h at 4 °C, then carefully detached from the membrane in 1× PBS and incubated in 30% sucrose overnight at 4 °C. The explants were then embedded in frozen section medium (VWR, no. 95057-838), cryosectioned at 16-µm thickness, and stored at − 20 °C. Sections were dried for 20 min in a 37 °C incubator and washed three times for 5 min with 0.2% Triton X-100 in PBS (PBST). EdU labeling was performed using a Click-iT EdU kit (Thermo Fisher Scientific, no. C10340) following the manufacturer’s instructions. Sections were then incubated in blocking buffer (10% horse serum (Thermo Fisher Scientific, no. 26050070), 0.2% Triton X-100 in 1× PBS) for 1 h at room temperature (RT) and subsequently incubated with primary antibodies in blocking buffer overnight at 4 °C.

Sections were washed three times for 5 min in PBST to remove excess primary antibodies and were then incubated in secondary antibodies in blocking buffer for 1 h at RT. The sections were washed three times for 5 min in PBST, then once for 5 min in PBS. Sections were then mounted with Fluoromount-G™ with DAPI (Thermo Fisher Scientific, no. 00-4959-52) under coverslips (VWR, no. 48404-453), air-dried, and stored at 4 °C. Fluorescent images were captured using a Zeiss LSM 700 confocal microscope.

Single-cell RNA-sequencing (scRNA-seq) library preparation

Electroporated retinal explants were dissociated using the Worthington Papain Dissociation Kit according to manufacturer’s instructions. scRNA-seq were prepared on GFP + cells isolated from dissociated retinal explants using the 10X Genomics Chromium Single Cell 3’ Reagents Kit v3.1 (10X Genomics, Pleasanton, CA). Libraries were constructed following the manufacturer’s instructions and were sequenced using Illumina NextSeq. Sequencing data were processed through the Cell Ranger 7.0.1 pipeline (10X Genomics) using default parameters.

scRNA-seq data analysis

ScRNA-Seq data were pre-processed using Cellranger v7.1.0 using custom mouse genome mm10 with an added GFP sequence. The cell-by-genes count matrices were further analyzed using Seurat V5.1.0²⁶.Cells with RNA counts less than 500 or greater than 50,000, and number of genes less than 200 or greater than 6000 were filtered out as low quality cells. Additionally, cells with a mitochondrial fraction of greater than 25% were removed. Lastly, cells with a GFP expression level of less than 1 were removed. For data visualization, UMAP was generated using the first 22 dimensions. Primary RPC and neurogenic RPC cell clusters were identified based on expression of published marker genes and subsetted for further analysis. Differential gene expression between the control group and MEIS1, MEIS2, or MEIS1 + MEIS2 was determined through Seurat’s built-in FindMarkers function.

Dataset analysis R packages

AUCell was used for the aggregating scoring plots in Figs. 1 and 4, and supplemental Figs. 2, 7, and 8²⁷. Pheatmap was used to generate the heatmaps in Supplemental Figs. 1, 3, 9, 10, and 11²⁸. ggplot2 was used to generate the Volcano Plots shown in Fig. 3 and Supplemental Fig. 6²⁹. ggplot2 was used to generate the GO term enrichment plots shown in Fig. 3²⁹. The Venn Diagram plots generated for Fig. 3 and Supplemental Fig. 6 created with ggvenn³⁰. ggplot2 was used to generate the box plot for the EdU incorporation in Fig. 5²⁹. Tricycle was used to generate the cell cycle stage density plots in Supplemental Fig. 12³¹. Monocle3 was used to generate the cell fate trajectory plots shown in Fig. 6 and Supplemental Fig. 13³².