Skip to main content
PMC home page
eLife logoLink to eLife
. 2022 Feb 17;11:e71864. doi: 10.7554/eLife.71864

Transcriptional heterogeneity of ventricular zone cells in the ganglionic eminences of the mouse forebrain

Dongjin R Lee 1, Christopher Rhodes 1, Apratim Mitra 2, Yajun Zhang 1, Dragan Maric 3, Ryan K Dale 2, Timothy J Petros 1,✉
Editors: Carol A Mason4, Marianne E Bronner5
PMCID: PMC8887903  PMID: 35175194

Abstract

The ventricular zone (VZ) of the nervous system contains radial glia cells that were originally considered relatively homogenous in their gene expression, but a detailed characterization of transcriptional diversity in these VZ cells has not been reported. Here, we performed single-cell RNA sequencing to characterize transcriptional heterogeneity of neural progenitors within the VZ and subventricular zone (SVZ) of the ganglionic eminences (GEs), the source of all forebrain GABAergic neurons. By using a transgenic mouse line to enrich for VZ cells, we characterize significant transcriptional heterogeneity, both between GEs and within spatial subdomains of specific GEs. Additionally, we observe differential gene expression between E12.5 and E14.5 VZ cells, which could provide insights into temporal changes in cell fate. Together, our results reveal a previously unknown spatial and temporal genetic diversity of VZ cells in the ventral forebrain that will aid our understanding of initial fate decisions in the forebrain.

Research organism: Mouse

Introduction

During embryogenesis, the telencephalon is derived from a neuroepithelium sheet at the anterior end of the neural plate. Along the neuraxis, the telencephalon develops an anterior-posterior and dorsal-ventral identity guided by extrinsic signals (Wilson and Houart, 2004). The dorsal telencephalon gives rise to excitatory glutamatergic cells of the cortex (CTX), whereas the ventral telencephalon is comprised of three transient structures, the lateral, medial and caudal ganglionic eminences (LGE, MGE, and CGE, respectively), that give rise to all forebrain GABAergic cells (Hébert and Fishell, 2008). Each of these regions contains progenitor domains known as the ventricular zone (VZ), which harbors Nestin-expressing radial glia cells (RGCs) and apical progenitors (APs), and the subventricular zone (SVZ) where basal progenitors (BPs) reside (Batista-Brito et al., 2008; Flames et al., 2007; Inan et al., 2012; Long et al., 2009; Petros et al., 2015; Rowitch and Kriegstein, 2010; Turrero García and Harwell, 2017; Wonders et al., 2008). Early studies supported the hypothesis that RGC progenitors in the cortex become progressively restricted over future cell divisions (Desai and McConnell, 2000; Gao et al., 2014; McConnell and Kaznowski, 1991). More recent experiments identified some heterogeneity of forebrain VZ cells (Franco and Müller, 2013; Llorca et al., 2019), but a comprehensive characterization of these VZ neural progenitor cells has not been performed.

There is a significant expansion in the number of SVZ cells as embryogenesis progresses which is likely related to developmental changes in cell fate (Lui et al., 2011). Several mechanisms have been shown to guide cell fate decisions in the ventral telencephalon, including spatial gradients of signaling factors (Brandão and Romcy-Pereira, 2015; Flames et al., 2007; McKinsey et al., 2013; Tyson et al., 2015; Wonders et al., 2008; Xu et al., 2010), cellular birthdates (Bandler et al., 2017; Inan et al., 2012; Miyoshi and Fishell, 2011; Rymar and Sadikot, 2007) and the mode of neurogenesis (Petros et al., 2015), all of which influence the cell cycle dynamics of VZ and SVZ cells. Additionally, many disease-associated genes are enriched in RGCs and interneuron progenitors (Schork et al., 2019; Trevino et al., 2020). Thus, a thorough characterization of gene expression in VZ and SVZ within the ganglionic eminences will increase our understanding of initial cell fate decisions of GABAergic cells in the embryonic brain and provide insight into possible disease mechanisms.

In recent years, single-cell RNA-sequencing (scRNAseq) studies revealed extensive neuronal diversity in the mature telencephalon whereby cells can be classified into relatively clean distinct neuronal subtypes based on their transcriptome (Economo et al., 2018; Tasic et al., 2016; Tasic et al., 2018; Zeisel et al., 2015). However, significantly less is known about neural progenitor diversity within the developing embryo where initial fate decisions arise. Several initial studies performed bulk RNA-sequencing on the embryonic ventral forebrain, but these studies were not able to differentiate molecular characteristics between progenitor domains (Tucker et al., 2008; Zechel et al., 2014). More recent studies performed comprehensive scRNAseq experiments on the developing neocortex, but they did not focus on the ventral telencephalon or interneuron development (Di Bella et al., 2021; Loo et al., 2019; Moreau et al., 2021; Telley et al., 2019; Telley et al., 2016). scRNAseq studies that targeted GEs found that initial signatures of mature interneuron (IN) subtypes appear in postmitotic precursors, whereas very little transcriptional diversity was detected in VZ and SVZ progenitors (Chen et al., 2017; Mayer et al., 2018; Mi et al., 2018). However, these studies were underpowered for detecting potential transcriptional diversity in VZ neural progenitors because the GEs are comprised primarily of SVZ and mantle zone (MZ) cells at these mid-embryonic ages. In other central nervous system regions, transcriptionally heterogeneous VZ cell populations have been reported (Johnson et al., 2015; Li et al., 2020; Ogawa et al., 2005; Yuzwa et al., 2017). Thus, we still lack a definitive characterization of differential gene expression in VZ and SVZ neural progenitors in the embryonic mouse forebrain.

In this study, we performed scRNAseq analyses on the LGE, MGE, CGE, and cortex from E12.5 and E14.5 mice to identify spatial and temporal genetic heterogeneity of VZ and SVZ neural progenitors. To specifically increase the number of VZ cells, we utilized a reporter mouse line containing a destabilized VenusGFP protein that is driven by the Nestin promoter, which limits GFP leakiness into non-VZ cells and ensures GFP is a reliable readout of VZ cells (Sunabori et al., 2008). Using this mouse line, we observed significant transcriptional diversity of VZ cells within the ventral forebrain, both between different GEs as well as spatially segregated expression profiles of VZ cells within specific GEs. Many of these intriguing gene expression patterns were confirmed via multiplex in situ hybridizations. We also uncovered spatially-restricted gene expression patterns in the SVZ between and within the LGE, MGE, and CGE. Last, integrated transcriptional analysis of GE cells from E12.5 and E14.5 mice highlighted differential gene expression between VZ cells during these embryonic timepoints, which could have important implications for the well-described temporal changes in interneuron cell fate (Bandler et al., 2017; Inan et al., 2012; Miyoshi and Fishell, 2011; Rymar and Sadikot, 2007). Thus, our data reveal both spatial and temporal transcriptional diversity in distinct VZ and SVZ progenitor cell populations within the ganglionic eminences, which will further our understanding of neurogenesis during initial fate decisions in the ventral forebrain.

Results

Identification of distinct cell groups in the E12.5 mouse telencephalon

To characterize cellular heterogeneity in the developing telencephalon, we used the 10X Genomics scRNAseq platform to profile the transcriptome of cells from the LGE, MGE, CGE and CTX of E12.5 wild-type (WT) mice. At E12.5, neurogenesis has been occurring in the GEs for several days, with most cells residing in the SVZ and MZ with significantly fewer cycling VZ neural progenitors (Turrero García and Harwell, 2017; Wichterle et al., 2001). Conversely, the vast majority of E12.5 dorsal cortical cells are VZ neural progenitors, with a small number of layer VI projection neurons starting to emerge at this time (Di Bella et al., 2021; Kwan et al., 2012). To ensure that we collected a sufficient number of VZ cells from the GEs to identify transcriptional diversity in this population, we also dissected E12.5 brains from transgenic reporter mice that express destabilized Venus driven by the Nestin promoter and intronic enhancer (Nes-dVenus) (Sunabori et al., 2008; Figure 1A–B). We used fluorescence-activated cell sorting (FACS) to isolate GFP-positive cells from Nes-dVenus E12.5 mouse telencephalon (Figure 1—figure supplement 1). Note that ~ 87% of cortical cells are GFP-positive at E12.5 compared to 41–53% of GE cells, consistent with different proportions of VZ cells in these regions (Figure 1—figure supplement 1B). After filtering out non-viable outliers via the 10X Genomics Cell Ranger pipeline, we obtained >84,000 potential cells (Figure 1—figure supplement 2). We then removed predicted doublets using Doubletfinder (McGinnis et al., 2019) and applied a stringent cutoff of >1500 genes per cell, which resulted in 36,428 E12.5 cells passing quality control that were used for downstream analysis. Cells harvested from both WT and Nes-dVenus mice were used for all subsequent analysis.

Figure 1. scRNAseq of distinct E12.5 forebrain regions from WT and Nes-dVenus mice.

(A) Coronal section through the LGE/MGE (left) and CGE (right) of a E12.5 Nes-dVenus telencephalon immunostained for GFP (green), Nestin (Red) and Dcx (Blue). Scale bar = 200 μm. (B) Schematic of telencephalic cell dissection and single-cell dissociation. (C) Uniform manifold approximation and projection (UMAP) plots of all cells labeled by brain region (left), mouse line (middle) or putative cell clusters (right). (D) Representative genes displaying enriched temporal and/or spatial expression patterns. CTX, cortex; LGE, lateral ganglionic eminence; MGE, medial ganglionic eminence; CGE, caudal ganglionic eminence; WT, wild-type.

Figure 1.

Figure 1—figure supplement 1. Flow cytometry plots depicting isolation of GFP-positive cells in Nes-dVenus mice.

Figure 1—figure supplement 1.

(A) Flow cytometry gating strategy: gated for potential live cells, then gated for DRAQ5+/DAPI- live cells, then gated by DRAQ5 width vs. height to select singlets, then EGFP vs. DAPI to collect GFP+ live cells. (B) FACS plots depicting GFP-positive cells collected from the CTX, LGE, MGE, and CGE in E12.5 Nes-dVenus mice. (C) FACS plots depicting GFP-positive cells collected from the LGE, MGE, and CGE in E14.5 Nes-dVenus mice.
Figure 1—figure supplement 2. Quality-control of sequenced cells.

Figure 1—figure supplement 2.

(A) Cells that had unique feature counts < 4500 or > 1500 were filtered out using Seurat (black horizontal line in top right plot approximates 1500 gene/cell cutoff). Mitochondrial counts over 0.05% and Hbb gene counts over 0.01% were also filtered out. (B) Scatterplots showing the total number of molecules detected versus the number of genes detected before and after the filtrations (top). Additional set of scatterplots showing the total number of molecules detected versus the percentage of mitochondrial genes before and after the filtrations (bottom).
Figure 1—figure supplement 3. Identification of region-specific genes from each telencephalic region at E12.5.

Figure 1—figure supplement 3.

(A) UMAP plot of all cells annotated by brain region. (B) Heatmap showing expression of top 20 differentially expressed genes (DEGs) in each region. Each column represents averaged expression in cells, color-coded as per the color scale. (C) UMAP plots depicting the top 5 DEGs from each region. Cells are color-coded for levels of gene expression as per the color scales.
Open in a new tab

We used the Seurat software package (Stuart et al., 2019) to integrate these cells and identified 20 cell clusters that were largely segregated by brain regions, with clean separation of the CTX and MGE cells while there was greater overlap between the LGE and CGE populations (Figure 1C). We observed common VZ markers (Nestin and Hes5) and SVZ/early neurogenic markers (Dcx and Ccnd2) present in cells from all four regions, as well as GABAergic-specific (Ascl1, Dlx5, and Gad2) and glutamatergic-specific (Pax6, Neurod6, Eomes, Tbr1, and Slc17a7) genes restricted to their expected populations (Figure 1D). By comparing each region individually within the dataset, we observed many genes strongly enriched in the CTX, LGE, MGE, and CGE. While some of these genes have well-described roles in specific brain regions, we did uncover many genes whose regional restricted patterns have not been previously described, such as the alpha-internexin encoding gene Ina and Insulin like Growth Factor Binding Protein five encoding gene Igfbp5 in the CGE (Figure 1—figure supplement 3f). Together, these data reveal different cohorts of telencephalic cells display distinct region-specific gene expression profiles exist at E12.5.

Common neurogenic zones within each ganglionic eminence

To identify specific cell clusters in the GEs, we removed cortex-derived cells and identified 17 GE-derived cell clusters (Figure 2A). Based on Nestin and Dcx expression patterns, we identify seven clusters that likely represent VZ cells (high Nestin), six clusters that likely represent postmitotic cells (high Dcx), and four clusters that likely represent SVZ basal progenitors (moderate levels of both Nestin and Dcx). To confirm successful enrichment of Nestin-expressing VZ cells with the Nes-dVenus mouse, we plotted GE cells based on which mouse line they were derived from. This approach clearly demonstrates that the vast majority of high Nestin-expressing cells are indeed derived from the Nes-dVenus mouse, with significantly fewer VZ cells captured in the WT mice (Figure 2A). Violin plots revealed that most cells expressing VZ-enriched genes such as Nestin, Hes5, and Ccnd1 were harvested from the Nes-dVenus mouse whereas a greater percentage of cells expressing more mature markers such as Dcx, Tubb3, and Gad2 arose from the WT mice (Figure 2—figure supplement 1A). Thus, we have a significantly greater percentage of VZ cells compared to previous studies (Mayer et al., 2018; Mi et al., 2018; Figure 2—figure supplement 2), which provides us greater power to identify transcriptional heterogeneity and smaller subgroups in this VZ cell population that may have been previously overlooked.

Figure 2. Common neurogenic lineages in each ganglionic eminence.

(A) UMAP plots displaying all GE-derived cells annotated from left to right by: brain region, putative cell clusters, mouse strain and expression of VZ marker Nestin and SVZ/MZ marker Dcx. (B) UMAP visualizations show transcriptional diversity in each GE region. Several genes are depicted to highlight distinct neurogenic stages: Nestin, Hes5, and Olig2 are indicative of VZ cells (green), Asc1 and Gadd45g label intermediate SVZ cells (blue), and Dcx, Sp9, and Mapt depict postmitotic neuronal precursors (red). (C) UMAP plots of MGE, LGE, and CGE cells annotated via putative cell clusters and including Slingshot analyses depicting developmental progression through neurogenic stages.

Figure 2.

Figure 2—figure supplement 1. Differential gene expression between mouse lines and brain regions.

Figure 2—figure supplement 1.

(A) Violin plots showing expression of pan-VZ-enriched genes from Nes-dVenus mice and expression of SVZ/early postmitotic genes enriched in cells from WT mice. (B–C) Violin plots depicting differential gene expression profiles of cells arising from the LGE (yellow), MGE (blue), and CGE (red). Highlighted genes include those that are predominantly expressed within one GE (B) or display intriguing expression profiles based on scRNAseq and/or in situ hybridization data (C).
Figure 2—figure supplement 2. Integration and visualization of multiple embryonic mouse scRNAseq datasets.

Figure 2—figure supplement 2.

(A–C) Integration of 3 scRNAseq datasets of embryonic mouse ventral telencephalon: This study, E13.5 MGE and E14.5 LGE and CGE dataset acquired via Drop seq from Mayer et al., 2018, and MGE and CGE dataset from two different timepoints (E12.5 and E14.5) acquired via Fluidigm C1 system by Mi et al., 2018. UMAP plots depicting all cells from the three studies annotated by dataset (A), putative cell clusters (B) and expression of Nestin and Dcx (C). (D–E) High Nestin-expressing cells (> 1.5 normalized dataset) were extracted from the combined dataset. UMAP plot annotated by dataset (D) and bar graphs showing the total number of high Nestin-expressing cells present in each study (E).
Figure 2—figure supplement 3. Genes defining common neurogenic cell types between brain regions.

Figure 2—figure supplement 3.

UMAP visualizations highlighting genes that help define VZ (green), SVZ (blue) and MZ (red) cell populations in the MGE, LGE, and CGE. Cells are color-coded according to their levels of expression as per the color scale. Red ovals in LGE and CGE highlight the Ebf1-positive, Mapt-positive, Sp9-negative population of putative immature GABAergic projection neurons.
Open in a new tab

The LGE, MGE, and CGE cells were individually extracted out and reanalyzed to reveal three common cell types shared between regions: (1) Nestin, Hes5, and Olig2 expressing mitotic VZ cells, (2) Ascl1, Ccnd2, and Gadd45g expressing SVZ/BP cells, and (3) Dcx, Sp9, and Mapt expressing postmitotic SVZ and MZ cells (Figure 2B and Figure 2—figure supplement 3). Focusing on MGE-specific genes, Nkx2.1 was expressed by most cells within the MGE whereas Lhx6 was restricted to postmitotic MGE cells, a subset of which also expressed the mature MGE-derived subtype marker Sst (Figure 1—figure supplement 2B and Figure 2—figure supplement 3). Within LGE and CGE clusters, Ebf1-positive, Mapt-positive and Sp9-negative clusters likely identify GABAergic projection neurons (Nery et al., 2002; Figure 2—figure supplement 3). Also, Npy- and Sst-positive clusters were captured within LGE and CGE dataset representing MGE-derived migrating cortical interneurons migrating through both regions towards the cortex (Kessaris et al., 2014; Figure 2—figure supplement 3). Pseudotime analysis with Slingshot (Street et al., 2018; Trapnell et al., 2014) showed clear trajectories in each region originating from Nestin-positive clusters progressing towards the postmitotic cell markers (Figure 2C). This analysis displays many branching patterns within VZ neural progenitors which could reflect distinct developmental trajectories among Nestin-positive VZ cells in GEs.

We next asked how GE-derived Nestin-expressing VZ neural progenitors are transcriptionally different from cycling Dcx-expressing SVZ cells and postmitotic SVZ/MZ cells. To specifically extract the VZ cells, we isolated cells that had a Nestin expression value above 1.5 from the log normalized GE count data (high Nestin cells). The value of this threshold was optimized by testing different expression values for subsetting, with the goal of harvesting a large number of high Nestin-expressing cells with minimal contamination of Dcx-expressing cells (Figure 4—figure supplement 1A-B). To obtain a relatively clean population of cycling SVZ/BP cells, we isolated cells that expressed Dcx, Mki67 and Ccnd2 but were negative for more mature cell markers like Mapt and Rbfox3. To enrich for postmitotic SVZ/MZ cells, we isolated cells that were negative for Nestin, Hes5, Mki67, Ccnd1, and Ccnd2 but were positive for Dcx (Figure 3A). The subsetting value for postmitotic cells (high Dcx cells) was optimized with the similar approach to that of Nestin-expressing cell enrichment (Figure 4—figure supplement 1C-D). These three distinct cell populations retained their regional segregation and were grouped into 11 clusters (Figure 3A). Comparative analysis between the Nestin-positive, mitotic Dcx-positive, and postmitotic Dcx-positive groups showed clear transcriptional diversity, with 1,083 genes enriched in VZ cells, 330 genes upregulated in mitotic SVZ cells, and 1,320 genes upregulated in postmitotic SVZ/MZ cells (minimum log-fold change = 0.25) (Supplementary file 1). Highlighting the top 20 enriched genes for each cell type revealed several neural stem cell-associated genes found in VZ cells (Fabp7, Hes5, and Ccnd1), BP-associated genes for cycling SVZ cells (Mpped2, Ccnd2, and Sp9) and mature neuronal makers (Mapt, Tubb2a, and Tubb3) were detected in the postmitotic SVZ/MZ cells (Figure 3B–C). Each of the 11 identified cell clusters contained several enriched genes that correlated with their cell types, demonstrating GE-specific gene expression within each VZ, SVZ and MZ domain (Figure 3D). Our data uncovers transcriptional diversity within distinct neurogenic domains that may provide insight into how specific genes regulate early cell fate decisions within the embryo.

Figure 3. Transcriptional heterogeneity within the VZ, SVZ and MZ in E12.5 GEs.

Figure 3.

Open in a new tab

(A) High Nestin- and Dcx-expressing cells (> 1.5 fold expression above normalized dataset) were extracted from the GE populations and replotted. UMAP plots are annotated by: expression of Nestin and Dcx (upper left), brain region (lower left), separated into VZ (high-Nestin, red), SVZ (Dcx-, Mki67-, and Ccnd2-expressing, light blue) and MZ (Dcx-positive and Ccnd2- and Mki67-negative, dark blue) (upper right), and putative cell clusters (lower right). (B) Heatmap of genes enriched in VZ, SVZ and MZ cells. Each column represents expression in a single cell, color-coded as per the color scale. (C) UMAP plots depicting the top five differentially expressed genes (DEGs) in the VZ, SVZ, and MZ regions. (D) Heatmap showing expression of top 5 DEGs in each cell cluster from (A), with colored bar depicting whether each cluster contains predominantly VZ (red), SVZ (light blue), or MZ (dark blue) cells.

Transcriptional heterogeneity throughout the ventricular zone

To uncover transcriptional heterogeneity specifically in VZ cells, we extracted and replotted the high Nestin-expressing cells (threshold >1.5) from the LGE, MGE, and CGE (Figure 4). This resulted in analysis of 9,308 high Nestin-expressing cells (3036 from LGE, 3890 from MGE, and 2302 from CGE). These cells were largely clustered based on brain region and were grouped into eight clusters. There was a large cohort of VZ-enriched genes that were restricted to one specific GE (Figure 4B–C), with some genes being confined to specific VZ clusters (subdomains) within a GE (Figure 4D). We selected ~20 genes that displayed intriguing spatial or cell-type-specific expression patterns within the VZ, the majority of which have not been previously described and visualized them with the RNAscope HiPlex in situ hybridization assay. We identified several pan-VZ genes such as Ednrb, Pkdcc, and Nrarp that are known to regulated by Wnt and Notch signaling pathways that are critical for embryonic development (Krebs et al., 2001; Takeo et al., 2016; Vitorino et al., 2015), as well as mitosis associated genes Prc1, Cenpf, and Ube2c (Engeland, 2018; Figure 5—figure supplement 1A-C). There were also genes that were strongly expressed in VZ cells in only two regions: Ptx3 was expressed in VZ cells throughout the LGE and CGE yet absent from the MGE, while Fgfr3 was strongly expressed in MGE and CGE VZ cells and absent in the LGE (Figure 5A, Figure 2—figure supplement 1C and Figure 2—figure supplement 3C). And several genes were restricted or strongly enriched in VZ cells in only one region, such as Shisa2 and Cntnap2 in the LGE, Igfbp5 in the CGE, and Asb4 in the MGE (Figure 5B–D, Figure 2—figure supplement 1C and Figure 2—figure supplement 3C).

Figure 4. Transcriptional heterogeneity of Nestin-expressing VZ cells in the ganglionic eminences.

(A) High Nestin-expressing (> 1.5 normalized dataset) cells were extracted from GE dataset and replotted as UMAP graphs annotated by brain region and putative cell clusters. (B) Heatmap of top 20 DEGs enriched in VZ cells from the LGE, MGE, and CGE. Each column represents expression in a single cell, color-coded as per the color scale. (C) Dot-plot depicting expression levels of DEGs within each brain region (LGE, MGE, and CGE) and cell cluster. (D) Heatmap showing expression of top 5 DEGs in each cluster from (A), with colored bar depicting whether each cluster contains cells from the LGE (yellow), MGE (blue), or CGE (red). Gray bar indicates cluster containing cells from all three GEs. Each column represents averaged expression in cells, color-coded as per the color scale.

Figure 4.

Figure 4—figure supplement 1. Thresholding optimization for subsetting high Nestin- and Dcx-expressing GE cells.

Figure 4—figure supplement 1.

(A–B) Ridgeline plots showing the distribution of Nestin (A, red) and Dcx (B, blue) expression when E12.5 GEs were isolated at different expression values from the log normalized count data. (C–D) Violin plot showing the median and interquartile range of Nestin and Dcx post-subsetting at Nestin > 1.5 (C) and Dcx > 1.5 (D).
Open in a new tab

Figure 5. Expression profiles of spatially-enriched genes in VZ cells in the ganglionic eminences.

(A–E) Left, RNAscope HiPlex and HiPlexUp in situ hybridization assays of differentially expressed VZ genes on E12.5 GE coronal sections. Right, the corresponding UMAP plots of VZ-enriched genes depicted in the in situs. Scale bar in A = 200 μm. (F) UMAP plot of Nes-enriched cells annotated by brain region. (G) Schematic of a whole mount view of the E12.5 ventral telencephalon depicting the approximate spatial location of VZ-enriched genes within the GEs.

Figure 5.

Figure 5—figure supplement 1. Expression profiles of genes enriched in VZ cells throughout the GEs.

Figure 5—figure supplement 1.

(A–C) Left, RNAscope HiPlex and HiPlexUp in situ hybridization assays of VZ-enriched genes expressed in all GEs. Right, the corresponding UMAP plots of SVZ/MZ-enriched genes depicted in the in situs. Scale bar in A = 200 μm. (D) UMAP plot of Nes-enriched cells annotated by brain region.
Open in a new tab

We also observed many genes that had more refined spatially restricted expression patterns within various GEs, highlighting several examples below. Gadd45g, a gene that regulates stem cell proliferation (Kaufmann and Niehrs, 2011), displayed a salt-and-pepper profile with enriched expression in the dorsal LGE (dLGE) and ventral MGE and CGE (vMGE and vCGE) (Figure 5—figure supplement 1C). The UMAP plots of Id4 and Mest look quite similar, but we observed slight differences in their VZ expression profiles. Id4 was expressed in VZ cells throughout the LGE, CGE, and MGE, whereas Mest was strongly enriched in the LGE and dorsal CGE (Figure 5C). These moderate differences highlight the need to confirm transcriptome expression profiles from scRNAseq experiments via in situ hybridizations.

Mt3 and Rbp1 are both strongly enriched in middle-to-ventral LGE VZ cells and in the dCGE, with only a smattering of Mt3- and Rbp1-expressing cells observed in the MGE (Figure 5D, Figure 2—figure supplement 1C and Figure 2—figure supplement 3C). Igfbp5 expression is exclusively restricted to vCGE VZ cells (Figure 5D, Figure 2—figure supplement 1C and Figure 2—figure supplement 3C). Bcan and Wnt7b are enriched in vMGE VZ cells, with Wnt7b expression extends into the vCGE, whereas Nkx6.2 is expressed at the LGE/MGE boundary as previously described (Sousa et al., 2009; Figure 5E, Figure 2—figure supplement 1C and Figure 2—figure supplement 3C). Of note, Fabp7 is strongly enriched in VZ cells in the dLGE, vMGE, and both dCGE and vCGE, as if forming longitudinal stripes along the dorsal and ventral GE boundaries (Figure 5E). In sum, our data reveal previously unidentified transcriptional heterogeneity within VZ cells throughout the GEs that define specific progenitor subdomains and likely regulate specific GABAergic cell types from these regions (Figure 5G).

Transcriptional heterogeneity in SVZ/MZ cells within the ventral telencephalon

We utilized a similar approach to extract SVZ/MZ cells from the GEs that had high expression levels of Dcx, which displayed fairly clean segregation between LGE, MGE and CGE resulting in 11 cell clusters (total of 12,307 high Dcx-expressing cells with 3900 from LGE, 3902 from MGE and 4505 from CGE; Figure 6—figure supplement 1A). We observed clusters containing previously described region-enriched genes such Nkx2.1, Lhx6 and Lhx8 in the MGE, and Nr2f1 and Nr2f2 for CGE (Kanatani et al., 2008; Lodato et al., 2011); the expression profile of these genes was confirmed with in situ hybridizations (Figure 6A–C, Figure 6—figure supplement 1A-C and Figure 6—figure supplement 2A). CGE-enriched genes Nr2f1 and Nr2f2 are also expressed in the dLGE and POA, in agreement with previous observations (Hu et al., 2017; Figure 6A).

Figure 6. Expression profiles of spatially-enriched genes in SVZ/MZ cells in the ganglionic eminences.

(A–E) Left, RNAscope HiPlex and HiPlexUp in situ hybridization assays of differentially expressed SVZ/MZ genes on E12.5 GE coronal sections. Right, the corresponding UMAP plots of SVZ/MZ-enriched genes depicted in the in situs. Scale bar in A = 200 μm. (F) UMAP plot of Dcx-enriched cells annotated by brain region. (G) Schematic of a coronal section through the E12.5 ventral telencephalon depicting the approximate spatial location of SVZ/MZ-enriched genes within the GEs.

Figure 6.

Figure 6—figure supplement 1. Characterization of genes enriched in high Dcx-expressing cells.

Figure 6—figure supplement 1.

(A) High Dcx-expressing cells (> 1.5 normalized dataset) were extracted from the full GE dataset and replotted, annotated by brain region (middle) and putative cell clusters (right). (B) Heatmap of top 20 SVZ/MZ-enriched genes showing differential expression between LGE, MGE, and CGE. Each column represents expression in a single cell, color-coded as per the color scale. (C) Dot-plot depicting expression levels of SVZ/MZ-enriched DEGs within each brain region (LGE, MGE, and CGE) and cell cluster. (D) Heatmap showing expression of top 5 DEGs in each cluster from (A), with colored bar depicting whether each cluster contains cells from the LGE (yellow), MGE (blue), or CGE (red). Each column represents averaged expression in cells, color-coded as per the color scale.
Figure 6—figure supplement 2. Expression profiles of additional SVZ/MZ-enriched genes in the ganglionic eminences.

Figure 6—figure supplement 2.

(A–D) Left, RNAscope HiPlex and HiPlexUp in situ hybridization assays of SVZ/MZ-enriched genes on E12.5 GE coronal sections. Right, the corresponding UMAP plots of SVZ/MZ-enriched genes depicted in the in situs. Scale bar in A = 200 μm. (E) UMAP plot of Dcx-enriched cells annotated by brain region.
Open in a new tab

We also identified markers that separated cycling SVZ/BP cells from postmitotic MZ cells. Genes such as Ascl1, Dlk2, E2f1, and Tcf4 are expressed in the VZ-SVZ boundary throughout all GEs and then downregulated in postmitotic MZ cells (Figure 6—figure supplement 2B). Conversely, Zfhx3, Ina, Mapt, Mpped2, and Stmn2 are predominantly expressed in Dcx-positive postmitotic cells in the MZ layers, with Zfhx3 and Mpped2 expressed near the SVZ-MZ boundary, whereas Ina, Mapt and Stmn2 are found in deeper MZ regions (Figure 6B and Figure 6—figure supplement 2C). These expression patterns reveal a conserved set of genes that are tightly regulated as cells transition from cycling progenitors to postmitotic GABAergic neurons throughout all GEs.

Other markers displayed more restricted expression patterns within the GEs. Most notably, we identified a ‘stripe’ of MZ cells in the ventral LGE that express a cohort of spatially enriched genes (Mpped2, Pcsk1n, Zfp503, Zfhx3, and Aldh1a3) (Figure 6B–E and Figure 6—figure supplement 2C-D). Both Zfhx3 and Zfp503/Nolz1 have recently been implicated in development of striatal medium spiny neurons (MSNs) (Soleilhavoup et al., 2020; Zhang et al., 2019), and Aldh1a3/Raldh3 is critical for generation of LGE-derived neurons (Chatzi et al., 2011). Ebf1, a marker for direct pathway striatal neurons (Lobo et al., 2008), is also expressed in this region but not as tightly restricted to this narrow domain as the other genes (Figure 6D). Thus, this vLGE population likely represents future striatal MSNs, possibly D1-expressing MSNs (Zhang et al., 2019).

Focusing specifically on the CGE, Cxcl14 is expressed by many CGE-derived interneurons in the mature cortex and hippocampus (Harris et al., 2018; Tasic et al., 2018), although it was reported to not be expressed in the embryonic mouse brain (Wei et al., 2019). We detected Cxcl14 expression in the MZ of the CGE with a bias toward the vCGE (Figure 6D), a similar expression profile to Zfhx3. Conversely, the non-coding RNA Rmst is enriched in the dCGE and displays a complementary pattern to Nr2f1/2 (Figure 6A). Several other genes are also enriched in the SVZ/MZ region of the dCGE, such as Ccnd2 and Flrt2 (Figure 6C–F). Thus, similar to the MGE, there are differential expression patterns along the dorsal-ventral axis of the CGE. How these distinct VZ and SVZ/MZ subdomains related to CGE-derived interneuron fate remains unknown.

To highlight a few more notable expression patterns, the pro-neural gene Pou3f1/Oct-6 is strongly enriched in LGE SVZ adjacent to the Ebf1 domain and is not expressed in the MGE or CGE (Figure 6D). Nrp2 is enriched in deep MZ cells within the dLGE, dMGE, and vCGE, whereas Pcsk1n displays a complementary pattern in the vLGE and vMGE but is not expressed within the CGE (Figure 6—figure supplement 2D). Ccnd2 and Flrt2 were enriched in the SVZ layer of both LGE and CGE with significantly weaker expression in the MGE (Figure 6C–E). Curiously, Rprm is enriched in the SVZ of both the dMGE and a small group of cells in the most ventral CGE (Figure 6C). In sum, these single-cell gene expression profiles provide new insight into the transcriptional diversity in Dcx-positive SVZ/MZ postmitotic GE cells (Figure 6G) and revealed subdomains of genetically defined SVZ/MZ cells that have not been previously described.

Developmental transition of ventral telencephalic VZ neural progenitors over time

There is ample evidence that the capacity for neural progenitors to generate specific neuronal subtypes changes over time (Gal et al., 2006; Pilz et al., 2013). In the MGE, production of SST-expressing interneurons is significantly decreased at E14.5 compared to E12.5 (Bandler et al., 2017; Inan et al., 2012; Miyoshi and Fishell, 2011). To characterize transcriptional changes in GE progenitors over time, we collected LGE, MGE and CGE cells from E14.5 WT and Nes-dVenus mice. These cells were largely segregated by brain region and consisted of 19 cell clusters (Figure 7—figure supplement 1A-C). As with the E12.5 population, significant heterogeneity was still observed when extracting out the highest Nestin- and Dcx-expressing cells (Figure 7—figure supplement 1D-G).

To compare the E12.5 and E14.5 cells, we integrated these datasets together. The majority of postmitotic cell clusters consisted of E14.5 cells, and most Nestin-expressing cells were derived from the Nes-dVenus mouse (Figure 7A). These integrated populations were still primarily segregated based on region and were divided into seventeen different clusters (Figure 7B–C). We isolated strong Nestin- and Dcx-expressing cells from the integrated dataset and performed a differential expression analysis on both sets of cells (Figure 7D–F). High Dcx-expressing from E12.5 and E14.5 were nearly completely overlapping with no significant differences, indicating that the global population of postmitotic cells from GEs are transcriptionally very similar at these ages (Figure 7F). However, when we compared the high Nestin-expressing cells, we observed more age-specific segregation in the dot plot clusters, with one subdomain predominantly consisting of E12.5 cells and another population containing E14.5 cells (Figure 7E).

Figure 7. Comparison of transcriptional changes between E12.5 and E14.5 ventral telencephalon.

(A–C) UMAP plots of all E12.5 and E14.5 cells annotated by mouse line and embryonic timepoint (A), brain region (B), and putative cell clusters (C). (D) UMAP plot of E12.5 and E14.5 cells depicting expression levels of Nestin and Dcx. Red box = approximate location of high Nestin-expressing cells (> 1.5 normalized dataset), blue box = approximate location of high Dcx-expressing cells (> 1.5 normalized dataset). (E) UMAP plot depicting high Nestin-expressing cells annotated by brain region (top), putative cell clusters (lower left) and timepoint (lower right). Note the clouds that are strongly enriched for E12.5 cells (dark green oval) and E14.5 cells (dark red oval). (F) UMAP plot depicting high Dcx-expressing cells annotated by brain region (top), putative cell clusters (lower right) and timepoint (lower left). Note that E12.5 and E14.5 cells are largely overlapping populations with no clear differential clustering. (G) Heatmap depicting the top 20 DEGs between the E12.5-enriched (dark green oval) and E14.5-enriched (dark red oval) clouds of VZ cells from (E). The E14.5-derived cells were removed from the E12.5-enriched cloud, and vice versa. Each column represents expression in a single cell, color-coded as per the color scale. (H) Heatmap depicting the top 5 DEGs in VZ cells from each GE region at E12.5 and E14.5. Each column represents averaged expression in cells, color-coded as per the color scale.

Figure 7.

Figure 7—figure supplement 1. Single-cell transcriptional profiling of E14.5 ganglionic eminences.

Figure 7—figure supplement 1.

(A) Chart describing the number of cells, with mean reads per cell and genes per cell from each E14.5 ganglionic eminence based on CellRanger output. (B–E) UMAP plots of all E14.5 cells annotated by putative cell clusters (B), brain region (C), mouse line (D) and expression of Nestin and Dcx (E). (F) UMAP plot of high Nestin-expressing E14.5 cells (> 1.5 normalized dataset) annotated by brain region (left) and putative cell clusters (right). (G) UMAP plot of high Dcx-expressing E14.5 cells (> 1.5 normalized dataset) annotated by brain region (left) and putative cell clusters (right).
Figure 7—figure supplement 2. Expression of pan-VZ genes in high Nestin-expressing E12.5 and E14.5 cells.

Figure 7—figure supplement 2.

UMAP visualizations show a high level of pan-VZ gene expressions throughout the high Nestin subset E12.5 and E14.5 dataset.
Open in a new tab

To compare these two cell clusters, we removed the E12.5 cells from the E14.5 cloud (and vice versa) and identified many genes enriched specifically in E12.5 or E14.5 Nestin-expressing cells (Figure 7G). Changes in VZ gene expression between these two timepoints were also observed within specific GEs (Figure 7H). Many genes enriched in E14.5 high Nestin-expressing are commonly associated with BPs or postmitotic cells such as Dcx, Tubb3, Mpped2, Arx, and Nrxn3 (Figure 7G). The adult neural progenitor marker Igfbpl1 (Artegiani et al., 2017) was also upregulated in E14.5 VZ cells when compared to E12.5 (Figure 7G). To ensure that the dVenus is not ‘leaking’ into non-VZ cells at E14.5 and contaminating this dataset with SVZ/BP cells, we confirmed that many previously reported pan-VZ cell markers are still expressed in the E14.5 high Nestin-expressing cells (Figure 7—figure supplement 2). Taken together, this data suggests that while Dcx-positive cells are transcriptionally very similar between E12.5 and E14.5, Nestin-positive VZ cells in the GEs display a transition in their gene expression profiles, with a greater number of neurogenic and postmitotic genes present in E14.5 VZ cells.

We next validated some of the differentially expressed genes between E12.5 and E14.5 via RNAscope HiPlexUp. As predicted from the scRNAseq data (Figure 7G), Sfrp1 and Sparc were strongly downregulated in E14.5 VZ GE cells compared to E12.5 (Figure 8A). Mir124a-1hg was strongly enriched in E14.5 VZ cells in the LGE compared to E12.5, agreeing the predictions from the scRNAseq data (Figure 7G), whereas Gucy1a1 expression was upregulated in the SVZ of E14.5 LGE and CGE (Figure 8B). Id4 was strongly downregulated in E14.5 VZ cells in the LGE and MGE, but was significantly upregulated in E14.5 SVZ cells in the LGE and CGE (Figure 8C). Notably, several of these genes displayed weaker expression in the MGE SVZ at E14.5 compared to E12.5 (Mir124a-1hg, Gucy1a1 and Cited2; Figure 8B–C). Focusing specifically on the CGE, we observe a strong downregulation of Hmga2 and Igfbp5 in E14.5 VZ cells and an upregulation of Sox6 in both VZ and SVZ cells (Figure 8D). We also observed numerous gene expression patterns that were consistent between E12.5 and E14.5, such as Slc1a3 in the dCGE VZ, Sp8 in the dLGE SVZ and Ebf1 throughout the LGE SVZ/MZ (Figure 8—figure supplement 1). These in situ hybridizations confirm our scRNAseq data indicating dynamic changes in gene expression patterns in VZ and SVZ cells between E12.5 and E14.5, likely reflecting changes in cell fate trajectories throughout embryonic development.

Figure 8. Genes with differential expression profiles between E12.5 and E14.5 ganglionic eminences.

(A–D) RNAscope HiPlexUp assay of genes differentially expressed in VZ and SVZ cells between E12.5 and E14.5 on coronal sections through the GEs. Dashed circles on the UMAP plot (A) indicate the clouds that are strongly enriched for E12.5 cells (dark green) and E14.5 cells (dark red). Scale bar in A = 200 μm. Dotplot in (A) contains high Nestin-expressing cells from E12.5 and E14.5, dotplots in B-D consist of all E12.5 and E14.5 cells.

Figure 8.

Figure 8—figure supplement 1. Genes with similar expression profiles between E12.5 and E14.5 ganglionic eminences.

Figure 8—figure supplement 1.

(A–C) Left, RNAscope HiPlexUp assay of genes displaying similar expression profiles in VZ and/or SVZ cells between E12.5 and E14.5. Right, UMAP plots of the genes depicted in the in situ panels. Scale bar in A = 200 μm. (D–E) UMAP plots of all E12.5 and E14.5 cells annotated by brain region (D) and mouse line and embryonic timepoint (E).
Open in a new tab

Discussion

We performed scRNAseq on distinct neurogenic forebrain regions to generate a comprehensive transcriptional overview of embryonic telencephalic progenitors. By utilizing the Nes-dVenus mouse, we were able to significantly increase the number of VZ cells and differentiate between cycling VZ cells (RGCs/APs) and cycling SVZ cells (BPs), a distinction that had not been explored in previous forebrain scRNAseq studies. Validating scRNA-seq findings via in situ hybridizations is critical to ensure accuracy of the dataset and to characterize additional spatial heterogeneity that is not obtainable from scRNA-seq data alone. Our in situs revealed a rich diversity of gene expression profiles in progenitors throughout the VZ and SVZ/MZ of the ventral forebrain. These insights could increase our understanding of GABAergic neuronal cell type patterning during neurogenesis and have important implications for understanding neurodevelopmental disorders because many disease-associated genes are enriched in radial glia/VZ cells and interneuron progenitors (Schork et al., 2019; Trevino et al., 2020).

We uncovered striking spatial transcriptional diversity throughout the VZ layers, with many genes displaying spatially-restricted expression patterns that had not been previously reported. We observe a variety of expression patterns, with some genes displaying sharp boundaries within subdomains of GEs while other genes are expressed as gradients across GEs. Some genes were restricted to one or two GEs, with several genes restricted or strongly enriched in the CGE. There are minimal tools to genetically target CGE-specific cell types during embryogenesis because of the lack of established CGE-specific genes, and even some traditionally described CGE-specific genes such as Nr2f1 and Nr2f2 are also expressed in the posterior MGE (Hu et al., 2017). Our finding that Igfbp5 is restricted to the CGE at E12.5 (Figure 5D) warrants further investigation into Igfbp5 as a promising tool to label CGE-derived cells.

Numerous genes display restricted expression patterns that are enriched in dorsal or ventral VZ cells in specific structures. This is particularly relevant due to the ample evidence that certain spatial domains are biased to generate particular interneuron subtypes (Brandão and Romcy-Pereira, 2015; Flames et al., 2007; McKinsey et al., 2013; Tyson et al., 2015; Wonders et al., 2008; Xu et al., 2010). While a relationship between spatial domain and interneuron subtype has been described in the MGE, whether distinct CGE-derived interneuron subtypes arise from specific CGE regions remains unknown. Several genes such as Rbp1, Id4, and Wnt7b are enriched in subdomains of VZ cells within the CGE, which may indicate that CGE subdomains preferentially give rise to specific interneuron subtypes. Another intriguing expression profile was Fabp7, which is traditionally believed to be a pan-VZ/radial glia cell marker (Arai et al., 2005; Schmid et al., 2006; Yun et al., 2012; Yuzwa et al., 2017). However, in the ventral telencephalon, Fabp7 is restricted to the dorsal and ventral GE boundaries, as if demarcating the GEs from the surrounding forebrain.

We also detected transcriptionally defined subdomains within the SVZ/MZ layers to a greater extent than was previously appreciated. For example, Mbip and Rprm show a complimentary expression profile at the VZ/SVZ and SVZ/MZ border in MGE, respectively. We observed several genes (Mpped2, Pcsk1n, Aldh1a3, Zfhx3, and Zfp503) strongly enriched at the SVZ/MZ boundary in the ventral LGE (Figure 6 and Figure 6—figure supplement 2), which likely represent future medium spiny neurons of the striatum (Chatzi et al., 2011; Soleilhavoup et al., 2020; Zhang et al., 2019). Within the CGE, Nrp2, Rprm, and Zfp503 are enriched in the ventral region whereas Mpped2, Ccnd2, and Flrt2 are enriched in the SVZ/MZ of the dorsal CGE. As cell fates become more defined as cells transition from cycling to postmitotic at this SVZ/MZ border, further work is needed to characterize the role of these spatially restricted genes in this process.

Additionally, we also identify a transition in the transcriptional profile of VZ cells from E12.5 to E14.5 that was not observed in SVZ/MZ cells, which were transcriptionally very similar between the two ages. Temporal transcriptional changes in VZ cells were most prominent in the LGE according to our comparative and RNAscope analyses (Figures 7E and 8). In general, E14.5 VZ cells from the GEs begin to express genes that are indicative of more mature cells, such as Dcx, Arx, and Gad2. Many of the VZ cell-enriched genes at E14.5 are also strongly expressed in the SVZ layer (Figures 7G–H ,–8B). Perhaps environment-dependent signals arising at later embryonic timepoints drive E14.5 VZ cells to become ‘extraverted’ when compared to their ‘introverted’ E12.5 counterparts (Telley et al., 2019). These transcriptional changes are occurring during a simultaneous increase in the number of SVZ cells/BPs with a proportional decrease in the number of proliferative VZ cells, as most MGE VZ cells give rise to SVZ/BP cells by E14.5 (Glickstein et al., 2007; Petros et al., 2015; Tsoa et al., 2014). Gene expression changes in VZ cells over time could guide this transition in cell cycle dynamics and in part regulate temporal changes in cell fate throughout neurogenesis (Bandler et al., 2017; Inan et al., 2012; Miyoshi and Fishell, 2011). It will be intriguing to explore how these changes in gene expression in VZ cells over time alters their neurogenic potential and cell fate capacity.

In sum, this study characterized the gene expression profiles of VZ and SVZ cells in distinct neurogenic regions of the embryonic telencephalon, providing important insights into the transcriptome of distinct regional domains that may guide early neuronal fate decisions. There is growing evidence that transcriptional diversity in distinct progenitor subtypes is critical for generating neuronal diversity (Cheung et al., 2010; Molnár et al., 2006). Hence, characterizing the developmental expression profiles in distinct progenitor cells in different brain regions during neurogenesis will further explain origins of cortical interneuron subtypes and how the neocortex is formed or altered in neurodevelopmental disorders.

Materials and methods

Key resources table.

Reagent type (species) or resource Designation Source or reference Identifiers Additional information
Strain, strain background (Mus musculus) Nes-dVenus RIKEN BioResource Research Center C57BL/6-Tg(Nes-d4YFP*)1HOkn,RRID:IMSR_RBRC04058 Pooled sexes
Strain, strain background (Mus musculus) Wild-Type The Jackson Laboratory C57BL/6 J000664,RRID:IMSR_JAX:000664 Pooled sexes
Sequence-based reagent Chromium Next GEM Single Cell 3' Kit 10X Genomics v2 (120267) v3 (1000092) v2 & v3
Commercial assay or kit RNAscope HiPlex Assays Advanced Cell Diagnostics 324,102 v1
Antibody Anti-GFP (Rabbit polyclonal) Invitrogen A11122,RRID:AB_221569 1:400
Antibody Anti-Doublecortin (Chicken polyclonal) Abcam AB153668,RRID:AB_2728759 1:1,000
Antibody Anti-Nestin (Rat Monoclonal) Millipore MAB353,RRID:AB_94911 1:100
Software, algorithm Cell Ranger 10X Genomics RRID:SCR_017344 v3.0.0
Software, algorithm Seurat the Satija Lab at the New York Genome Center RRID:SCR_007322 v3.0.0
Software, algorithm Slingshot Kelly Street and Sandrine Dudoit at UC Berkeley RRID:SCR_017012 https://github.com/kstreet13/slingshotv2.2.0
Software, algorithm DoubletFinder Christopher S. McGinnis and Zev J. Gartner at UCSF RRID:SCR_018771 https://github.com/chris-mcginnis-ucsf/DoubletFinderv3.0
Software, algorithm RNAscope HiPlexImage Registration Advanced Cell Diagnostics v1 and v2
Software, algorithm Photoshop CC Adobe RRID:SCR_014199 20.0.9
Software, algorithm ImageJ National Health Institution RRID:SCR_003070 http://imagej.nih.gov/ij/
Open in a new tab

Animals

All mouse colonies were maintained in accordance with protocols approved by the Animal Care and Use Committee at the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD). Wild-type (WT) C57BL/6 mice were obtained from The Jackson Laboratory whereas Nestin-d4-Venus (Nes-dVenus) mice were provided by RIKEN BioResource Research Center (Sunabori et al., 2008). For all embryonic experiments, the day on which a vaginal plug was found was considered as embryonic day (E) 0.5.

Single-cell isolation

WT and Nes-dVenus pregnant dams at stage E12.5 and E14.5 were euthanized with Euthasol. Embryos were removed and incubated on ice in oxygenated artificial cerebrospinal fluid (ACSF) throughout dissection. ACSF, in mM: 87 NaCl, 26 NaHCO3, 2.5 KCl, 1.25 NaH2PO4, 0.5 CaCl2, 7 MgCl2, 10 glucose, 75 sucrose saturated with 95% O2, 5% CO2, pH 7.4. Cortices and all three GEs were microdissected and tissue collected into labeled tubes with ACSF. Tissue was pooled from > 4 embryos for each experiment and was then enzymatically dissociated with 1 mg/ml of Pronase (Sigma-Aldrich #10165921001) in ACSF for 15–20 min. Pronase solution was removed and 1–2 ml of reconstitution solution (ACSF +1:100 fetal bovine serum (FBS) +0.01% DNase) was added to each tube before mechanically dissociating with fire-polished glass pipettes of large, medium and small-bore openings. DAPI (1 μl) and DRAQ5 (5 μM, ThermoFisher #62251) were then added and cell solution was passed through a pre-wetted 35 μm filter prior to sorting (Nes-dVenus) or cell counting (WT).

Fluorescence-activated cell sorting

Dissociated cell solution from Nes-dVenus mice were sorted with Beckman Coulter MoFlo Astrios cell sorter or Sony SH800S sorter with 100 μm chips. Cells were first gated with forward scatter (FSC) vs. side scatter (SSC) to remove debris, then gated with DAPI vs. DRAQ5 to select for live cells (DRAQ5+/DAPI-), then gated with GFP 488 vs. FSC to harvest GFP-expressing cells. Cells were collected into DNA LoBind microcentrifuge tubes (Eppendorf #022431021) containing cold oxygenated ACSF supplemented with 1% FBS.

Single-cell RNA sequencing and library generation

For each experiment, ~ 15,000 cells from WT or sorted Nes-dVenus mice were run through the 10X Genomics Single Cell controller. Chromium Single Cell 3’ GEM (versions 2 and 3), Library and Gel Bead Kits were used according to the manufacturer’s instructions. Libraries were sequenced on Illumina HiSeq 2500 by the NICHD Molecular Genomics Core. For the E12.5 MGE/LGE/CGE, three biological replicates were used to make independent libraries (1 WT and 2 Nes-dVenus). For the E12.5 CTX and the E14.5 MGE/LGE/CGE samples, two replicates were used for library construction (1 WT and 1 Nes-dVenus). Library quality and DNA content were assessed using a BioAnalyzer (Agilent) and Qubit (ThermoFisher), respectively. Libraries from each experiment were balanced by DNA quantity, pooled, and sequenced on an Illumina HiSeq 2500 by the NICHD Molecular Genomics Core.

Single-cell RNA sequencing analysis

Reads were aligned to the mouse genome (mm10) before generating a gene by barcode count matrix using Cell Ranger (10X Genomics) with default parameters. In total, ~120,000 cells were obtained from the original Cell Ranger pipeline. To analyze high quality cells only, we first removed cells that had unique molecular identifier (UMI) counts > 4500 or < 1500. We also ran Doubletfinder (McGinnis et al., 2019) to identify doublets and removed data points when the doublet score was greater than 0.3 from further analysis in each of the datasets. Cells were further filtered based on the percentage of UMI counts associated with mitochondrial mapped reeds and number of detected genes per cell, and we removed cells > 3 median absolute deviations (MADs) from the median of the total population. For cells derived from WT mice, an additional step was taken by filtering out cells based on the percentage of UMI associated with hemoglobin subunit beta (Hbb) transcripts (since Nes-dVenus cells were sorted, they did not require this additional filtering step). Filtration of cells based on quality control metrics, data normalization, scaling, dimensionality reduction, highly variable feature detections, population subsetting and data integration were all performed with the Seurat package using the standard gene expression workflow and default parameters except when scaling counts, where the mitochondrial transcript and ribosomal transcript percentages were regressed out, before non-linear dimensional reduction and community detection were performed (Stuart et al., 2019). Applying these stringent filtration steps resulted in a total of 36,428 E12.5 cells and 24,218 E14.5 cells that were analyzed in this manuscript, with the following breakdown: E12.5: 12,052 MGE cells, 10,316 LGE cells, 9275 CGE cells, 4785 CTX cells. E14.5: 8389 MGE cells, 8658 LGE cells, 7171 CGE cells.

To characterize lineage arising from progenitors, Slingshot was used for single cell trajectory inference using UMAP projections (Street et al., 2018). For the comparative analysis of VZ, SVZ, and MZ cells, the effects of cell cycle heterogeneity were mitigated by calculating cell-cycle phase scores based on known canonical markers (Nestorowa et al., 2016) and the scores were regressed out during the data scaling using ScaleData. To distinguish cycling and postmitotic cells from Dcx-expressing population subsets, Mki67- and Ccnd2-expressing cells (> 0.5 the log normalized count data) were considered SVZ cells whereas Mki67- and Ccnd2-negative cells (< 0.5 the log normalized count data) were labeled as MZ cells. To integrate multiple datasets, standard Seurat integration workflow was followed by performing the log normalization method. However, when combining previously reported datasets (GES103983 and GES109796; Mayer et al., 2018; Mi et al., 2018) with our data, the SCTransform normalization method was used under Seurat v3 integration workflow. For GSE109796, CGE, dMGE, and vMGE datasets from both E12.5 and E14.5 were analyzed. Cells that expressed the dorsal telencephalic cortical markers such as Tbr1, Eomes, Neurod2, and Neurod6 were considered contamination from the microdissected GE samples and removed prior to SCTransform normalization.

Fluorescent in situ hybridization and immunostaining

E12.5 and E14.5 embryonic brains were removed and drop-fixed in 4% paraformaldehyde overnight at 4°C. Fixed brains were washed in PBS, incubated in 30% sucrose overnight at 4°C and the cryopreserved. Tissues were cryosectioned at 14–16 µm in the coronal plane. RNAscope HiPlex and HiPlexUp in situ hybridization assays (Advanced Cell Diagnostics) were performed according to the manufacturer’s instructions. In situ hybridization images were taken using Zeiss AxioImager.M2 with or without ApoTome.2. Autofluorescence signals from blood vessels in embryonic brain tissues were corrected using Photoshop (Adobe) layer masking strategies. Image outlining was performed using ImageJ’s (National Institutes of Health) Canny Edge Detector and superimpose of all images was conducted with RNAscope HiPlex Image Registration software (Advanced Cell Diagnostics).

For immunofluorescence, 16 μm sections were blocked for > 1 hr at RT in blocking buffer (10% Normal Donkey Serum in PBS + 0.3% Triton X-100) and incubated at 4 °C overnight with primary antibodies in blocking buffer. Sections were washed 3 × 5 min at RT in PBS and incubated > 1 hr at RT with fluorescent secondary antibodies and DAPI in blocking buffer. Before mounting, sections were washed again 3 × 5 at RT in PBS and images were captured with AxioImager.M2 with ApoTom.2. Primary antibodies used were rabbit anti-GFP (Invitrogen A11122, 1:400), chicken anti-doublecortin (Abcam AB153668, 1:1000) and mouse anti-Nestin (Millipore MAB353, 1:100).

Acknowledgements

We thank Steven L Coon, James R Iben and Tianwei Li at the NICHD Molecular Genomics Core for sequencing services and initial Cell Ranger analyses; Dr. Hideyuki Okano for the Nestin-dVenus mice; and Dr. Dae-sung Kim and members of the Petros lab for helpful comments on the manuscript.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Timothy J Petros, Email: tim.petros@nih.gov.

Carol A Mason, Columbia University, United States.

Marianne E Bronner, California Institute of Technology, United States.

Funding Information

This paper was supported by the following grant:

  • Eunice Kennedy Shriver National Institute of Child Health and Human Development Intramural Award to Timothy J Petros.

Additional information

Competing interests

No competing interests declared.

Author contributions

Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing - original draft, Writing - review and editing.

Data curation, Formal analysis, Methodology, Writing - review and editing.

Data curation, Formal analysis, Software, Writing - review and editing.

Investigation, Writing - review and editing.

Investigation, Writing - review and editing.

Formal analysis, Software, Supervision, Visualization, Writing - review and editing.

Conceptualization, Investigation, Methodology, Project administration, Resources, Supervision, Visualization, Writing - original draft, Writing - review and editing.

Ethics

All mouse colonies were maintained in accordance with protocols approved by the Animal Care and Use Committee at the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) under animal study protocol ASP #20-047.

Additional files

Supplementary file 1. Differentially expressed genes in the VZ and SVZ.

Table contains the genes that are enriched in the Nestin-positive VZ cells (1,083 genes), mitotic Dcx-positive SVZ cells (330 genes) and postmitotic Dcx-positive SVZ/MZ cells (1,320 genes), all with a minimum log-fold change = 0.25.

elife-71864-supp1.xlsx (239.7KB, xlsx)
Transparent reporting form

Data availability

All of our sequencing data has been deposited in GEO under accession code GSE167013 and GSE190593.

The following datasets were generated:

Lee DR, Rhodes CT, Mitra A, Zhang Y, Maric D, Dale RK, Petros TJ. 2021. Transcriptional heterogeneity of ventricular zone cells throughout the embryonic mouse forebrain. NCBI Gene Expression Omnibus. GSE167013

Lee DR, Rhodes CT, Mitra A, Zhang Y, Maric D, Dale RK, Petros TJ. 2021. Transcriptional heterogeneity of ventricular zone cells throughout the embryonic mouse forebrain. NCBI Gene Expression Omnibus. GSE190593

The following previously published datasets were used:

Mayer C, Hafemeister C, Bandler RC, Machold R, Batista Brito R, Jaglin X, Allaway K, Butler A, Fishell G, Satija R. 2018. Developmental diversification of cortical inhibitory interneurons. NCBI Gene Expression Omnibus. GSE103983

Mi D, Li Z, Lim L, Li M, Moissidis M, Yang Y, Gao T, Hu TX, Pratt T, Price DJ, Sestan N, Marin O. 2018. Early emergence of cortical interneuron diversity in the mouse embryo. NCBI Gene Expression Omnibus. GSE109796

References

  1. Arai Y, Funatsu N, Numayama-Tsuruta K, Nomura T, Nakamura S, Osumi N. Role of Fabp7, a downstream gene of Pax6, in the maintenance of neuroepithelial cells during early embryonic development of the rat cortex. The Journal of Neuroscience. 2005;25:9752–9761. doi: 10.1523/JNEUROSCI.2512-05.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Artegiani B, Lyubimova A, Muraro M, van Es JH, van Oudenaarden A, Clevers H. A Single-Cell RNA Sequencing Study Reveals Cellular and Molecular Dynamics of the Hippocampal Neurogenic Niche. Cell Reports. 2017;21:3271–3284. doi: 10.1016/j.celrep.2017.11.050. [DOI] [PubMed] [Google Scholar]
  3. Bandler RC, Mayer C, Fishell G. Cortical interneuron specification: the juncture of genes, time and geometry. Current Opinion in Neurobiology. 2017;42:17–24. doi: 10.1016/j.conb.2016.10.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Batista-Brito R, Machold R, Klein C, Fishell G. Gene expression in cortical interneuron precursors is prescient of their mature function. Cerebral Cortex (New York, N.Y. 2008;18:2306–2317. doi: 10.1093/cercor/bhm258. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Brandão JA, Romcy-Pereira RN. Interplay of environmental signals and progenitor diversity on fate specification of cortical GABAergic neurons. Frontiers in Cellular Neuroscience. 2015;9:149. doi: 10.3389/fncel.2015.00149. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Chatzi C, Brade T, Duester G. Retinoic acid functions as a key GABAergic differentiation signal in the basal ganglia. PLOS Biology. 2011;9:e1000609. doi: 10.1371/journal.pbio.1000609. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chen Y-JJ, Friedman BA, Ha C, Durinck S, Liu J, Rubenstein JL, Seshagiri S, Modrusan Z. Single-cell RNA sequencing identifies distinct mouse medial ganglionic eminence cell types. Scientific Reports. 2017;7:45656. doi: 10.1038/srep45656. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Cheung AFP, Kondo S, Abdel-Mannan O, Chodroff RA, Sirey TM, Bluy LE, Webber N, DeProto J, Karlen SJ, Krubitzer L, Stolp HB, Saunders NR, Molnár Z. The subventricular zone is the developmental milestone of a 6-layered neocortex: comparisons in metatherian and eutherian mammals. Cerebral Cortex (New York, N.Y. 2010;20:1071–1081. doi: 10.1093/cercor/bhp168. [DOI] [PubMed] [Google Scholar]
  9. Desai AR, McConnell SK. Progressive restriction in fate potential by neural progenitors during cerebral cortical development. Development (Cambridge, England) 2000;127:2863–2872. doi: 10.1242/dev.127.13.2863. [DOI] [PubMed] [Google Scholar]
  10. Di Bella DJ, Habibi E, Stickels RR, Scalia G, Brown J, Yadollahpour P, Yang SM, Abbate C, Biancalani T, Macosko EZ, Chen F, Regev A, Arlotta P. Molecular logic of cellular diversification in the mouse cerebral cortex. Nature. 2021;595:554–559. doi: 10.1038/s41586-021-03670-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Economo MN, Viswanathan S, Tasic B, Bas E, Winnubst J, Menon V, Graybuck LT, Nguyen TN, Smith KA, Yao Z, Wang L, Gerfen CR, Chandrashekar J, Zeng H, Looger LL, Svoboda K. Distinct descending motor cortex pathways and their roles in movement. Nature. 2018;563:79–84. doi: 10.1038/s41586-018-0642-9. [DOI] [PubMed] [Google Scholar]
  12. Engeland K. Cell cycle arrest through indirect transcriptional repression by p53: I have a DREAM. Cell Death and Differentiation. 2018;25:114–132. doi: 10.1038/cdd.2017.172. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Flames N, Pla R, Gelman DM, Rubenstein JLR, Puelles L, Marín O. Delineation of multiple subpallial progenitor domains by the combinatorial expression of transcriptional codes. The Journal of Neuroscience. 2007;27:9682–9695. doi: 10.1523/JNEUROSCI.2750-07.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Franco SJ, Müller U. Shaping our minds: stem and progenitor cell diversity in the mammalian neocortex. Neuron. 2013;77:19–34. doi: 10.1016/j.neuron.2012.12.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gal JS, Morozov YM, Ayoub AE, Chatterjee M, Rakic P, Haydar TF. Molecular and morphological heterogeneity of neural precursors in the mouse neocortical proliferative zones. The Journal of Neuroscience. 2006;26:1045–1056. doi: 10.1523/JNEUROSCI.4499-05.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gao P, Postiglione MP, Krieger TG, Hernandez L, Wang C, Han Z, Streicher C, Papusheva E, Insolera R, Chugh K, Kodish O, Huang K, Simons BD, Luo L, Hippenmeyer S, Shi S-H. Deterministic progenitor behavior and unitary production of neurons in the neocortex. Cell. 2014;159:775–788. doi: 10.1016/j.cell.2014.10.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Glickstein SB, Alexander S, Ross ME. Differences in cyclin D2 and D1 protein expression distinguish forebrain progenitor subsets. Cerebral Cortex (New York, N.Y. 2007;17:632–642. doi: 10.1093/cercor/bhk008. [DOI] [PubMed] [Google Scholar]
  18. Harris KD, Hochgerner H, Skene NG, Magno L, Katona L, Bengtsson Gonzales C, Somogyi P, Kessaris N, Linnarsson S, Hjerling-Leffler J. Classes and continua of hippocampal CA1 inhibitory neurons revealed by single-cell transcriptomics. PLOS Biology. 2018;16:e2006387. doi: 10.1371/journal.pbio.2006387. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hébert JM, Fishell G. The genetics of early telencephalon patterning: some assembly required. Nature Reviews. Neuroscience. 2008;9:678–685. doi: 10.1038/nrn2463. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hu JS, Vogt D, Lindtner S, Sandberg M, Silberberg SN, Rubenstein JLR. Coup-TF1 and Coup-TF2 control subtype and laminar identity of MGE-derived neocortical interneurons. Development (Cambridge, England) 2017;144:2837–2851. doi: 10.1242/dev.150664. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Inan M, Welagen J, Anderson SA. Spatial and temporal bias in the mitotic origins of somatostatin- and parvalbumin-expressing interneuron subgroups and the chandelier subtype in the medial ganglionic eminence. Cerebral Cortex (New York, N.Y. 2012;22:820–827. doi: 10.1093/cercor/bhr148. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Johnson MB, Wang PP, Atabay KD, Murphy EA, Doan RN, Hecht JL, Walsh CA. Single-cell analysis reveals transcriptional heterogeneity of neural progenitors in human cortex. Nature Neuroscience. 2015;18:637–646. doi: 10.1038/nn.3980. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Kanatani S, Yozu M, Tabata H, Nakajima K. COUP-TFII is preferentially expressed in the caudal ganglionic eminence and is involved in the caudal migratory stream. The Journal of Neuroscience. 2008;28:13582–13591. doi: 10.1523/JNEUROSCI.2132-08.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kaufmann LT, Niehrs C. Gadd45a and Gadd45g regulate neural development and exit from pluripotency in Xenopus. Mechanisms of Development. 2011;128:401–411. doi: 10.1016/j.mod.2011.08.002. [DOI] [PubMed] [Google Scholar]
  25. Kessaris N, Magno L, Rubin AN, Oliveira MG. Genetic programs controlling cortical interneuron fate. Current Opinion in Neurobiology. 2014;26:79–87. doi: 10.1016/j.conb.2013.12.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Krebs LT, Deftos ML, Bevan MJ, Gridley T. The Nrarp gene encodes an ankyrin-repeat protein that is transcriptionally regulated by the notch signaling pathway. Developmental Biology. 2001;238:110–119. doi: 10.1006/dbio.2001.0408. [DOI] [PubMed] [Google Scholar]
  27. Kwan KY, Sestan N, Anton ES. Transcriptional co-regulation of neuronal migration and laminar identity in the neocortex. Development (Cambridge, England) 2012;139:1535–1546. doi: 10.1242/dev.069963. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Li Z, Tyler WA, Zeldich E, Santpere Baró G, Okamoto M, Gao T, Li M, Sestan N, Haydar TF. Transcriptional priming as a conserved mechanism of lineage diversification in the developing mouse and human neocortex. Science Advances. 2020;6:eabd2068. doi: 10.1126/sciadv.abd2068. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Llorca A, Ciceri G, Beattie R, Wong FK, Diana G, Serafeimidou-Pouliou E, Fernández-Otero M, Streicher C, Arnold SJ, Meyer M, Hippenmeyer S, Maravall M, Marin O. A stochastic framework of neurogenesis underlies the assembly of neocortical cytoarchitecture. eLife. 2019;8:e51381. doi: 10.7554/eLife.51381. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Lobo MK, Yeh C, Yang XW. Pivotal role of early B-cell factor 1 in development of striatonigral medium spiny neurons in the matrix compartment. Journal of Neuroscience Research. 2008;86:2134–2146. doi: 10.1002/jnr.21666. [DOI] [PubMed] [Google Scholar]
  31. Lodato S, Tomassy GS, De Leonibus E, Uzcategui YG, Andolfi G, Armentano M, Touzot A, Gaztelu JM, Arlotta P, Menendez de la Prida L, Studer M. Loss of COUP-TFI alters the balance between caudal ganglionic eminence- and medial ganglionic eminence-derived cortical interneurons and results in resistance to epilepsy. The Journal of Neuroscience. 2011;31:4650–4662. doi: 10.1523/JNEUROSCI.6580-10.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Long JE, Cobos I, Potter GB, Rubenstein JLR. Dlx1&2 and Mash1 transcription factors control MGE and CGE patterning and differentiation through parallel and overlapping pathways. Cerebral Cortex (New York, N.Y) 2009;19 Suppl 1:i96–i106. doi: 10.1093/cercor/bhp045. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Loo L, Simon JM, Xing L, McCoy ES, Niehaus JK, Guo J, Anton ES, Zylka MJ. Single-cell transcriptomic analysis of mouse neocortical development. Nature Communications. 2019;10:134. doi: 10.1038/s41467-018-08079-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Lui JH, Hansen DV, Kriegstein AR. Development and evolution of the human neocortex. Cell. 2011;146:18–36. doi: 10.1016/j.cell.2011.06.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Mayer C, Hafemeister C, Bandler RC, Machold R, Batista Brito R, Jaglin X, Allaway K, Butler A, Fishell G, Satija R. Developmental diversification of cortical inhibitory interneurons. Nature. 2018;555:457–462. doi: 10.1038/nature25999. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. McConnell SK, Kaznowski CE. Cell cycle dependence of laminar determination in developing neocortex. Science (New York, N.Y.) 1991;254:282–285. doi: 10.1126/science.254.5029.282. [DOI] [PubMed] [Google Scholar]
  37. McGinnis CS, Murrow LM, Gartner ZJ. DoubletFinder: Doublet Detection in Single-Cell RNA Sequencing Data Using Artificial Nearest Neighbors. Cell Systems. 2019;8:329–337. doi: 10.1016/j.cels.2019.03.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. McKinsey GL, Lindtner S, Trzcinski B, Visel A, Pennacchio LA, Huylebroeck D, Higashi Y, Rubenstein JLR. Dlx1&2-dependent expression of Zfhx1b (Sip1, Zeb2) regulates the fate switch between cortical and striatal interneurons. Neuron. 2013;77:83–98. doi: 10.1016/j.neuron.2012.11.035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Mi D, Li Z, Lim L, Li M, Moissidis M, Yang Y, Gao T, Hu TX, Pratt T, Price DJ, Sestan N, Marín O. Early emergence of cortical interneuron diversity in the mouse embryo. Science (New York, N.Y.) 2018;360:81–85. doi: 10.1126/science.aar6821. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Miyoshi G, Fishell G. GABAergic interneuron lineages selectively sort into specific cortical layers during early postnatal development. Cerebral Cortex (New York, N.Y. 2011;21:845–852. doi: 10.1093/cercor/bhq155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Molnár Z, Métin C, Stoykova A, Tarabykin V, Price DJ, Francis F, Meyer G, Dehay C, Kennedy H. Comparative aspects of cerebral cortical development. European Journal of Neuroscience. 2006;23:921–934. doi: 10.1111/j.1460-9568.2006.04611.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Moreau MX, Saillour Y, Cwetsch AW, Pierani A, Causeret F. Single-cell transcriptomics of the early developing mouse cerebral cortex disentangle the spatial and temporal components of neuronal fate acquisition. Development (Cambridge, England) 2021;148:dev197962. doi: 10.1242/dev.197962. [DOI] [PubMed] [Google Scholar]
  43. Nery S, Fishell G, Corbin JG. The caudal ganglionic eminence is a source of distinct cortical and subcortical cell populations. Nature Neuroscience. 2002;5:1279–1287. doi: 10.1038/nn971. [DOI] [PubMed] [Google Scholar]
  44. Nestorowa S, Hamey FK, Pijuan Sala B, Diamanti E, Shepherd M, Laurenti E, Wilson NK, Kent DG, Göttgens B. A single-cell resolution map of mouse hematopoietic stem and progenitor cell differentiation. Blood. 2016;128:e20–e31. doi: 10.1182/blood-2016-05-716480. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Ogawa Y, Takebayashi H, Takahashi M, Osumi N, Iwasaki Y, Ikenaka K. Gliogenic radial glial cells show heterogeneity in the developing mouse spinal cord. Developmental Neuroscience. 2005;27:364–377. doi: 10.1159/000088452. [DOI] [PubMed] [Google Scholar]
  46. Petros TJ, Bultje RS, Ross ME, Fishell G, Anderson SA. Apical versus Basal Neurogenesis Directs Cortical Interneuron Subclass Fate. Cell Reports. 2015;13:1090–1095. doi: 10.1016/j.celrep.2015.09.079. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Pilz G-A, Shitamukai A, Reillo I, Pacary E, Schwausch J, Stahl R, Ninkovic J, Snippert HJ, Clevers H, Godinho L, Guillemot F, Borrell V, Matsuzaki F, Götz M. Amplification of progenitors in the mammalian telencephalon includes a new radial glial cell type. Nature Communications. 2013;4:2125. doi: 10.1038/ncomms3125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Rowitch DH, Kriegstein AR. Developmental genetics of vertebrate glial-cell specification. Nature. 2010;468:214–222. doi: 10.1038/nature09611. [DOI] [PubMed] [Google Scholar]
  49. Rymar VV, Sadikot AF. Laminar fate of cortical GABAergic interneurons is dependent on both birthdate and phenotype. The Journal of Comparative Neurology. 2007;501:369–380. doi: 10.1002/cne.21250. [DOI] [PubMed] [Google Scholar]
  50. Schmid RS, Yokota Y, Anton ES. Generation and characterization of brain lipid-binding protein promoter-based transgenic mouse models for the study of radial glia. Glia. 2006;53:345–351. doi: 10.1002/glia.20274. [DOI] [PubMed] [Google Scholar]
  51. Schork AJ, Won H, Appadurai V, Nudel R, Gandal M, Delaneau O, Revsbech Christiansen M, Hougaard DM, Bækved-Hansen M, Bybjerg-Grauholm J, Giørtz Pedersen M, Agerbo E, Bøcker Pedersen C, Neale BM, Daly MJ, Wray NR, Nordentoft M, Mors O, Børglum AD, Bo Mortensen P, Buil A, Thompson WK, Geschwind DH, Werge T. A genome-wide association study of shared risk across psychiatric disorders implicates gene regulation during fetal neurodevelopment. Nature Neuroscience. 2019;22:353–361. doi: 10.1038/s41593-018-0320-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Soleilhavoup C, Travaglio M, Patrick K, Garção P, Boobalan E, Adolfs Y, Spriggs RV, Moles-Garcia E, Dhiraj D, Oosterveen T, Ferri SL, Abel T, Brodkin ES, Pasterkamp RJ, Brooks BP, Panman L. Nolz1 expression is required in dopaminergic axon guidance and striatal innervation. Nature Communications. 2020;11:3111. doi: 10.1038/s41467-020-16947-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Sousa VH, Miyoshi G, Hjerling-Leffler J, Karayannis T, Fishell G. Characterization of Nkx6-2-derived neocortical interneuron lineages. Cerebral Cortex (New York, N.Y. 2009;19 Suppl 1:i1–i10. doi: 10.1093/cercor/bhp038. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Street K, Risso D, Fletcher RB, Das D, Ngai J, Yosef N, Purdom E, Dudoit S. Slingshot: cell lineage and pseudotime inference for single-cell transcriptomics. BMC Genomics. 2018;19:477. doi: 10.1186/s12864-018-4772-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Stuart T, Butler A, Hoffman P, Hafemeister C, Papalexi E, Mauck WM, Hao Y, Stoeckius M, Smibert P, Satija R. Comprehensive Integration of Single-Cell Data. Cell. 2019;177:1888–1902. doi: 10.1016/j.cell.2019.05.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Sunabori T, Tokunaga A, Nagai T, Sawamoto K, Okabe M, Miyawaki A, Matsuzaki Y, Miyata T, Okano H. Cell-cycle-specific nestin expression coordinates with morphological changes in embryonic cortical neural progenitors. Journal of Cell Science. 2008;121:1204–1212. doi: 10.1242/jcs.025064. [DOI] [PubMed] [Google Scholar]
  57. Takeo M, Lee W, Rabbani P, Sun Q, Hu H, Lim CH, Manga P, Ito M. EdnrB Governs Regenerative Response of Melanocyte Stem Cells by Crosstalk with Wnt Signaling. Cell Reports. 2016;15:1291–1302. doi: 10.1016/j.celrep.2016.04.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Tasic B, Menon V, Nguyen TN, Kim TK, Jarsky T, Yao Z, Levi B, Gray LT, Sorensen SA, Dolbeare T, Bertagnolli D, Goldy J, Shapovalova N, Parry S, Lee C, Smith K, Bernard A, Madisen L, Sunkin SM, Hawrylycz M, Koch C, Zeng H. Adult mouse cortical cell taxonomy revealed by single cell transcriptomics. Nature Neuroscience. 2016;19:335–346. doi: 10.1038/nn.4216. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Tasic B, Yao Z, Graybuck LT, Smith KA, Nguyen TN, Bertagnolli D, Goldy J, Garren E, Economo MN, Viswanathan S, Penn O, Bakken T, Menon V, Miller J, Fong O, Hirokawa KE, Lathia K, Rimorin C, Tieu M, Larsen R, Casper T, Barkan E, Kroll M, Parry S, Shapovalova NV, Hirschstein D, Pendergraft J, Sullivan HA, Kim TK, Szafer A, Dee N, Groblewski P, Wickersham I, Cetin A, Harris JA, Levi BP, Sunkin SM, Madisen L, Daigle TL, Looger L, Bernard A, Phillips J, Lein E, Hawrylycz M, Svoboda K, Jones AR, Koch C, Zeng H. Shared and distinct transcriptomic cell types across neocortical areas. Nature. 2018;563:72–78. doi: 10.1038/s41586-018-0654-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Telley L, Govindan S, Prados J, Stevant I, Nef S, Dermitzakis E, Dayer A, Jabaudon D. Sequential transcriptional waves direct the differentiation of newborn neurons in the mouse neocortex. Science (New York, N.Y.) 2016;351:1443–1446. doi: 10.1126/science.aad8361. [DOI] [PubMed] [Google Scholar]
  61. Telley L, Agirman G, Prados J, Amberg N, Fièvre S, Oberst P, Bartolini G, Vitali I, Cadilhac C, Hippenmeyer S, Nguyen L, Dayer A, Jabaudon D. Temporal patterning of apical progenitors and their daughter neurons in the developing neocortex. Science (New York, N.Y.) 2019;364:eaav2522. doi: 10.1126/science.aav2522. [DOI] [PubMed] [Google Scholar]
  62. Trapnell C, Cacchiarelli D, Grimsby J, Pokharel P, Li S, Morse M, Lennon NJ, Livak KJ, Mikkelsen TS, Rinn JL. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nature Biotechnology. 2014;32:381–386. doi: 10.1038/nbt.2859. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Trevino AE, Sinnott-Armstrong N, Andersen J, Yoon S-J, Huber N, Pritchard JK, Chang HY, Greenleaf WJ, Pașca SP. Chromatin accessibility dynamics in a model of human forebrain development. Science (New York, N.Y.) 2020;367:eaay1645. doi: 10.1126/science.aay1645. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Tsoa RW, Coskun V, Ho CK, de Vellis J, Sun YE. Spatiotemporally different origins of NG2 progenitors produce cortical interneurons versus glia in the mammalian forebrain. PNAS. 2014;111:7444–7449. doi: 10.1073/pnas.1400422111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Tucker ES, Segall S, Gopalakrishna D, Wu Y, Vernon M, Polleux F, Lamantia AS. Molecular specification and patterning of progenitor cells in the lateral and medial ganglionic eminences. The Journal of Neuroscience. 2008;28:9504–9518. doi: 10.1523/JNEUROSCI.2341-08.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Turrero García M, Harwell CC. Radial glia in the ventral telencephalon. FEBS Letters. 2017;591:3942–3959. doi: 10.1002/1873-3468.12829. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Tyson JA, Goldberg EM, Maroof AM, Xu Q, Petros TJ, Anderson SA. Duration of culture and sonic hedgehog signaling differentially specify PV versus SST cortical interneuron fates from embryonic stem cells. Development (Cambridge, England) 2015;142:1267–1278. doi: 10.1242/dev.111526. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Vitorino M, Silva AC, Inácio JM, Ramalho JS, Gur M, Fainsod A, Steinbeisser H, Belo JA. Xenopus Pkdcc1 and Pkdcc2 Are Two New Tyrosine Kinases Involved in the Regulation of JNK Dependent Wnt/PCP Signaling Pathway. PLOS ONE. 2015;10:e0135504. doi: 10.1371/journal.pone.0135504. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Wei S, Du H, Li Z, Tao G, Xu Z, Song X, Shang Z, Su Z, Chen H, Wen Y, Liu G, You Y, Zhang Z, Yang Z. Transcription factors Sp8 and Sp9 regulate the development of caudal ganglionic eminence-derived cortical interneurons. The Journal of Comparative Neurology. 2019;527:2860–2874. doi: 10.1002/cne.24712. [DOI] [PubMed] [Google Scholar]
  70. Wichterle H, Turnbull DH, Nery S, Fishell G, Alvarez-Buylla A. In utero fate mapping reveals distinct migratory pathways and fates of neurons born in the mammalian basal forebrain. Development (Cambridge, England) 2001;128:3759–3771. doi: 10.1242/dev.128.19.3759. [DOI] [PubMed] [Google Scholar]
  71. Wilson SW, Houart C. Early steps in the development of the forebrain. Developmental Cell. 2004;6:167–181. doi: 10.1016/s1534-5807(04)00027-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Wonders CP, Taylor L, Welagen J, Mbata IC, Xiang JZ, Anderson SA. A spatial bias for the origins of interneuron subgroups within the medial ganglionic eminence. Developmental Biology. 2008;314:127–136. doi: 10.1016/j.ydbio.2007.11.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Xu Q, Guo L, Moore H, Waclaw RR, Campbell K, Anderson SA. Sonic hedgehog signaling confers ventral telencephalic progenitors with distinct cortical interneuron fates. Neuron. 2010;65:328–340. doi: 10.1016/j.neuron.2010.01.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Yun SW, Leong C, Zhai D, Tan YL, Lim L, Bi X, Lee JJ, Kim HJ, Kang NY, Ng SH, Stanton LW, Chang YT. Neural stem cell specific fluorescent chemical probe binding to FABP7. PNAS. 2012;109:10214–10217. doi: 10.1073/pnas.1200817109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Yuzwa SA, Borrett MJ, Innes BT, Voronova A, Ketela T, Kaplan DR, Bader GD, Miller FD. Developmental Emergence of Adult Neural Stem Cells as Revealed by Single-Cell Transcriptional Profiling. Cell Reports. 2017;21:3970–3986. doi: 10.1016/j.celrep.2017.12.017. [DOI] [PubMed] [Google Scholar]
  76. Zechel S, Zajac P, Lönnerberg P, Ibáñez CF, Linnarsson S. Topographical transcriptome mapping of the mouse medial ganglionic eminence by spatially resolved RNA-seq. Genome Biology. 2014;15:486. doi: 10.1186/s13059-014-0486-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  77. Zeisel A, Muñoz-Manchado AB, Codeluppi S, Lönnerberg P, La Manno G, Juréus A, Marques S, Munguba H, He L, Betsholtz C, Rolny C, Castelo-Branco G, Hjerling-Leffler J, Linnarsson S. Brain structure Cell types in the mouse cortex and hippocampus revealed by single-cell RNA-seq. Science (New York, N.Y.) 2015;347:1138–1142. doi: 10.1126/science.aaa1934. [DOI] [PubMed] [Google Scholar]
  78. Zhang Z, Wei S, Du H, Su Z, Wen Y, Shang Z, Song X, Xu Z, You Y, Yang Z. Zfhx3 is required for the differentiation of late born D1-type medium spiny neurons. Experimental Neurology. 2019;322:113055. doi: 10.1016/j.expneurol.2019.113055. [DOI] [PubMed] [Google Scholar]

Editor's evaluation

Carol A Mason 1

Your study is a thorough transcriptomics study demonstrating previously uncharacterized transcriptional diversity, discriminating the embryonic domains that produce interneurons. You have responded well to the requests for amendments, adjusted figures and better emphasized gene diversity in the ganglionic eminence.

Decision letter

Editor: Carol A Mason1

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Transcriptional heterogeneity of ventricular zone cells throughout the embryonic mouse forebrain" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided that this work will not be considered further for publication by eLife at this time.

The three reviewers found your study of transcriptional diversity of cortex and ganglionic eminences (GE) of the mouse embryo, focusing on progenitors that produce cortical interneurons of the brain, of importance to the field. They appreciated your use of a Nestin GFP line enabling enrichment for progenitors, and were positive about your finding of previously uncharacterized transcriptional diversity discriminating the embryonic domains producing interneurons.

Nonetheless, even with your novel findings and extensive analysis, the reviewers concurred that potential impact of the study is weakened by technical concerns on the sc RNA Seq analysis as well as by the lack of data supporting relevance of observed sub-types. They call for lineage analysis of at least one potential progenitor subtype to bolster the notion that transcriptionally heterogeneous progenitors are relevant to different fate outcomes. There was praise for the in situs but a call for additional in situ analysis, adding more genes and looking at different embryonic stages. Finally, they suggest that you pose your findings more broadly and beyond the lists of genes, classifying potential subtypes in a model to implicate which are primarily early born and those that are late born. Some of these revisions would be possible without performing additional work (cleaning up the sc RNA Seq analysis) but other inquiries would require further bench work. We hope that the reviewers' comments will help you to strengthen your story based on this data set.

Reviewer #1:

The manuscript describes the RNA fingerprint of developing telencephalic cells. To focus on progenitors, the authors use a transgenic line that drives the expression of a destabilized-venus GFP under the Nestin promoter. With this strategy, the authors find higher diversity of developing precursors than when they dissociate single cells from unlabeled WT brains because the latter contains higher proportions of postmitotic cells. The main finding is that early VZ precursors are more heterogeneous than inferred from previous studies. In situs in brain sections confirm molecularly distinct clusters of progenitors in medial, lateral, and dorsal VZ. The work is nicely written and organized. The results support the main conclusions. The findings advance the field, and the data will be useful in future research.

1. The work is nicely written and the conclusions about the heterogeneity of ventral precursors are well supported and important contribution to the field. However, the study of dorsal cell is minor. The analysis of dorsal cells seems to include many dorsal venus-expressing cells that are postmitotic neurons, and the number of analyzed cells from the dorsal origin is small.

My advice is to include the data of the study of dorsal cells as supplementary data, not as principal data.

2. On the other hand, I consider that the current supplementary Figure 1 showing the histological expression of the venus reporter should be presented as the main figure.

3. The field is rapidly offering new reports describing single cell analysis of telencephalic progenitors. The authors should revise and update the bibliography and discussion accordingly, including: "Single-cell transcriptomics of the early developing mouse cerebral cortex disentangle the spatial and temporal components of neuronal fate acquisition. Moreau et al., Development (2021) 148 (14): dev197962"; "Molecular logic of cellular diversification in the mouse cerebral cortex. Di Bella et la., Nature. 2021 Jul;595(7868):554-559".

Reviewer #2:

In this study, Lee et al. investigate the molecular diversity of cortex and ganglionic eminences (GE) of the mouse embryo. The authors used single-cell transcriptomics technology to acquire gene expression profiles of telencephalic regions, which they separated by microdissection of the lateral, caudal and medial ganglionic eminences along with the cortex in E12.5 and E14.5 mouse embryos. Using the Nes-dVenus line combined with Fluorescence-activated sorting they could enrich their dataset in ventricular zone cells, mainly neuronal progenitors. With this dataset, they could identify common differentiation lineages between the different GEs. Using the RNAscope technology, the authors validated unreported genes expressed in a specific ganglionic eminence and compartment within the GEs (VZ, SVZ, or MZ). In addition, the manuscript shows a bigger temporal mark on VZ cells compared to post-mitotic neurons between E12.5 and E14.5.

This manuscript presents a complete dataset of single-cell sequencing of ganglionic eminences during development. The effort put in the microdissection of the different GEs, central for the understanding of spatial implication in molecular identity and diversity, along with the validation of newly discovered genes with the RNAscope technique is very appreciated. Nevertheless, the manuscript shows significant weaknesses in the analysis of the transcriptomics data and quality controls. The overall tone is very descriptive such that this manuscript is rather intended for a specialized readership. I have the following comments:

1. In the methods, the authors describe the quality control applied to exclude bad quality cells from the data set (Supplementary Figure 2). At line 384, the authors say: "we first removed cells that had unique molecular identifies (UMI) counts over 4500 or less than 200." From what is shown in supplementary figure 2 we observe that this filtering is done on the number of expressed genes and not on the UMI counts. In addition, in the violin plot of a number of genes per cell, the distribution looks binomial with a population of cells expressing around 500 genes (highlighted in orange) and another population expressing around 3000 genes per cell (highlighted in blue below). The population with a low number of genes expressed should be excluded for further analysis because it is likely that it represents 10x GEMs containing ambient RNA or dying cells instead of GEMS containing a whole cell. This is supported by the post-filtering graph at the bottom of the figure (% of Mitochondrion vs Counts Depth) where, strikingly for the cortex, we observe a population of cells with low % of mitochondrial RNA and high Count depth and a band of cells formed by low Count Depth cells with a variety of mitochondrial RNA (highlighted in red). I strongly recommend the authors filter out the low number of genes population (in orange) by setting a threshold of around 1500 genes per cell which will probably clean the populations with low count depth and thus avoid including bad quality GEMs for further analysis.

2. At the lines 90-92 of the manuscript, the authors write: "Conversely, cortical neurogenesis does not begin until ~E14 (Noctor et al., 2004; Sessa et al., 2008; Tyler et al., 2015), so the majority of E12.5 cortical cells are mainly VZ neural progenitors, specifically RGCs." Indeed, at E12.5 the ventricular zone is the predominant compartment of the developing pallium and most of the cells present at this stage are radial glial cells, nevertheless, asymmetric division of radial glia to produce neurons has been reported as soon as E11.5 in the mouse (Kwan et al. Development, 2012, doi: 10.1242/dev.069963), explaining the small but present proportion of neurons found in the dataset.

3. In line 258 the authors write: "To compare the E12.5 and E14.5 cells, we integrated these datasets together". It is unclear whether the two datasets were just merged in a common Seurat object or if they were integrated with SCTransform normalization and integrated with the Seurat v3 pipeline. I would suggest that this sort of integration not be applied when combining datasets generated in such similar conditions because it can remove the differences between datasets, in this case, age of collection.

4. In Figure 7, the authors nicely show a big overlap and few differentially expressed genes in post-mitotic cells between E12.5 and E14.5. It would be very interesting to know if this overlap is shared for different areas of the telencephalon such as the pallium or if it is specific to the subpallium.

5. In order to isolate basal progenitor cells, the authors selected cells expressing both neuronal markers such as Dcx and cycling marker as Mki67. Since in the 10x experiments it is frequent to have doublets and their exclusion is usually very hard in standard quality control based on genes and UMI counts, I would recommend running a doublet prediction analysis (as DoubletFinder McGinnis et al., 2019 doi:10.1016/j.cels.2019.03.003 ), in order to be sure that the population isolated are indeed basal progenitors and not doublets composed of progenitor and neuron.

6. The sequencing of spatially resolved data is extremely interesting and opens the possibility for addressing the contribution of space to the transcriptome. Here it would be extremely interesting to examine whether the different ganglionic eminences have sharp distinct molecular identities or if there is a gradient of identity between the different eminences.

7. From line 187 on, the supplementary figures references are wrongly numbered.

Reviewer #3:

This study investigates transcriptional diversity of cortical interneuron progenitors in embryonic mice. A major strength of this study is the use of a Nestin GFP line which enables the authors to enrich for progenitors, more so than has been done in previous studies. Using this line the authors uncover previously uncharacterized transcriptional diversity discriminating embryonic domains which produce interneurons at an early stage of development (E12.5). Importantly, the authors make use of smFISH to validate many of these changes. The study provides new lists of transcripts which may highlight diversity of progenitors across domains. Finally the authors perform some analysis at E14.5 to evaluate temporal differences.

While the study has some important novelty and generates evidence of transcriptional diversity, the potential impact of this study is weakened by technical concerns as well as lack of data supporting relevance of observed sub-types. Lineage analysis of at least one potential progenitor subtype would further support the notion that transcriptional heterogeneous progenitors are important (give rise to different fates) as opposed to transcriptional noise.

1. A major conclusion of this paper is that progenitors exhibit transcriptional heterogeneity. When focusing specifically on VZ cells with high Nestin expression (figure 4), how many cells are the authors examining-ie how many cells is this conclusion based upon? Likewise, it would be helpful to better understand how many samples were sequenced to conclude heterogeneity as well as how replicates compare. From the methods it appears that E12.5 is based on 2 Nestin GFP mice. This raises the question as to what extent is this transcriptional diversity reflecting noise of different samples (which could be slightly different stages?) or even technical variation of sequencing.

2. Previous studies have used scSEQ of GE at E13.5 and E14.5 (Fishell lab) and E12.5-E14.5 (Marin lab). The discriminating point here is that the current study uses FACS and a Nestin reporter to isolate and sequence specifically VZ cells. The previous studies also examined VZ cells within their sequenced population, with the Marin study highlighting heterogeneous progenitors evident at E12.5 as well as across MGE and CGE. In the current study the authors conclude their gene expression profiles provide new insight into transcriptional diversity of interneurons. However mainly this study is a list of differential genes (albeit which extends that previously identified). To make this study more impactful it would be helpful if the authors could provide some data such as lineage analysis to support the conclusion that observed diversity impacts fate.

3. Further classification of potential subtypes in a model of some kind would also be helpful-ie are there subtypes which are primarily early born and others that are late born, etc?

4. The authors use Nestin GFP to sort out VZ cells in this study. In Figure 1 they show that Nestin- GFP cells also express markers of newborn neurons (Dcx positive), neurons (Tbr1), and IPS/neurons (Neurod6), particularly in the cortex. The authors should clarify if this is reflecting some perdurance of the GFP into post-mitotic cells? As figure 1 includes WT cells, it is unclear in subsequent figures (such as Figure 3) if they are only focusing on GFP sorted cells or if they are segregating data solely by looking at Nestin transcript. This may be a simple text fix but these missing details make the paper confusing. Due to potential perdurance of GFP protein and even transcript these are not necessarily the same.

5. Analysis of E14.5 is limited-especially as the authors again just sample 1 nestin GFP brain at this stage. The authors conclude that there is more diverse neurogenic and postmitotic expression at E14.5. Additional analysis of cell populations as well as validation would be important to tease out these differences and support this conclusion.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Transcriptional heterogeneity of ventricular zone cells throughout the embryonic mouse forebrain" for further consideration by eLife. Your revised article has been evaluated by Marianne Bronner (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

The reviewers find your paper valuable and suggest several revisions that should improve the presentation of your study. Although Reviewer #1 appears satisfied with the study in the appended review, in the Reviewer Consultations offline, this reviewer agreed with the other two that a deeper analysis of the expression (in situ and immunostaining) at different developmental times (at least the 2 stages you sequenced) would verify whether a specific marker is stable or not in CGE MGE and LGE. Fabp7 characterization in particular would be welcome; if the results show selectivity, you could minimally address the request of Reviewer 4 to highlight a marker.

This reviewer would also like to revisit the previous request to move data concerning the dorsal progenitors to supplementary data so as to further focus on the GE, to emphasize the ganglionic eminences.

Reviewer 2 asks for highlighting the impact of the study with respect to the ganglionic eminences, to better showcase the special focus of your work. This can be done by modifying the summary to better represent content of the paper.

Also recommended is to address in the Discussion the discordance between spatial (in situ) patterns vs sc seq for Id4 and Mest.

Finally, and readily done, is to place in situs next to the relevant UMAPs in the figures; this would aid the reader's navigation through your very detailed and rich data.

Reviewer 4 is new and asks for proof of some genes suggested to be region-specific, such as Igfbp5, and also linking progenitor heterogeneity to proliferation or fate differences or even protein levels, toward using these differences to isolate subsets by FACS. Proof that expression differences link to functional differences, by, for example, targeting candidate genes with a Cre construct, or doing localization/birthdating. Targeting genes experimentally would involve much more extensive work at this point, and Reviewer 1 especially agrees. But all three reviewers believe that because your paper is very descriptive, if you could add some functional evidence, it would increase the paper's impact. At the very least, you could try to do birthdating/cell cycle analyses, and should they work, this would help the significance of the paper. Minimally, the deeper analysis of Fabp7 expression is warranted, and should address the "ask" of Reviewer 4 to provide a marker for further cell isolation.

Reviewer #1:

The authors have resolved the main concerns of the manuscript. I consider that the addition of the new controls improves the quality of the manuscript. The figures showing comparisons with previous studies explain the added value of this study.

Reviewer #3:

The authors have been generally attentive to concerns raised. They have adjusted their analysis of scSEQ datasets, included new validation experiments for E14.5, and made some additional minor changes. They have revised the writing of the manuscript to explain the novelty of their findings relative to other papers (one of my original concerns) as well as to flesh out more details which increases readability. While the paper is still largely descriptive, the authors have generated some valuable new insights for the field. I do have a few remaining minor suggestions.

1. The impact of this study is mainly their analysis of the ganglionic eminences where interneurons are born. In this regard the authors might consider adjusting the summary to reflect that emphasis even more. This may attract more readers to their work and increase impact.

2. The finding that Id4 and Mest spatial patterns by in situ are different than that predicted from the scseq is interesting (but also somewhat concerning for the field). It would be nice if the authors comment on this discordance in the discussion as an important point of consideration for sc seq experiments.

3. Figure 7-supplement 1F labels Nestin expressing cells. The use of red for labeling nestin is confusing here as red also labels CGE.

4. The representation of the in situs adjacent to UMAPS in Figure 8 is really helpful. For this figure I think they should also add in annotations with the UMAPs labeled for each respective region/stage. There are labels for panel A but not the plots below. In addition, I wish the authors would use this approach in prior in situ figures-ie include the relevant UMAPs beside the in situs (for perhaps a subset), as opposed to putting in supplement. It facilitates analysis of their data.

5. Line 366 in discussion (types may be singular)

Reviewer #4:

In this manuscript the authors profile nestin-expressing cells from the 3 ganglionic eminences in E12 and E14 murine embryos by scRNA-seq. This yields several region-specific genes, some of which could serve as a new regional marker, such as Igfbp5 for CGE, but they do not prove this. Most importantly, they do not analyze the region-specific differences across at least the 2 stages they sequenced to verify if these markers are stable at least for these 2 days in development.

I fully agree with the authors that it is important to explore progenitor subtypes at higher resolution, i.e. many cells, to explore heterogeneity. This is also highly relevant for the diverse output from these regions, including the generation of adult neural stem cells. However, the manuscript falls short of linking either progenitor heterogeneity or regional heterogeneity to a specific function – e.g. proliferation differences, fate differences or protein levels to allow subset isolation by FACS. Please provide at least 1 example to showcase that the expression differences link to functional differences in output (in utero-electroporation of a Cre construct targeted to a specific subset) or highlighting by RNAscope and EdU cell cycle monitoring that the marker population has indeed specific cell cycle properties (slow division, for example).

One specific example that raises doubts about the usefulness of the markers proposed is Fabp7, as indeed staining seems to label all cells in the ventricular zone, even though levels are always higher at boundary regions. This would mean that the "absent" expression of Fabp7 may simply be due to the depth of sequencing. Therefore, I suggest the authors perform immunostainings to verify or falsify the usefulness of this "marker".

This leads to my last suggestion namely to focus the analysis to some extent on genes encoding cell surface proteins to allow sorting subsets of progenitors.

eLife. 2022 Feb 17;11:e71864. doi: 10.7554/eLife.71864.sa2

Author response

[Editors’ note: The authors appealed the original decision. What follows is the authors’ response to the first round of review.]

Comments to the Authors:

Reviewer #1:

The manuscript describes the RNA fingerprint of developing telencephalic cells. To focus on progenitors, the authors use a transgenic line that drives the expression of a destabilized-venus GFP under the Nestin promoter. With this strategy, the authors find higher diversity of developing precursors than when they dissociate single cells from unlabeled WT brains because the latter contains higher proportions of postmitotic cells. The main finding is that early VZ precursors are more heterogeneous than inferred from previous studies. In situs in brain sections confirm molecularly distinct clusters of progenitors in medial, lateral, and dorsal VZ. The work is nicely written and organized. The results support the main conclusions. The findings advance the field, and the data will be useful in future research.

We thank the reviewer for their positive comments on our manuscript.

1. The work is nicely written and the conclusions about the heterogeneity of ventral precursors are well supported and important contribution to the field. However, the study of dorsal cell is minor. The analysis of dorsal cells seems to include many dorsal venus-expressing cells that are postmitotic neurons, and the number of analyzed cells from the dorsal origin is small.

My advice is to include the data of the study of dorsal cells as supplementary data, not as principal data.

The reviewer is correct that our main focus is on the ventral telencephalon, and all figures and analysis after Figure 1 are focused on the GEs. However, we feel that presenting all of the dorsal and ventral sequenced cells together in one introductory figure is the most comprehensive way to display our entire dataset, and thus we prefer to keep it in the main text. This approach allows us to compare pan-excitatory vs. inhibitory cell markers to verify the quality of the dataset, as in Figure 1C-D. And it’s unclear how useful it would be to segregate the dorsal cells into their own supplementary figure, as it removes the benefit of comparing/contrasting to the GE dataset. For these reasons, we prefer to keep the dorsal cells in Figure 1 and not move extract them into a supplement figure.

Also, one point of clarification: neurogenesis in the dorsal cortex begins later that in the GEs with only layer VI cortical pyramidal cells beginning to be born around E12.5 (please see comments to Reviewer #2 below). Thus, the majority of all E12.5 dorsal cortical cells (and especially Venus-expressing cells) are still cycling progenitors, likely ~80% based on Di Bella…Arlotta Nature 2021; there are very few dorsal cortex postmitotic neurons at E12.5.

2. On the other hand, I consider that the current supplementary Figure 1 showing the histological expression of the venus reporter should be presented as the main figure.

We have moved the immunostaining of the Nes-dVenus mouse from Figure 1—figure supplement 1 into the main text as new Figure 1A. Now Figure 1—figure supplement 1 is entirely devoted to the flow cytometry gating strategy and plots.

3. The field is rapidly offering new reports describing single cell analysis of telencephalic progenitors. The authors should revise and update the bibliography and discussion accordingly, including: "Single-cell transcriptomics of the early developing mouse cerebral cortex disentangle the spatial and temporal components of neuronal fate acquisition. Moreau et al., Development (2021) 148 (14): dev197962"; "Molecular logic of cellular diversification in the mouse cerebral cortex. Di Bella et la., Nature. 2021 Jul;595(7868):554-559".

We thank the reviewer for pointing out these relevant references and have added them (and several others) in the introduction of the manuscript on p. 3, lines 59-62.

Reviewer #2:

In this study, Lee et al. investigate the molecular diversity of cortex and ganglionic eminences (GE) of the mouse embryo. The authors used single-cell transcriptomics technology to acquire gene expression profiles of telencephalic regions, which they separated by microdissection of the lateral, caudal and medial ganglionic eminences along with the cortex in E12.5 and E14.5 mouse embryos. Using the Nes-dVenus line combined with Fluorescence-activated sorting they could enrich their dataset in ventricular zone cells, mainly neuronal progenitors. With this dataset, they could identify common differentiation lineages between the different GEs. Using the RNAscope technology, the authors validated unreported genes expressed in a specific ganglionic eminence and compartment within the GEs (VZ, SVZ, or MZ). In addition, the manuscript shows a bigger temporal mark on VZ cells compared to post-mitotic neurons between E12.5 and E14.5.

This manuscript presents a complete dataset of single-cell sequencing of ganglionic eminences during development. The effort put in the microdissection of the different GEs, central for the understanding of spatial implication in molecular identity and diversity, along with the validation of newly discovered genes with the RNAscope technique is very appreciated. Nevertheless, the manuscript shows significant weaknesses in the analysis of the transcriptomics data and quality controls. The overall tone is very descriptive such that this manuscript is rather intended for a specialized readership. I have the following comments:

1. In the methods, the authors describe the quality control applied to exclude bad quality cells from the data set (Supplementary Figure 2). At line 384, the authors say: "we first removed cells that had unique molecular identifies (UMI) counts over 4500 or less than 200." From what is shown in supplementary figure 2 we observe that this filtering is done on the number of expressed genes and not on the UMI counts. In addition, in the violin plot of a number of genes per cell, the distribution looks binomial with a population of cells expressing around 500 genes (highlighted in orange) and another population expressing around 3000 genes per cell (highlighted in blue below). The population with a low number of genes expressed should be excluded for further analysis because it is likely that it represents 10x GEMs containing ambient RNA or dying cells instead of GEMS containing a whole cell. This is supported by the post-filtering graph at the bottom of the figure (% of Mitochondrion vs Counts Depth) where, strikingly for the cortex, we observe a population of cells with low % of mitochondrial RNA and high Count depth and a band of cells formed by low Count Depth cells with a variety of mitochondrial RNA (highlighted in red). I strongly recommend the authors filter out the low number of genes population (in orange) by setting a threshold of around 1500 genes per cell which will probably clean the populations with low count depth and thus avoid including bad quality GEMs for further analysis.

We thank the reviewer for carefully exploring our dataset and pointing out the bimodal distributions within our graphs in Figure 1—figure supplement 2. With our goal of maximizing cell numbers, it’s possible we under-filtered our dataset. ~84,000 E12.5 cells successfully passed through the Cell Ranger pipeline (which was depicted in the chart in old Figure 1). Then the combination of removing cells with > 4500 or < 200 genes per cell, cells with high mitochondria transcripts, and cells with hemoglobin subunit Hbb from WT mice (all described in the original Methods) resulted in ~43,000 cells being analyzed in the original manuscript.

As the reviewer suggested, we increased the threshold for number of genes from > 200 to > 1,500 per cell to filter out potential ambient RNA, dying cells, etc. with low count depth that the reviewer noted (Figure 1—figure supplement 3b). (Relatedly, both DAPI and DRAQ5 were added to the single cell suspension prior to flow cytometry so that we could collect live cells (DRAQ5+/DAPI-) and remove dead/dying cells (DRAQ5+/DAPI+), so there should be very few dead/dying cells in our analysis.) We also removed those predicted doublets that had DoubletFinder scores greater than 0.3 from further downstream analysis (see point #5 below). Using the > 1,500 gene/cell cutoff, this reduced our final counts from ~43,000 E12.5 cells to ~36,500 E12.5 cells. (We performed identical analysis on the E14.5 population in Figure 7—figure supplement 1, which resulted in ~24,000 high quality E14.5 cells). As you can see from the figure below, we still observe a clean separation of the CTX and MGE cells whereas the LGE and CGE populations are still largely intermingled; the new dotplot looks very similar to the original plot (Author response image 1 and Figure 1—figure supplement 3a). Furthermore, the top 20 DEGs between the regions are very similar to our original analysis (Figure 1—figure supplement 3b). Thus, we strongly believe that no significant biological misinterpretation was made using the previous > 200 gene/cell cutoff, and that possible contamination from ambient mRNA or dead/dying cell datapoints did not affect any data interpretation or conclusions from our dataset. This notion is strengthened by the fact that the overwhelming majority of genes characterized by our in situ hybridizations were in agreement with their predicted expression profiles from the scRNAseq data.

Author response image 1.

Author response image 1.

Open in a new tab

Using the more stringent > 1,500 gene per cell threshold with the ~36,500 E12.5 cells, we have updated the dotplots and heatmaps in Figures 1 and 2, as well as related Figure 1—figure supplement 3 and Figure 2—figure supplement 3. We have also updated Figure 1—figure supplement 2B and Figure 7—figure supplement 1 using the > 1,500 gene per cell threshold.Note that these low genes per cell datapoints were already excluded from the Nestin- and Dcx-enriched datasets in Figures 3, 4, 7, Figure 6—figure supplement 1 and related analysis because they did not reach the gene expression threshold anyway; thus, this portion of our analysis in this manuscript was unaffected by this change in gene per cell threshold.

Again, we thank the reviewer for this observation and now have more confidence in the quality of cells in our dataset. And importantly, there are no significant changes in the overall conclusions and gene expression profiles originally detected.

2. At the lines 90-92 of the manuscript, the authors write: "Conversely, cortical neurogenesis does not begin until ~E14 (Noctor et al., 2004; Sessa et al., 2008; Tyler et al., 2015), so the majority of E12.5 cortical cells are mainly VZ neural progenitors, specifically RGCs." Indeed, at E12.5 the ventricular zone is the predominant compartment of the developing pallium and most of the cells present at this stage are radial glial cells, nevertheless, asymmetric division of radial glia to produce neurons has been reported as soon as E11.5 in the mouse (Kwan et al. Development, 2012, doi: 10.1242/dev.069963), explaining the small but present proportion of neurons found in the dataset.

The reviewer is correct that there is growing evidence that the earliest cortical neurons (layer VI) are born prior to E14, likely starting to be generated ~E11.5-E12. My lab has performed EdU labeling at E11.5 and found VERY few cortical neurons that are EdU-labeled in the adult, but we do see subplate cells that are EdU+. And based on the recent Di Bella…Arlotta 2021 Nature paper, it appears that ~80% of E12.5 dorsal cortical cells are classified as APs or intermediate progenitors whereas < 5% of cells are classified as ‘immature neurons’ (see Extended Data Figure 1). Based on these findings, we have modified the sentence to more accurately reflect the timing of cortical development, p.4 line 94-96: “Conversely, the vast majority of E12.5 dorsal cortical cells are VZ neural progenitors (specifically RGCs), with a small number of layer VI projection neurons starting to emerge at this time (Kwan et al., 2012; Di Bella et al., 2021).”

3. In line 258 the authors write: "To compare the E12.5 and E14.5 cells, we integrated these datasets together". It is unclear whether the two datasets were just merged in a common Seurat object or if they were integrated with SCTransform normalization and integrated with the Seurat v3 pipeline. I would suggest that this sort of integration not be applied when combining datasets generated in such similar conditions because it can remove the differences between datasets, in this case, age of collection.

We apologize for not being clearer in the main text and Methods about how we merged these two datasets. For the dataset comparing the E12.5 and E14.5 cells in Figure 7, standard Seurat integration workflow was followed performing the log normalization method as opposed to SCTransform, using E12.5 and E14.5 as a reference list. To clarify this point, we have updated the Methods section with more details about the normalization methods we used for each analysis in the ‘Single-cell RNA sequencing Analysis’ section (p. 18-19).

4. In Figure 7, the authors nicely show a big overlap and few differentially expressed genes in post-mitotic cells between E12.5 and E14.5. It would be very interesting to know if this overlap is shared for different areas of the telencephalon such as the pallium or if it is specific to the subpallium.

We did not generate scRNAseq experiments on the E14.5 pallium/CTX, so we are unable to perform the requested comparison. As the reviewer accurately noted in the above comment, there is a very small number of postmitotic cortical cells at E12.5, while there would be a significant expansion of postmitotic cells at E14.5. I do not know if our dataset contains enough postmitotic E12.5 cortical cells to definitely state one way or another if this overlap would be subpallium-specific or if it applied more generally to other brain regions. Since the cortex is producing primarily layer VI cells at E12.5 and layer III-IV cells at E14.5, I would not be surprised if there were more distinct genetic profiles of postmitotic cells at these ages. There are more comprehensive dorsal cortex datasets available to better answer this specific question, such as the one generated in the Di Bella…Arlotta 2021 Nature paper.

5. In order to isolate basal progenitor cells, the authors selected cells expressing both neuronal markers such as Dcx and cycling marker as Mki67. Since in the 10x experiments it is frequent to have doublets and their exclusion is usually very hard in standard quality control based on genes and UMI counts, I would recommend running a doublet prediction analysis (as DoubletFinder McGinnis et al., 2019 doi:10.1016/j.cels.2019.03.003 ), in order to be sure that the population isolated are indeed basal progenitors and not doublets composed of progenitor and neuron.

We thank the reviewer for pointing out the limitation of 10X experiments with regards to dealing with doublets. Seurat attempts to identify and filter out suspected doublets with a clear outlier number of UMIs and genes detected, which we have incorporated to our analyses. However, this pipeline is not perfect in identifying and removing doublets. As the reviewer suggested, we employed the doublet prediction algorithm DoubletFinder to identify possible doublets in our dataset (scores greater than 0.3). The resulting DoubletFinder plot is shown in author response image 2. As you can see in Author response image 2, DoubletFinder identified a group of cells in the middle of the plot (cluster #11) as high potential doublets. Notably, this was also the only cluster that had a pretty even mix of MGE/CGE/LGE-derived cells. Importantly, there was only a smattering of potential doublets scattered amongst the other clusters, indicating that there is little reason for believing that there are any potential doublet issues with the overwhelming majority of cells outside of cluster #11. Thus, there is no evidence that the basal progenitors are ‘false positives’ consisting of a progenitor and neuron doublet.

Author response image 2.

Author response image 2.

Open in a new tab

Based on this observation, we reran the genetic analysis in Figure 3 and all subsequent figures after removing predicted doublets. We have generated new dotplots with these predicted doublets removed for Figures 3, 4, 7 and Figure 5—figure supplement 1, Figure 6—figure supplement 1, Figure 6—figure supplement 2 and Figure 7—figure supplement 2. We also generated new heatmaps in Figures 3, 4, 7 and Figure 6—figure supplement 1 after removal of these predicted doublets. Importantly, these adjustments did not generate any significant changes in the data. In some cases, the gene order may have jumped around a little (e.g., the second most enriched gene is now the fifth, or vice versa). All conclusions we originally made are still supported by the data, which is expected as we only removed ~4% of all cells/doublets that likely had minimal influence in the overall genetic comparisons between regions, cell types and time points. We thank the reviewer for bringing Doubletfinder to our attention to help clean up our data, and we plan to use it for future experiments to ensure that we remove all predicted doublets from our dataset. We have updated the Results text on p. 5 and Methods text in the ‘Single-cell RNA Sequencing Analysis’ section on p. 18-19 to reflect these changes.

6. The sequencing of spatially resolved data is extremely interesting and opens the possibility for addressing the contribution of space to the transcriptome. Here it would be extremely interesting to examine whether the different ganglionic eminences have sharp distinct molecular identities or if there is a gradient of identity between the different eminences.

Based on our in situ data, we see both types of expression patterns: some genes display sharp, distinct boundaries that seem to adhere to the GE boundaries, such as Mt3 and Rbp1 in the MGE/LGE boundary (Figure 5G). Other genes display gradient expression profiles that may blur or ignore the GE boundaries such as Id4 and Nkx6.2 at the MGE/LGE boundary (Figure 5F, H). Some genes even display sharp boundaries within a specific GE, such as Rbp1 and Igfbp5 in the dCGE and vCGE, respectively (Figure 5G). The idea of sharp molecular boundaries vs. gradients of gene expression has important implications in gene function throughout developmental biology, with gradients of various signaling factors playing critical roles in initial patterning and fate determination throughout the early stages of neurogenesis. We agree with the reviewer that these patterns are extremely interesting, and the function of various genes in specific brain regions with either gradient and/or sharp boundaries warrant future exploration. We have added a sentence in the discussion noting the observation of both sharp boundaries and gradients, p. 14 lines 338-339.

7. From line 187 on, the supplementary figures references are wrongly numbered.

Thank you for pointing this out, we apologize for these errors and have corrected the Supplement Figure numbers in the revised submission.

Reviewer #3:

This study investigates transcriptional diversity of cortical interneuron progenitors in embryonic mice. A major strength of this study is the use of a Nestin GFP line which enables the authors to enrich for progenitors, more so than has been done in previous studies. Using this line the authors uncover previously uncharacterized transcriptional diversity discriminating embryonic domains which produce interneurons at an early stage of development (E12.5). Importantly, the authors make use of smFISH to validate many of these changes. The study provides new lists of transcripts which may highlight diversity of progenitors across domains. Finally the authors perform some analysis at E14.5 to evaluate temporal differences.

While the study has some important novelty and generates evidence of transcriptional diversity, the potential impact of this study is weakened by technical concerns as well as lack of data supporting relevance of observed sub-types. Lineage analysis of at least one potential progenitor subtype would further support the notion that transcriptional heterogeneous progenitors are important (give rise to different fates) as opposed to transcriptional noise.

Please see responses to Reviewer #2 above and specific comments below regarding the technical concerns. And please see our response to point #2 below regarding lineage analysis.

1. A major conclusion of this paper is that progenitors exhibit transcriptional heterogeneity. When focusing specifically on VZ cells with high Nestin expression (figure 4), how many cells are the authors examining-ie how many cells is this conclusion based upon? Likewise, it would be helpful to better understand how many samples were sequenced to conclude heterogeneity as well as how replicates compare. From the methods it appears that E12.5 is based on 2 Nestin GFP mice. This raises the question as to what extent is this transcriptional diversity reflecting noise of different samples (which could be slightly different stages?) or even technical variation of sequencing.

To answer the reviewer’s first point, the number of high Nestin-expressing cells from Figure 4 is 9,308 (with 3036 from LGE, 3890 from MGE and 2382 from CGE). This is the number reported in Sup Figure 5 where we compare our study to the Fishell & Marin studies. Similarly, the number of high DCX-expressing cells from Sup Figure 9 is 12,307 (3900 LGE, 3902 MGE, 4505 CGE). We have included these cell numbers in the main text on lines p.8 line 194-95 and p.10 lines 243-44.

We noted the total number of sequenced cells and replicates in the Methods, but it certainly could have been presented more clearly. In total, our manuscript analyzed ~60,000 high quality E12.5 and E14.5 cells which can be broken down as follows: (A) 3 scRNAseq reactions for each E12.5 MGE/LGE/CGE (1 WT and 2 Nes-dVenus) resulting in 9,275-12,052 cells per GE, (B) 2 scRNAseq reactions for E12.5 cortex (1 WT and 1 Nes-dVenus) resulting in ~4,785 CTX cells, and (C) 2 scRNAseq reactions for each E14.5 MGE/LGE/CGE (1 WT and 1 Nes-dVenus) resulting in 7,171-8,658 cells per GE. For each reaction, tissue from a minimum of 4 embryos were combined. We have reworded the Methods section to more clearly state this breakdown of cell numbers in the ‘Single-cell RNA sequencing Analysis’ section, p. 19.

Author response image 3 displays the 3 E12.5 replicates (1 WT and 2 Nes-dVenus) overlaid on top of each other: on the left are the 3 GEs combined (comparable to Figure 2A in the manuscript), and on the right is the MGE extracted out (comparable to Figure 2B in the manuscript). As you can see, the replicates have a high degree of overlap, with very little evidence for diversity being due to differences in replicates or technical variation of sequencing. The main difference is the bias for WT-derived cells to be on the right side of the figure (DCX+-region) and Nes-dVenus on the left side (Nes+-region), as expected. Similar results are observed when extracting out the LGE and CGE.

Author response image 3.

Author response image 3.

Open in a new tab

For comparison to other recent high-quality single cell sequencing papers of embryonic mouse forebrain: (1) Di Bella…Arlotta 2021 Nature paper performed a comprehensive scRNAseq analysis of mouse embryonic dorsal cortex every day from E10.5-E17.5. This study consisted of 1 replicate on all cortical embryonic timepoints from E10.5-E17.5, with a range of ~3,000-11,600 cells per timepoint (mean ~7,000/timepoint). (2) Moreau…Causeret Development 2021 paper analyzed a total of 4,225 cells from 1 replicate near the pallial/subpallial boundary. (3) Telley…Jabaudon Science 2019 paper analyzed a total of 2,756 cortical cells spread across 12 different embryonic cell labeling and collection conditions, 1 replicate each, which averages out to ~230 cells/timepoint. (4) Preissel…Ren Nat Neuro 2018 performed scATACseq on ~12.7K cells from 7 different timepoints ranging from E11.5 to P0, with 2 replicates/timepoint, which averages out to ~900 cells per replicate (or ~1800/timepoint, see Sup Table 2 in this paper). (5) Mi…Marin Nature 2018 paper sequenced a total of 2,000 cells from 3 different brain regions at 2 different embryonic timepoints (1 replicate each), which averages out to ~330 cells/timepoint. (6) Mayer…Fishell Nature 2018 paper performed scRNAseq on a total of 5,600-8,500 cells from the MGE, LGE and CGE, with multiple replicates of each region.Thus, we believe that our significant increase cell number (and thus power) compared to many other single cell sequencing studies, combined with our specific enrichment of VZ cells with the Nes-dVenus mouse, is why we detect many novel gene expression profiles and candidate progenitor subgroups that could not have been detected in other studies. Especially in regards to VZ cells/RGCs which are underrepresented in these other studies.

2. Previous studies have used scSEQ of GE at E13.5 and E14.5 (Fishell lab) and E12.5-E14.5 (Marin lab). The discriminating point here is that the current study uses FACS and a Nestin reporter to isolate and sequence specifically VZ cells. The previous studies also examined VZ cells within their sequenced population, with the Marin study highlighting heterogeneous progenitors evident at E12.5 as well as across MGE and CGE. In the current study the authors conclude their gene expression profiles provide new insight into transcriptional diversity of interneurons. However mainly this study is a list of differential genes (albeit which extends that previously identified). To make this study more impactful it would be helpful if the authors could provide some data such as lineage analysis to support the conclusion that observed diversity impacts fate.

The reviewer is correct that the previous studies from the Fishell (Mayer…Fishell 2018) and Marin lab (Mi…Marin 2018) performed scRNAseq of the mouse ventral telencephalon. A couple of notes on the Marin paper: First, as stated above, they sequenced a total of 2,003 cells from the vMGE, dMGE and CGE at E12.5 and E14.5 (six different data points), around half of which were classified as ‘progenitors’ (VZ and SVZ cells). This averages out to ~150 progenitor cells per region per age, which is significantly underpowered for making strong conclusions about distinct VZ and SVZ progenitor cell types. And based on our analysis, the entire Marin dataset contains only 147 high Nestin expressing/VZ cells (our Sup Figure 5). Additionally, they do not perform any confirmation of these gene expression profiles (in situs, immunostaining, etc.).

Comparatively, our dataset averages ~9,300 cells per GE per age, with the vast majority of collected cells being VZ or SVZ progenitors because of using the Nes-dVenus mouse. And our in situs confirm spatially and temporally restricted expression patterns of numerous genes from the scRNAseq dataset. For these reasons, we argue that our dataset does more than ‘…extend previously identified [genes]’. The Marin study was significantly underpowered to make any definitive claims about differential gene expression at these timepoints, especially regarding cycling VZ and SVZ cells. And the Fishell study does not closely explore these cycling cell populations; their focus is more on the developmental progression of postmitotic MGE cells. Thus, we argue that our study has uncovered an entirely new genetic organization and diversity in the embryonic GEs, and particularly in cycling VZ progenitors, that has not been previously identified. To our knowledge, the majority of gene expression patterns confirmed by our in situs are novel.

The author raises in interesting and important point about linking gene expression patterns to specific GABAergic cell fates. This is an important and challenging question to tackle throughout developmental biology. We are confident that some of these genes we characterize here likely play important roles in fate determination of specific interneuron subtypes. For example, it has been shown that there is a bias of SST+ cells born from the dorsal MGE and PV+ cells born from the ventral MGE (Wonders…Anderson 2008; Sousa…Fishell 2009; Inan…Anderson 2012). In particular, there is a strong bias for Nkx6.2 lineage cells from the dMGE to generate SST+ cells (Figure 5H; Sousa…Fishell 2009). However, almost nothing is known about the relationship between spatial domains and cell fate in the LGE and CGE.

For the genes with the most intriguing, spatially refined expression profiles (such as Igfbp5, Wnt7b Rmst and Cxcl14 in the CGE), we searched the literature for Cre-drivers that would allow us to fate-map these populations in the CGE, but could not find live mice with these drivers. We also did not find live floxed alleles for the relevant genes of interest that would allow us to study their function. A Cxcl14 floxed mouse was generated in 2010, but it’s unclear if this line is still maintained as a live colony. Plus, the lack of a specific CGE-Cre driver makes this approach challenging. In utero electroporation, which our lab has extensive experience with, do not provide the necessary spatial targeting resolution to label or manipulate subdomains of GEs. Thus, without proper genetic tools, it is extremely challenging (if not impossible) to definitively target specific subpopulations of the mouse GEs and correlate these populations with mature interneuron fate. We are planning on generating Cre-drivers for 1-2 of the most intriguing candidates we have uncovered, but this will take ~6-12 months to generate the mice, then cross them to appropriate reporters, etc.; this is a much longer-term project. Therefore, while we agree with the reviewer’s interest in lineage tracing, this is beyond the scope of the current manuscript because we currently lack the genetic tools to properly characterize embryonic gene expression with mature cell fate in a timely manner.

3. Further classification of potential subtypes in a model of some kind would also be helpful-ie are there subtypes which are primarily early born and others that are late born, etc?

In the MGE, there is a bias for SST+ interneurons being born during the early stages of neurogenesis (~E11.5-E15) while PV+ interneurons are born throughout the neurogenic period (~E12-P0), see Bandler…Fishell 2017 review and references therein. In the LGE, striosomal spiny projection neurons appear to be born before matrix spiny projection neurons (Kelly…Huang Neuron 2018). To our knowledge, almost nothing is known about the temporal birthdates of specific CGE-derived interneuron subtypes. Similarly, there are spatial domains in the MGE that preferentially give rise to specific subtypes, with a bias of SST+ interneurons from the dMGE and PV+ interneurons from the vMGE (see above and Bandler…Fishell 2017 review). We are not aware of any studies relating spatial microdomains of the LGE or CGE to specific GABAergic cell types derived from these regions.

This knowledge allows us to make some predictions about specific spatial and temporal gene patterns in relation to cell fate in the MGE (but much less so in the LGE and CGE). For example, vMGE-enriched genes such as Bcan, Fabp7 and Nr2f1 (Figures 5H and 6B) are more likely to play a role in PV+ fate whereas dMGE-enriched genes such as Nkx6.2, Id4 and Mpped2 (Figures 5F, 5H and 6E) may regulate production of SST+ cells. The downregulation of 472 genes from E12.5 MGE to E14.5 MGE is correlative for these genes being more involved in production of SST+ interneurons, and conversely upregulation of 522 genes over time could indicate their association with PV+ interneurons (log-fold change threshold > 0.25). Further studies are needed to characterize the functional role of these candidate genes in MGE-derived fate determination. And while similar spatial and temporal gene expression patterns exist in the LGE and CGE, it is nearly impossible to link these profiles to mature cell types due to our lack of understanding in this organization and lack of genetic tools to target these populations, as noted above.

4. The authors use Nestin GFP to sort out VZ cells in this study. In Figure 1 they show that Nestin- GFP cells also express markers of newborn neurons (Dcx positive), neurons (Tbr1), and IPS/neurons (Neurod6), particularly in the cortex. The authors should clarify if this is reflecting some perdurance of the GFP into post-mitotic cells? As figure 1 includes WT cells, it is unclear in subsequent figures (such as Figure 3) if they are only focusing on GFP sorted cells or if they are segregating data solely by looking at Nestin transcript. This may be a simple text fix but these missing details make the paper confusing. Due to potential perdurance of GFP protein and even transcript these are not necessarily the same.

We agree with the reviewer’s observation that the perdurance of GFP into SVZ/postmitotic cells is the most likely reason why some GFP+ cells may express more mature markers such as Dcx, Tbr1 and Neurod6. The picture from the original Nestin-dVenus mouse paper (Sunabori…Okano J. Cell Science 2008) shows the destabilization motif significantly reduces the perdurance of GFP after the cells have left the VZ and presumably downregulated Nestin mRNA and Nestin protein. Their figure looks very similar to what we saw (new Figure 1A).

Even with the destabilized GFP, there will still be a delay from when the Nestin mRNA is downregulated vs. when the GFP protein is eliminated. Also, flow cytometry is more sensitive than immunohistochemistry to detect fluorescence, especially in live cells vs. fixed tissue. It’s likely that by sorting the cells, we are collecting cells with lower GFP levels than can be detected via immunohistochemistry, which would likely represent SVZ cells and newly born neurons. Additionally, while we always used WT controls to establish the proper gating (Sup Figure 1), we did push the GFP+ gate very close to the border of the WT cells to maximize harvesting both dim and strong GFP+ cells. Thus, we likely collected GFP+ cells with low levels of GFP protein that have already progressed out of the VZ/Nestin+ stage towards a SVZ/postmitotic fate.

Regarding the reviewer’s second point, we apologize for not being clearer in the text regarding the WT vs. Nes-dVenus cells. Cells collected from both WT and Nes-dVenus mice are included in all of the figures and analysis throughout the manuscript. Several times we note whether the cells are derived from WT or Nes-dVenus mice to highlight the enrichment of Nestin+/VZ cells from the Nes-dVenus mice (such as Figures2A, 7A, Sup Figure 4 and Sup Figure 12).

To clarify this point, we have added the following line in the main text, p. 5 line 109: Cells harvested from both WT and Nes-dVenus mice were used for all subsequent analysis.

5. Analysis of E14.5 is limited-especially as the authors again just sample 1 nestin GFP brain at this stage. The authors conclude that there is more diverse neurogenic and postmitotic expression at E14.5. Additional analysis of cell populations as well as validation would be important to tease out these differences and support this conclusion.

We performed 2 sequencing regions for each E14.5 GE, 1 WT and 1 Nes-dVenus, totaling ~7,100-8,650 cells per GE. As detailed in response to Reviewer #3 point #1 above, our E14.5 dataset comprises more cells and more (or equal) replicates than many recently published high-profile scRNAseq manuscripts.

To the reviewer’s second point, we have performed additional RNAscope HiPlexUP in situ hybridizations at E14.5 to (1) validate our scRNAseq data at this timepoint, and (2) confirm differential gene expression patterns between these two ages that were predicted from the scRNAseq data. Many of the genes displayed the predicted changes in gene expression patterns from the scRNAseq data, such as downregulation of Sparc1, Sfrp1 and Id4 in E14.5 VZ cells and upregulation of Mir124a-1hg and Gucy1a1 at E14.5. This new data highlights the developmental change in transcription profiles of neural progenitors over time (specifically VZ and SVZ cells), which could have important implications in differential neuronal fates related to birthdates. These new images are now included in new Figure 8 and the corresponding text in the Results section on p. 13. We’ve also included a section in the Discussion relating these observations to the hypothesis proposed in Telley…Jabaudon 2019 that transcriptional progression of cortical APs is critical for progression of cell fate, on p. 15.

[Editors’ note: what follows is the authors’ response to the second round of review.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

The reviewers find your paper valuable and suggest several revisions that should improve the presentation of your study.

Although Reviewer #1 appears satisfied with the study in the appended review, in the Reviewer Consultations offline, this reviewer agreed with the other two that a deeper analysis of the expression (in situ and immunostaining) at different developmental times (at least the 2 stages you sequenced) would verify whether a specific marker is stable or not in CGE MGE and LGE.

Figure 8 in our previous submission does directly compare in situ expression of 9 different genes between E12.5 and E14.5 in the MGE, LGE and CGE. In this case, we were focusing on genes that demonstrated different expression patterns between the 2 ages, many of which we highlighted in this Figure 8 and the corresponding last paragraph of the Results section. In this latest resubmission, we have added a Supplement figure to highlight some genes that display stable expression between E12.5 and E14.5 in the GEs.

Fabp7 characterization in particular would be welcome; if the results show selectivity, you could minimally address the request of Reviewer 4 to highlight a marker.

We don’t understand Reviewer #4’s point about Fabp7 (see below), the scRNA-seq data and in situs are both in agreement. Fabp7 is clearly not expressed in all VZ cells as Reviewer #4 claims, see green channel in in situ and corresponding UMAP plot in Figure 5E. We have added a Fabp7/BLBP immunostaining image below to confirm that the protein expression aligns with the in situs.

This reviewer would also like to revisit the previous request to move data concerning the dorsal progenitors to supplementary data so as to further focus on the GE, to emphasize the ganglionic eminences.

Reviewer 2 asks for highlighting the impact of the study with respect to the ganglionic eminences, to better showcase the special focus of your work. This can be done by modifying the summary to better represent content of the paper.

We have modified the Title, Summary and Introduction to better highlight that our study is truly focused on the GEs per Reviewer #2’s suggestion. We don’t see in the specific Reviewer comments below about moving cortical progenitor data from Figure 1 to the supplementary data. As we stated in our previous Response to Reviewer’s, we believe that presenting all of the dorsal and ventral sequenced cells together in one introductory figure is the most comprehensive way to display our entire dataset, and thus we prefer to keep it in Figure 1. This approach allows us to compare pan-excitatory vs. inhibitory cell markers to verify the quality of the dataset (Figure 1C-D), an important initial verification to confirm the integrity of our dataset. And it’s unclear how useful it would be to segregate the dorsal cells into their own supplementary figure, as it removes the benefit of comparing/contrasting to the GE dataset. We would like to follow the common practice of snRNAseq studies, presenting the entire comprehensive dataset in the first figure and then subdividing the dataset into more refined subsets in future figures. For these reasons, we prefer to keep the dorsal cells in Figure 1 and not move extract them into a supplement figure.

Also recommended is to address in the Discussion the discordance between spatial (in situ) patterns vs sc seq for Id4 and Mest.

Our original statement of Id4 and Mest as ‘divergent’ mischaracterizes these expression patterns, and we apologize for this confusion. The relevant portion of the text and figure address expression in VZ cells. Based on our in situs, Id4 is expressed throughout the VZ of all GEs whereas Mest displays a more restricted expression pattern in VZ cells. There are differences, but the profiles are not divergent. However, Id4 and Mest do display divergent (complementary?) expression patterns in the SVZ/MZ GE regions. This data is NOT shown in the adjacent UMAP dotplot because we’ve selected for high-Nestin VZ cells in this case. We have clarified our description of Id4 and Mest in the text to more accurately reflect these observations.

Finally, and readily done, is to place in situs next to the relevant UMAPs in the figures; this would aid the reader's navigation through your very detailed and rich data.

We thank the reviewer for this suggestion and agree that having the UMAP plots adjacent to the in situs allows for better visual comparison between scRNA and in situ data. We have updated Figure 5, Figure 5—figure supplement 1, Figure 6 and Figure 6—figure supplement 2 so that the UMAP plots are adjacent to the in situs, with the figure legends and text references updated accordingly.

Reviewer 4 is new and asks for proof of some genes suggested to be region-specific, such as Igfbp5, and also linking progenitor heterogeneity to proliferation or fate differences or even protein levels, toward using these differences to isolate subsets by FACS. Proof that expression differences link to functional differences, by, for example, targeting candidate genes with a Cre construct, or doing localization/birthdating. Targeting genes experimentally would involve much more extensive work at this point, and Reviewer 1 especially agrees. But all three reviewers believe that because your paper is very descriptive, if you could add some functional evidence, it would increase the paper's impact. At the very least, you could try to do birthdating/cell cycle analyses, and should they work, this would help the significance of the paper. Minimally, the deeper analysis of Fabp7 expression is warranted, and should address the "ask" of Reviewer 4 to provide a marker for further cell isolation.

We discuss Reviewer #4’s concerns about these issues in more detail below.

Reviewer #3:

The authors have been generally attentive to concerns raised. They have adjusted their analysis of scSEQ datasets, included new validation experiments for E14.5, and made some additional minor changes. They have revised the writing of the manuscript to explain the novelty of their findings relative to other papers (one of my original concerns) as well as to flesh out more details which increases readability. While the paper is still largely descriptive, the authors have generated some valuable new insights for the field. I do have a few remaining minor suggestions.

1. The impact of this study is mainly their analysis of the ganglionic eminences where interneurons are born. In this regard the authors might consider adjusting the summary to reflect that emphasis even more. This may attract more readers to their work and increase impact.

We have modified the Title to read ‘Transcriptional heterogeneity of ventricular zone cells in the ganglionic eminences of the mouse forebrain’. And we have modified several sentences in the Summary and Introduction (highlighted in red) to emphasize that the primary focus of our study is on the ganglionic eminences.

2. The finding that Id4 and Mest spatial patterns by in situ are different than that predicted from the scseq is interesting (but also somewhat concerning for the field). It would be nice if the authors comment on this discordance in the discussion as an important point of consideration for sc seq experiments.

Our original statement of Id4 and Mest as ‘divergent’ mischaracterizes these expression patterns, and we apologize for this confusion. The relevant portion of the text and Figure 5 specifically address expression in VZ cells. Based on our in situs, Id4 is expressed throughout the VZ of all GEs whereas Mest displays a more restricted expression pattern in VZ cells. There are differences, but their profiles are not ‘divergent’. However, Id4 and Mest do display divergent/complementary expression patterns in the SVZ/MZ GE regions. This data is NOT shown in the adjacent UMAP dotplot because we’ve selected for high-Nestin VZ cells in this case. We have clarified our description of Id4 and Mest in the text (p. 9, lines 230-234) to accurately reflect these observations.

That being said, there is always going to be some variability between scRNA-seq and in situ datasets due to many reasons (sampling variability in scRNAseq data, different splice variants that can’t be detected with standard 3’ dropseq reactions like 10X Genomics, cell viability with scRNAseq, lack of spatial information in scRNAseq, etc.). The reviewer is correct to be somewhat skeptical of scRNAseq datasets that don’t perform some confirmation (either via in situs, immunostaining, Western blots, etc.). That’s why confirming gene expression from scRNAseq data is so critically important, and why we utilized the HiPlexUp platform to characterize 30+ genes on one brain section. The fact that the overwhelming majority of genes in our study displayed similar expression profiles between our scRNAseq and in situs instills great confidence in the accuracy of our dataset. We have added a sentence in the discussion (p. 13, lines 354-356.) noting the importance of confirming predicted gene expression patterns from scRNAseq datasets.

3. Figure 7-supplement 1F labels Nestin expressing cells. The use of red for labeling nestin is confusing here as red also labels CGE.

We have adjusted this figure, both the ‘Nestin’ and ‘Dcx’ text in panels F and G are black to eliminate any confusion with the color scheme in other panels.

4. The representation of the in situs adjacent to UMAPS in Figure 8 is really helpful. For this figure I think they should also add in annotations with the UMAPs labeled for each respective region/stage. There are labels for panel A but not the plots below. In addition, I wish the authors would use this approach in prior in situ figures-ie include the relevant UMAPs beside the in situs (for perhaps a subset), as opposed to putting in supplement. It facilitates analysis of their data.

We thank the reviewer for this suggestion and agree that having the UMAP plots adjacent to the in situs allows for better visual comparison between scRNA and in situ data. We have updated Figure 5, Figure 5—figure supplement 1, Figure 6 and Figure 6—figure supplement 2 so that the UMAP plots are now adjacent to the in situs, with the figure legends and text references updated accordingly. We hope this helps readers compare the scRNAseq and in situ data.

The dotplot in Figure 8A displays the Nestin-enriched population (from Figure 7E). This subpopulation clearly has distinct E12.5 and E14.5 clouds, which is why they are highlighted in that dotplot. The dotplots in Figures 8B-D contain ALL cells from E12.5 and E14.5 (from Figure 7A-D), which do not have clear, distinct E12.5 and E14.5 populations like in the Nestin-enriched population. That is why the E12.5 and E14.5 ovals are only shown in Figure 8A and not in Figures 8B-D.

5. Line 366 in discussion (types may be singular)

We have adjusted this line to read ‘…GABAergic neuronal cell type patterning…’.

Reviewer #4:

In this manuscript the authors profile nestin-expressing cells from the 3 ganglionic eminences in E12 and E14 murine embryos by scRNA-seq. This yields several region-specific genes, some of which could serve as a new regional marker, such as Igfbp5 for CGE, but they do not prove this. Most importantly, they do not analyze the region-specific differences across at least the 2 stages they sequenced to verify if these markers are stable at least for these 2 days in development.

Figure 8 in our revised submission did directly compare expression profiles of 9 different genes between E12.5 and E14.5 in the MGE, LGE and CGE. In this figure, we focused on genes which had different expression patterns between the E12.5 and E14.5, which we described in detail in the last paragraph of the Results (p. 13-4, lines 332-347). To assess the temporal stability of some markers as suggested by this reviewer, we have added a new Supplementary figure (Figure 8—figure supplement 1) that depicts numerous genes with consistent expression patterns between E12.5 and E14.5. We also added a sentence in the results (p. 13, lines 342-344) referring to the new Supplementary figure.

I fully agree with the authors that it is important to explore progenitor subtypes at higher resolution, i.e. many cells, to explore heterogeneity. This is also highly relevant for the diverse output from these regions, including the generation of adult neural stem cells. However, the manuscript falls short of linking either progenitor heterogeneity or regional heterogeneity to a specific function – e.g. proliferation differences, fate differences or protein levels to allow subset isolation by FACS. Please provide at least 1 example to showcase that the expression differences link to functional differences in output (in utero-electroporation of a Cre construct targeted to a specific subset)

This question is quite similar to Reviewer #3, point #2 in our original submission, which recommended additional data linking specific genes to cell fate. Relating gene function to cell fate is an important and challenging question to tackle throughout developmental biology. We are confident that some of the genes characterized in our study likely play critical roles in cell fate, maturation and/or function, as this was a large motivation for our study.

In order to test gene function, we need the necessary genetic tools which unfortunately are not currently available for our most intriguing candidates. Our main interest is in CGE-restricted genes, as the genetic logic of CGE-derived cells is poorly understood. While the reviewer’s suggestion to introduce Cre into a targeted brain region via IUE is feasible (the Petros lab has extensive experience with this technique), this requires a mouse with the desired floxed allele to remove a gene of interest. We’ve searched the literature and databases for transgenic mice with floxed alleles for genes we are most interested in (such as Igfbp5, Wnt7b, Rmst and Cxcl14), but these do not exist. The complimentary strategy would be to have a transgenic line expressing Cre driven by genes of interest. This would allow us to fate map and characterize specific cell populations in the adult mouse, but again we could not find the desired mouse lines. Additionally, the field does not have a CGE-specific Cre-driver and thus specifically removing genes from the CGE is quite challenging (as stated in the discussion, lines 364-369). Thus, targeted gene manipulation would require extensively more work (as Reviewer #1 noted) and we don’t currently have the necessary genetic tools to perform these types of experiments. As noted in our previous Response, we hope to generate Cre-drivers for 1-2 of the most intriguing candidate genes, but this will take 6-12 months and is beyond the scope of this study.

…or highlighting by RNAscope and EdU cell cycle monitoring that the marker population has indeed specific cell cycle properties (slow division, for example).

We do not understand the reviewer’s suggested association between characterizing cell cycle dynamics with EdU and the regional expression patterns of specific genes. What does the reviewer mean by ‘marker’ population? Is the reviewer implying that certain cell populations expressing specific genes (e.g., Igfbp5 in the CGE) have different cell cycle properties? Several studies from the 1990’s carefully explored cell cycle dynamics in the GEs and noted slightly different cell cycle times between different ages (E11 vs. E12) and different LGE domains (rostral, middle, caudal) in WT mice (Bhide, JCN 1996; Sheth & Bhide, JCN 1997). And there definitely are (subtle?) differences in cell cycle dynamics between VZ and SVZ cells that have been previously reported. Is this what the reviewer is referring to regarding ‘slow division’? We don’t make any claims about specific genes regulating the cell cycle properties in the manuscript, or even if specific GE subdomains have differential cell cycle dynamics. This type of EdU pulse-chase experiment could make sense if we were looking at a knockout mouse where there might be changes in cell cycle properties, but that is not the case here. Thus, we don’t understand the logic for performing cell cycle analysis in this context, and we apologize if we misunderstood what the reviewer is asking.

One specific example that raises doubts about the usefulness of the markers proposed is Fabp7, as indeed staining seems to label all cells in the ventricular zone, even though levels are always higher at boundary regions. This would mean that the "absent" expression of Fabp7 may simply be due to the depth of sequencing. Therefore, I suggest the authors perform immunostainings to verify or falsify the usefulness of this "marker".

We don’t fully understand what the reviewer is referring to regarding Fabp7 expression. Fabp7 is not expressed in all VZ cells based on our scRNA-seq data of the high-Nestin/VZ (new Figure 5E); note that there are clear regions of the MGE and LGE clouds where little/no Fabp7 is expressed (less evident for CGE). Similarly, in the in situ panels (new Figure 5E green pseudocolor), Fabp7 is only expressed in the most dorsal LGE and ventral MGE VZ cells; there is essentially no Fabp7 expression between these border zones. In the CGE, there appears to be a smaller region where Fabp7 is absent (in the middle CGE), which correlates with fewer CGE cells being Fabp7­-negative in the adjacent UMAP plots.

Thus, we believe that the scRNA-seq and in situs are largely in agreement and not in conflict. Based on this, we don’t know how to address the reviewer’s point about depth of sequencing except to say that our samples were sequenced to a sufficient depth to detect all gene expression patterns. This can be confirmed in the same image panel in Figure 5E, where Bcan (light blue) is expressed by a very small number of cells in the ventral MGE, which was still detected in our scRNA-seq in the lower left side of the adjacent UMAP plot.

Regardless, to convince the reviewer that the Fabp7 expression is legit, we immunostained an E14.5 brain section for Fabp7/BLBP, Author response image 4. Note that Fabp7/BLBP protein is restricted to VZ cells in the most dorsal LGE and ventral MGE, mimicking our in situ expression pattern. We don’t believe it is worth adding this immunostained image to the manuscript because it seems odd to just have 1 immunostained image to confirm an expression pattern that is not in conflict between our scRNAseq and in situ data. But if the reviewer or editor insists, we can include this image as a supplementary figure.

Author response image 4.

Author response image 4.

Open in a new tab

This leads to my last suggestion namely to focus the analysis to some extent on genes encoding cell surface proteins to allow sorting subsets of progenitors.

Single cell sequencing experiments generate many candidate genes of interests, and a significant challenge is figuring out how best to follow up on these candidates. If our primary interest was to identify candidates for sorting different progenitor subsets, then focusing on cell surface proteins certainly does make sense. One could also focus on transcription factors (due to their critical role in regulating gene expression and cell fate), receptors for diffusible cues that may influence cell cycle, or many other gene categories. Rather than focusing on specific gene families, we biased our analysis to genes displaying the most intriguing spatial or temporal expression profiles (regardless of gene family/function) to highlight previously overlooked heterogeneity of these cells. We invite all researchers to download our dataset and explore the expression pattern and function of their favorite gene categories.

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Data Citations

    1. Lee DR, Rhodes CT, Mitra A, Zhang Y, Maric D, Dale RK, Petros TJ. 2021. Transcriptional heterogeneity of ventricular zone cells throughout the embryonic mouse forebrain. NCBI Gene Expression Omnibus. GSE167013 [DOI] [PMC free article] [PubMed]
    2. Lee DR, Rhodes CT, Mitra A, Zhang Y, Maric D, Dale RK, Petros TJ. 2021. Transcriptional heterogeneity of ventricular zone cells throughout the embryonic mouse forebrain. NCBI Gene Expression Omnibus. GSE190593 [DOI] [PMC free article] [PubMed]
    3. Mayer C, Hafemeister C, Bandler RC, Machold R, Batista Brito R, Jaglin X, Allaway K, Butler A, Fishell G, Satija R. 2018. Developmental diversification of cortical inhibitory interneurons. NCBI Gene Expression Omnibus. GSE103983 [DOI] [PMC free article] [PubMed]
    4. Mi D, Li Z, Lim L, Li M, Moissidis M, Yang Y, Gao T, Hu TX, Pratt T, Price DJ, Sestan N, Marin O. 2018. Early emergence of cortical interneuron diversity in the mouse embryo. NCBI Gene Expression Omnibus. GSE109796 [DOI] [PMC free article] [PubMed]

    Supplementary Materials

    Supplementary file 1. Differentially expressed genes in the VZ and SVZ.

    Table contains the genes that are enriched in the Nestin-positive VZ cells (1,083 genes), mitotic Dcx-positive SVZ cells (330 genes) and postmitotic Dcx-positive SVZ/MZ cells (1,320 genes), all with a minimum log-fold change = 0.25.

    elife-71864-supp1.xlsx (239.7KB, xlsx)
    Transparent reporting form
    elife-71864-transrepform1.pdf (346.4KB, pdf)

    Data Availability Statement

    All of our sequencing data has been deposited in GEO under accession code GSE167013 and GSE190593.

    The following datasets were generated:

    Lee DR, Rhodes CT, Mitra A, Zhang Y, Maric D, Dale RK, Petros TJ. 2021. Transcriptional heterogeneity of ventricular zone cells throughout the embryonic mouse forebrain. NCBI Gene Expression Omnibus. GSE167013

    Lee DR, Rhodes CT, Mitra A, Zhang Y, Maric D, Dale RK, Petros TJ. 2021. Transcriptional heterogeneity of ventricular zone cells throughout the embryonic mouse forebrain. NCBI Gene Expression Omnibus. GSE190593

    The following previously published datasets were used:

    Mayer C, Hafemeister C, Bandler RC, Machold R, Batista Brito R, Jaglin X, Allaway K, Butler A, Fishell G, Satija R. 2018. Developmental diversification of cortical inhibitory interneurons. NCBI Gene Expression Omnibus. GSE103983

    Mi D, Li Z, Lim L, Li M, Moissidis M, Yang Y, Gao T, Hu TX, Pratt T, Price DJ, Sestan N, Marin O. 2018. Early emergence of cortical interneuron diversity in the mouse embryo. NCBI Gene Expression Omnibus. GSE109796

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES