The neurons of the three cerebellar nuclei (CN) are the primary output neurons of the cerebellum. The excitatory neurons (e) of the medial (m) CN (eCNm) were recently divided into molecularly defined subdomains in the adult; however, how they are established during development is not known. We define molecular subdomains of the mouse embryonic eCNm using single-cell RNA-sequencing and spatial expression analysis, showing that they evolve during embryogenesis to prefigure the adult. Furthermore, eCNm are transcriptionally divergent from cells in the other nuclei by embryonic day 14.5. We previously showed that loss of the homeobox genes En1 and En2 leads to loss of approximately half of the embryonic eCNm. We demonstrate that mutation of En1/2 in the embryonic eCNm results in death of specific posterior eCNm molecular subdomains and downregulation of TBR2 (EOMES) in an anterior embryonic subdomain, as well as reduced synaptic gene expression. We further reveal a similar function for EN1/2 in mediating TBR2 expression, neuron differentiation and survival in the other excitatory neurons (granule and unipolar brush cells). Thus, our work defines embryonic eCNm molecular diversity and reveals conserved roles for EN1/2 in the cerebellar excitatory neuron lineage.

Crucial to defining the development and function of an organ is determining the molecular heterogeneity of cell types and identifying the cellular properties of each subtype. The cerebellum provides an excellent system for exploring this question as it has only three excitatory cell types in addition to several inhibitory neurons, and the two lineages are derived from distinct embryonic progenitor zones (Leto et al., 2016; Joyner and Bayin, 2022). The adult cerebellum has an outer folded cerebellar cortex consisting of lobules situated above three bilaterally symmetrical groups of neurons called the cerebellar nuclei (CN) (Glickstein and Voogd, 1995; Voogd and Glickstein, 1998). The two groups of CN are subdivided into medial, intermediate and lateral nuclei (CNm, CNi and CNl) based on their location (Kebschull et al., 2023; Paxinos and Watson, 2006). There is preservation of a spatial map between the cells in the cortex and the CN, such that inhibitory Purkinje cells in the cortex located from medial to lateral project their axons to corresponding medial to lateral CN neurons (Apps et al., 2018). The CN are composed of excitatory cerebellar nuclei neurons (referred to as eCN), which have long-range projections to the rest of the brain/spinal cord, as well as inhibitory neurons (INs), which primarily project locally (Fujita et al., 2020; Judd et al., 2021; Kebschull et al., 2020; 2023). The CN are thus the primary output neurons of the cerebellum and mediate its wide array of motor and non-motor functions. During development, the eCN play a pivotal role in ensuring the proper expansion of several cell types in the lobules and, thus, growth of each lobule is dependent on accurate mapping of Purkinje cells to specific eCN (Willett et al., 2019). Given that the position of eCN in 3D space relates to the final morphology of the cerebellar cortex, it is important to determine whether there is an underlying molecular map of the developing eCN that provides a framework for growth and circuit formation in the lobules.

The eCN are the first-born neurons of the cerebellum, arising during embryonic day (E) 9.5-12.5 from an excitatory neuron progenitor zone called the rhombic lip (RL), soon after the cerebellar anlage is specified (Machold and Fishell, 2005; Wang et al., 2005). The newborn eCN migrate tangentially along the surface of the cerebellum and reside in the anterior-dorsal portion of the cerebellum in a region called the nuclear transitory zone (NTZ) until E15.5 (Altman and Bayer, 1985). They then rearrange to form the bilaterally symmetrical sets of three nuclei. Importantly, eCN promote survival of their upstream presynaptic partner Purkinje cells (Willett et al., 2019). The Purkinje cells then stimulate proliferation of postnatal progenitors of the excitatory granule cells, interneurons and astrocytes of the cerebellar cortex through secretion of sonic hedgehog (Corrales et al., 2004; Fleming et al., 2013). The eCN, together with the Purkinje cells, thereby regulate the size of the cerebellum (Willett et al., 2019).

Each nucleus of the adult mouse CN has molecularly defined subdomains based on their transcriptome and location. Using single-nuclear RNA sequencing (snRNA-seq) of dissected mouse CN, it has been shown that eCN in all three nuclei (eCNm, eCNi and eCNl) can be divided into two classes of neurons differing in size, electrophysiological properties and gene expression profiles, and further subdivided into spatially distinct subpopulations within each nucleus (Kebschull et al., 2020). Curiously, they found that the eCNm is distinct from the eCNi and eCNl, which are transcriptionally more similar to each other. In a separate study, the mouse adult eCNm was subdivided into four subpopulations based on the expression of three marker proteins (SPP1, SNCA and CALB2) and their distinct location and neural circuit organization (Fujita et al., 2020) (Fig. 1A). Similar to the snRNA-seq analysis, the four subpopulations were found to have different cell sizes and spatial distributions within the eCNm. Although these studies provide insights into the molecularly defined subpopulations of the adult eCN, how and when this molecular diversity emerges during development remains to be determined.

Fig. 1.

Two posterior subpopulations of the eCNm, defined by SPP1 laterally and SNCA medially, are absent in eCNm-En1/2-tdT CKO animals and anterior domains are diminished. (A) Table listing adult eCNm populations defined by Fujita et al., 2020 and Kebschull et al., 2020. (B) Cerebellar schematic indicating the anatomical levels shown in C-R. A, anterior; L, lateral; M, medial; P, posterior. (C) ISH for tdT in Allen Brain Mouse Connectivity Atlas of SepW1-Cre_NP39:Ai14 experiment 488246145 (https://connectivity.brain-map.org/transgenic/experiment/488246145) anterior coronal section with CN outlined. (D-R′) IF analysis of SPP1 and CALB2 and RNA in situ analysis of Snca as indicated in the eCNm on sagittal sections at lateral levels of the cerebellum of 11-week-old eCNm-En1/2-tdT CKOs and controls. Squares indicate locations from which higher power images (indicated by prime symbol) were taken. Filled arrowheads indicate tdT+ SPP1+ cells (E′), tdT+ CALB2+ cells (G′,H′) or tdT+ SPP+ cells (J′-M′). Unfilled arrowheads indicate tdT SPP1+ cells (F′) or tdT– Snca cells (P′,R′). White dashed outlines highlight the eCNm; gray dashed outlines highlight ectopically positioned UBCs based on lack of expression of MEIS2 (see Fig. S2). Schematics on the left summarize the data shown. Scale bars: 100 µm.

Fig. 1.

Two posterior subpopulations of the eCNm, defined by SPP1 laterally and SNCA medially, are absent in eCNm-En1/2-tdT CKO animals and anterior domains are diminished. (A) Table listing adult eCNm populations defined by Fujita et al., 2020 and Kebschull et al., 2020. (B) Cerebellar schematic indicating the anatomical levels shown in C-R. A, anterior; L, lateral; M, medial; P, posterior. (C) ISH for tdT in Allen Brain Mouse Connectivity Atlas of SepW1-Cre_NP39:Ai14 experiment 488246145 (https://connectivity.brain-map.org/transgenic/experiment/488246145) anterior coronal section with CN outlined. (D-R′) IF analysis of SPP1 and CALB2 and RNA in situ analysis of Snca as indicated in the eCNm on sagittal sections at lateral levels of the cerebellum of 11-week-old eCNm-En1/2-tdT CKOs and controls. Squares indicate locations from which higher power images (indicated by prime symbol) were taken. Filled arrowheads indicate tdT+ SPP1+ cells (E′), tdT+ CALB2+ cells (G′,H′) or tdT+ SPP+ cells (J′-M′). Unfilled arrowheads indicate tdT SPP1+ cells (F′) or tdT– Snca cells (P′,R′). White dashed outlines highlight the eCNm; gray dashed outlines highlight ectopically positioned UBCs based on lack of expression of MEIS2 (see Fig. S2). Schematics on the left summarize the data shown. Scale bars: 100 µm.

The homeobox genes Engrailed1/2 (En1/2) encode transcription factors that are key regulators of mouse cerebellar growth, patterning, and circuit formation (Cheng et al., 2010; Joyner, 1996; Joyner et al., 1991; Kuemerle et al., 1997; Millen et al., 1994, 1995; Orvis et al., 2012; Sgaier et al., 2007; Sillitoe et al., 2008, 2010; Willett et al., 2019; Wurst et al., 1994), as well of survival of specific mid- and hindbrain neuron types (Altieri et al., 2015; Fox and Deneris, 2012; Sgadò et al., 2006; Simon et al., 2001). These genes therefore provide a powerful entry point to study cerebellar development. Of interest for dissecting molecular characteristics of the eCN, we showed that ablation of En1/2 in eCN leads to death of a subset of cells in the eCNm and eCNi between E15.5 and E17.5 with a preferential loss of posteriorly located neurons (Willett et al., 2019). En1 and En2 are initially expressed broadly in the eCN at E14.5 and then become restricted to overlapping subsets of eCN during later embryonic development (Wilson et al., 2011; Willett et al., 2019). En1/2 are expressed in all cerebellar cell types in a similar manner, more broadly at earlier developmental stages (Sgaier et al., 2007). The finding that EN1/2 promote the survival of specific subpopulations of eCN indicates there could be molecular diversity of the eCN as early as E15.5. However, whether embryonic eCN subdomains can be defined and how they evolve and are related to the adult eCN subpopulations is not known.

Given the recently defined subpopulations of the mouse adult eCNm and the role of En1/2 in regulating survival of some posterior eCNm cells, we focused our current study on defining the molecular diversity of the embryonic eCNm and its temporal evolution in normal embryos and those lacking En1/2. Using single-cell RNA sequencing (scRNA-seq) and immunohistochemical spatial analysis, we reveal that the eCNm is transcriptionally distinct from the eCNi and eCNl by E14.5 (shortly after eCN neurogenesis). Furthermore, the spatial expression domains of six proteins specific to the eCNm [TBR1, CALB2, TBR2 (EOMES), PAX6, BARHL1 and LMX1A] become refined during embryonic development and at E17.5 their combined expression defines four subdomains. scRNA-seq analysis of E14.5 mutants lacking En1/2 in all eCN prior to their cell death revealed that EN1/2 promote expression of synaptic genes in the eCNm. En1/2 also regulate patterning of an anterior TBR2-expressing eCNm region, but cell viability of two posterior BARHL1+ eCNm subdomains. Uncovering a general role for EN1/2, the proteins upregulate TBR2 expression in the two other excitatory derivatives of the rhombic lip [granule cell precursors (GCPs) and unipolar brush cells (UBCs)] and promote their differentiation and the viability of UBCs. Our study thus defines four spatially segregated eCNm subpopulations in the embryo and demonstrates that EN1/2 play sequential roles in first setting up spatial patterning of the eCNm and later promoting neural specification and/or viability of specific subpopulations of the three excitatory cerebellar neuron types (eCN, granule cells, UBCs).

Adult eCNm subdomains are differentially vulnerable to embryonic loss of En1/2

As a step towards gaining a deeper understanding of the molecular subdivisions of the adult eCN, we determined whether the neurons lost in the eCNm of En1/2 conditional knockout mutants (CKOs) belong to one of the four spatial subdomains that can be identified based on preferential expression of SPP1, CALB2 and/or SNCA expression (F1-F4; Fujita et al., 2020) (Fig. 1A). We performed immunofluorescence (IF) and in situ hybridization (ISH) analysis on cerebellar sagittal sections of adult CKOs lacking En1/2 in eCN and GCPs and that carry a nuclear tdTomato (tdT) reporter (Atoh1-Cre/+; En1lox/lox, En2lox/lox; R26LSL-tdT/+ mice), referred to as eCN+GCP-En1/2-tdT CKOs, and littermate controls (En1lox/lox; En2lox/lox; R26LSL-tdT/+), referred to as En1/2f/f mice (Fig. S1, S2). We also analyzed a new allele in which En1/2 are deleted in the eCNm using a Selenow Bac construct to drive Cre (Gerfen et al., 2013) (SepW1-Cre/+; En1lox/lox, En2lox/lox; R26LSL-tdT/+ mice or eCNm-En1/2-td CKOs) and controls in which the eCNm was labeled using SepW1-Cre/+ with tdT (eCNm-tdT) (n=3 animals per genotype; Fig. 1B-R′). The SepW1-Cre lineage also includes differentiating GCPs and UBCs, and UBCs lacking En1/2 do not migrate properly (Fig. S2; see below and Lee at al., 2024 preprint). Strikingly, at all mediolateral levels, the posterior domain of the eCNm was absent. At lateral and intermediate levels of eCNm-En1/2-tdT CKO cerebella, where SPP1 alone marks most of the cells in the posterior dorsal region (F2) of the eCNm, no SPP1+/tdT+ cells were detected, whereas in eCNm-tdT controls a large distinct tdT+ domain was present (Fig. 1D-F′). The anterior and ventral region of the eCNm, which contains SPP1+ cells at medial and intermediate levels (F1) and scattered CALB2+ neurons more laterally (F3), was reduced in size in the eCNm-En1/2-tdT CKO mutants compared with eCNm-tdT controls (Fig. 1G-M′). At the most medial level, where Snca normally labels the posterior dorsal region of the eCNm (F4), no Snca mRNA was detected in eCNm-En1/2-tdT CKO mice and the tdT+ posterior dorsal eCNm domain was absent (Fig. 1N-R′). Similar results were seen in eCN+GCP-En1/2-tdT CKOs and littermate controls (n=3 animals per genotype; Fig. S1). Thus, absence of En1/2 in eCN results in loss of the SPP1+ and Snca posterior dorsal domains (F2 and F4) of the eCNm throughout the mediolateral axis and a partial loss of the remaining SPP1+ or CALB2+ anterior ventral eCNm (F1 and F3).

In the embryonic NTZ, TBR1 and OLIG2 mark complementary medial and lateral eCN domains, respectively

As a basis for determining whether the subpopulations of the eCNm that die in En1/2 eCN conditional mutants are transcriptionally distinct embryonically, we defined the transcriptional landscape of control eCN at E14.5, when they still reside in the NTZ. We performed scRNA-seq on tdT+ cells isolated by fluorescence activated cell sorting (FACS) from E14.5 eCN+GCP-tdT embryos (n=2 pools; Atoh1-Cre; R26LSL-tdT/+) using a 10x Genomics platform (Fig. S3A,B). After filtering out low-quality cells and integrating both replicates using Seurat v4.1 (see Materials and Methods), a total of 9620 cells were obtained (Fig. 2A, Fig. S3C). Unsupervised clustering of the combined cells yielded ten clusters that, as expected, were either enriched for Meis2, which marks the eCN (adjusted P≤ 0.05; Table S1) or Pax6, which is enriched in GCPs compared with eCN (adjusted P≤ 0.05; Table S1) (Fig. 2A′-A″, Fig. S3D-F) (Willett et al., 2019; Morales and Hatten, 2006; Engelkamp et al., 1999). En1 and En2 were detected across all clusters (Fig. S3G). Outside the cerebellum, Meis2 is expressed in neurons of the anterior NTZ (Fig. S3H,I) destined for the mid/hindbrain, including the isthmic nuclei (Wizeman et al., 2019), as well as in cell populations adjacent to the cerebellum (Fig. S3J,K). To identify which of the Meis2-enriched clusters were eCN, we isolated clusters 2, 3, 4, 6 and 7 and reiterated unsupervised clustering (Fig. 2B). Nine clusters were obtained, of which the largest cluster (0) was marked by Tbr1 (808 cells) and two other clusters (1 and 7) were enriched for Olig2 (1040 cells total) (Fig. 2B-B″). Tbr1 and Olig2 have been reported to be expressed in the E13.5 NTZ (Wizeman et al., 2019) and at E18.5 to mark the eCNm and eCNi+eCNl, respectively (Ju et al., 2016; Seto et al., 2014). The remaining clusters appeared to contain cells outside the cerebellum based on the expression patterns of their significantly enriched markers in the Allen Developing Mouse Brain Atlas E13.5 and E15.5 RNA ISH datasets (Table S2, Fig. S3G-I). We therefore excluded clusters 2-6 and 8, and repeated clustering on the Tbr1+ and Olig2+ clusters, which again yielded three clusters – one Tbr1-expressing cluster (0) and two Olig2+ clusters (Fig. 2C-C″), which were enriched for either Nrp1 (1) or Nr2f1 (2) (Fig. S3J).

Fig. 2.

TBR1 and OLIG2 label complementary medial and lateral domains, respectively, within the eCNm-containing region of the NTZ. (A-A″) Uniform Manifold Approximation and Projection (UMAP) visualization of tdT+ cell clusters from eCN+GCP-tdT (Atoh1-Cre; R26LSL-tdT/+) E14.5 cerebella (n=2 pools), with Meis2 (A′) and Pax6 (A″) expression. (B-B″) UMAP visualization of Meis2-enriched cluster reclustering, with Tbr1 (B′) and Olig2 (B″) expression. (C-C″) UMAP visualization of Tbr1+ and Olig2+ reclustering with Tbr1 (C′) and Olig2 (C″) expression. (D) Schematic of an E14.5 cerebellum showing the anatomical levels shown in E-H and J-M, and below a medial sagittal section. (E-H) Merge (E) and tdT (F), TBR1 (G) and OLIG2 (H) single-channel IF images of medial E14.5 eCN+GCP-tdT NTZ. (I) Schematic of a lateral sagittal section. (J-M) Merge (J) and tdT (K), TBR1 (L) and OLIG2 (M) single-channel IF images of lateral E14.5 eCN+GCP-tdT NTZ. (N) Schematics of E14.5 cerebellum showing the anatomical levels shown in O-R and T-W, and below a medial sagittal section. (O-R) Merge (O) and tdT (P), TBR1 (Q) and OLIG2 (R) single-channel IF images of medial E14.5 eCNm-tdT NTZ. (S) Schematic of a lateral sagittal section. (T-W) Merge (T) and tdT (U), TBR1 (V) and OLIG2 (W) single-channel IF images of lateral E14.5 eCN+GCP-tdT NTZ. Dashed lines delineate the eCN-containing domain of the NTZ. (X) Schematic summary of expression results. A, anterior; L, lateral; M, medial; P, posterior. Scale bars: 100 µm.

Fig. 2.

TBR1 and OLIG2 label complementary medial and lateral domains, respectively, within the eCNm-containing region of the NTZ. (A-A″) Uniform Manifold Approximation and Projection (UMAP) visualization of tdT+ cell clusters from eCN+GCP-tdT (Atoh1-Cre; R26LSL-tdT/+) E14.5 cerebella (n=2 pools), with Meis2 (A′) and Pax6 (A″) expression. (B-B″) UMAP visualization of Meis2-enriched cluster reclustering, with Tbr1 (B′) and Olig2 (B″) expression. (C-C″) UMAP visualization of Tbr1+ and Olig2+ reclustering with Tbr1 (C′) and Olig2 (C″) expression. (D) Schematic of an E14.5 cerebellum showing the anatomical levels shown in E-H and J-M, and below a medial sagittal section. (E-H) Merge (E) and tdT (F), TBR1 (G) and OLIG2 (H) single-channel IF images of medial E14.5 eCN+GCP-tdT NTZ. (I) Schematic of a lateral sagittal section. (J-M) Merge (J) and tdT (K), TBR1 (L) and OLIG2 (M) single-channel IF images of lateral E14.5 eCN+GCP-tdT NTZ. (N) Schematics of E14.5 cerebellum showing the anatomical levels shown in O-R and T-W, and below a medial sagittal section. (O-R) Merge (O) and tdT (P), TBR1 (Q) and OLIG2 (R) single-channel IF images of medial E14.5 eCNm-tdT NTZ. (S) Schematic of a lateral sagittal section. (T-W) Merge (T) and tdT (U), TBR1 (V) and OLIG2 (W) single-channel IF images of lateral E14.5 eCN+GCP-tdT NTZ. Dashed lines delineate the eCN-containing domain of the NTZ. (X) Schematic summary of expression results. A, anterior; L, lateral; M, medial; P, posterior. Scale bars: 100 µm.

Having identified E14.5 eCN clusters, we confirmed that TBR1 and OLIG2 were indeed expressed in cells derived from cells that express the Atoh1-Cre transgene by performing IF on cerebellar sagittal sections across the mediolateral extent of E14.5 eCN+GCP-tdT embryos (Atoh1-Cre; R26LSL-tdT/+). TBR1 and OLIG2 were detected in complementary domains throughout the tdT+ NTZ (Fig. 2D-M,X). TBR1 was present in the entire NTZ at medial levels (Fig. 2E,G), and was restricted to the posterior NTZ laterally (Fig. 2J,L). In contrast, OLIG2 was absent from the medial NTZ (Fig. 2E,H) and marked the anterior NTZ, complementary to TBR1, at lateral levels (Fig. 2J,M). Because we observed some TBR1+ eCN at lateral levels, we tested whether TBR1 is exclusive to the eCNm at this stage by analyzing E14.5 eCNm-tdT cerebella (Fig. 2N-W). Strikingly, we found that TBR1 was co-expressed with nearly all tdT+ cells in eCNm-tdT cerebella (Fig. 2O,Q,T,V), whereas OLIG2 was excluded from tdT+ cells (Fig. 2O,R,T,W, X). As the eCNm-Cre transgene marks only the eCNm in adult mice, these results demonstrate that eCNm cells are transcriptionally distinct from cells in the remaining eCNi+eCNl shortly after they are generated and are uniquely marked by TBR1. Furthermore, some eCNm neurons are initially located laterally (Fig. 2T-W).

Markers enriched in E14.5 eCNm have distinct anteroposterior and mediolateral distributions

We next generated a list of genes enriched in the Tbr1+ eCNm cluster 0 at E14.5 (Table S3). Interestingly, of the three genes used to subdivide the adult eCNm into subpopulations, only Calb2 was enriched in the Tbr1+ eCNm cluster in our scRNA-seq dataset (Fig. 3A). Snca was present in a few cells in both the eCNm and eCNi+eCNl clusters (Fig. S4A) and the Allen Developing Mouse Brain Atlas shows Snca expressed broadly in the eCN at E15.5 and E18.5 (Fig. S4B-K). Spp1 was not detected in our scRNA-seq dataset or using IF analysis of E14.5 or E17.5 eCNm-tdT cerebellar sections (Fig. S4L-U). Thus, our expression results indicate that the transcriptome of the embryonic eCNm is distinct from that of the mature eCNm and thus must evolve during postnatal development.

Fig. 3.

Protein markers enriched in eCNm have distinct spatial patterns in the NTZ at E14.5. (A) Expression of eCNm, eCNi+eCNl and pan-eCN markers across all eCN clusters. (B) Percentage of double-positive (tdT and eCNm marker) cells in the NTZ at nine medio-lateral levels in the E14.5 eCNm-tdT cerebellum (n=3 animals). Data are presented as mean±s.e.m. (C-Z) Colocalization of tdT and each eCNm marker in E14.5 eCNm-tdT cerebella at medial and lateral levels, with schematics of marker expression projected onto a dorsal view of E14.5 eCNm 3D-reconstruction (C,G,K,O,S,W) and sagittal section schematics (D,H,L,T,P,X). Dashed outlines delineate the eCNm. A, anterior; L, lateral; M, medial; P, posterior. Scale bars: 100 µm.

Fig. 3.

Protein markers enriched in eCNm have distinct spatial patterns in the NTZ at E14.5. (A) Expression of eCNm, eCNi+eCNl and pan-eCN markers across all eCN clusters. (B) Percentage of double-positive (tdT and eCNm marker) cells in the NTZ at nine medio-lateral levels in the E14.5 eCNm-tdT cerebellum (n=3 animals). Data are presented as mean±s.e.m. (C-Z) Colocalization of tdT and each eCNm marker in E14.5 eCNm-tdT cerebella at medial and lateral levels, with schematics of marker expression projected onto a dorsal view of E14.5 eCNm 3D-reconstruction (C,G,K,O,S,W) and sagittal section schematics (D,H,L,T,P,X). Dashed outlines delineate the eCNm. A, anterior; L, lateral; M, medial; P, posterior. Scale bars: 100 µm.

To determine whether molecular subpopulations of the eCNm exist at E14.5, we selected four significantly enriched eCNm marker genes (Tbr2, Pax6, Barhl1 and Lmx1a; Table S3) in addition to Calb2 and Tbr1, for spatial expression analysis because validated antibodies were available and they had been described as expressed in the NTZ or the eCNm (Chizhikov et al., 2006; Fink et al., 2006; Rose et al., 2009; Yeung et al., 2016) (Fig. 3A). We quantified the percentage of tdT+ eCNm NTZ cells in eCNm-tdT E14.5 embryos positive for each marker on every tenth sagittal section from medial to lateral levels (Fig. 3B). To represent the expression domain of each marker in the anteroposterior plane, we performed a 3D reconstruction of the E14.5 eCNm-tdT+ cells and projected the expression of each marker onto a dorsal view of the reconstruction. TBR1 (Fig. 3C-F) and LMX1A (Fig. 3G-J) were expressed across most of the tdT+ eCNm except for a few anteriorly located cells. Strikingly, the other markers were expressed in distinct domains restricted along the mediolateral and/or anteroposterior axes. We found that CALB2 (Fig. 3K-N) labels a subset of cells across the mediolateral extent of the tdT+ NTZ, aside from some anterior cells at medial levels where TBR2 (Fig. 3O-R) is expressed. PAX6 (Fig. 3S-V) and BARHL1 (Fig. 3W-Z) were enriched in the posterior-most region of the tdT+ NTZ laterally and present in most tdT+ cells medially.

We then performed double staining for proteins to gain a better understanding of the molecular subdomains. Sixty percent of TBR2+ cells expressed TBR1 (Fig. S5A-E), and most TBR2+/TBR1 cells were located laterally. At medial levels, there was a distinct, anteriorly located domain where cells expressed PAX6, TBR2 and/or BARHL1, but not CALB2 (Fig. S5F-K″). Interestingly, within the remaining TBR2 region where CALB2 is expressed we observed both single-positive CALB2+, BARHL1+ or PAX6+ cells as well as double-positive CALB2+/BARHL1+ or CALB2+/PAX6+ cells, suggesting further molecular complexity within a single domain. Taken together, these findings demonstrate that the eCNm already has molecular subpopulations at E14.5 that correlate with their 3D positions in the NTZ, suggesting that subpopulations of neurons are transcriptionally defined soon after they are generated.

eCNm molecular subpopulations evolve between E14.5 and E17.5

We next investigated whether the molecular diversity of the eCNm at E14.5 is stable or evolves during development and establishment of the CN. We generated 3D reconstructions of the expression domains of the same set of markers at E17.5, the earliest stage when CN morphology and positioning closely resembles that of the adult (Fig. S6). Whereas in eCNm-tdT embryos some tdT+ cells were located laterally at E14.5, by E17.5 the tdT+ cells were restricted to medial sections (Fig. S7). To test whether cell death removes the lateral eCNm cells, we performed terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining on E15.5 eCNm-tdT cerebella and did not detect increased cell death in lateral compared with medial tdT+ cells (Fig. S8A-D). Nevertheless, the total number of tdT+ cells was slightly decreased (14%) at 17.5 compared with E14.5, indicating some lateral cells could die (Fig. S8E).

Interestingly, all six markers labeled a lower percentage of the tdT+ cells in eCNm-tdT E17.5 cerebella than at E14.5 (Fig. 4A), and the expression domains of most of the markers changed between the time points (compare Fig. 3B with Fig. 4B). Additionally, it became apparent that the expression domains of most markers developed additional spatial segregation along the dorsoventral axis. TBR1 was found to be slightly more enriched laterally than medially at E17.5, and throughout the mediolateral axis it appeared more highly expressed ventrally (Fig. 4C-F). LMX1A became restricted at E17.5 to anterior and ventral regions in the eCNm across the mediolateral axis (Fig. 4G-J). CALB2 became enriched in the ventral regions of the tdT+ eCNm (Fig. 4K-N). Interesting, TBR2 showed a very similar pattern at E17.5 and E14.5 with TBR2+ cells solely in a small anterior and ventral region of the tdT+ eCNm with enrichment medially (Fig. 4O-R). PAX6, like LMX1A, became restricted at E17.5 to anterior and ventral regions in the eCNm, although LMX1A+ cells were less dense (Fig. 4S-V). BARHL1 was highly enriched in the posterior dorsal eCNm at all levels, aside from a small group of cells in the anterior medial eCNm (Fig. 4W-Z).

Fig. 4.

Protein markers enriched in the eCNm evolve between E14.5 and E17.5. (A) Comparison of the percentage of double-positive (tdT and marker) eCNm cells at E14.5 and E17.5 in eCNm-tdT mice (n=3 animals per age, unpaired, two-tailed Student's t-test). Error bars represent s.d. (B) Percentage of double-positive eCNm cells at six medio-lateral levels in the E17.5 eCNm-tdT cerebellum (n=3 animals). Data are presented as mean±s.e.m. (C-Z) Colocalization of tdT and each eCNm marker in E17.5 eCNm-tdT cerebella at medial and lateral levels, accompanied by schematics of eCNm marker expression projected onto a dorsal view of E17.5 eCNm 3D-reconstruction (C,G,K,O,S,W) and sagittal section schematics (D,H,L,T,P,X). Dashed outlines delineate the eCNm. A, anterior; D, dorsal; P, posterior; V, ventral. Scale bars: 100 μm.

Fig. 4.

Protein markers enriched in the eCNm evolve between E14.5 and E17.5. (A) Comparison of the percentage of double-positive (tdT and marker) eCNm cells at E14.5 and E17.5 in eCNm-tdT mice (n=3 animals per age, unpaired, two-tailed Student's t-test). Error bars represent s.d. (B) Percentage of double-positive eCNm cells at six medio-lateral levels in the E17.5 eCNm-tdT cerebellum (n=3 animals). Data are presented as mean±s.e.m. (C-Z) Colocalization of tdT and each eCNm marker in E17.5 eCNm-tdT cerebella at medial and lateral levels, accompanied by schematics of eCNm marker expression projected onto a dorsal view of E17.5 eCNm 3D-reconstruction (C,G,K,O,S,W) and sagittal section schematics (D,H,L,T,P,X). Dashed outlines delineate the eCNm. A, anterior; D, dorsal; P, posterior; V, ventral. Scale bars: 100 μm.

Double staining (Fig. S9) showed that TBR1 and TBR2 co-expression was increased at E17.5 (∼90% of TBR2+ cells expressed TBR1) compared with E14.5 (Fig. S9A-D″). Additionally, at medial levels, although there were a few anterior PAX6+, TBR2+ or BARHL1+ cells that were CALB2, the domain was not as distinct as the anterior CALB2 domain at E14.5 (Fig. S9E-K″). The proteins that showed the most complementarity were CALB2 and BARHL1: at all levels CALB2 was absent from the posterior dorsal BARHL1+ eCNm (Fig. S9I-K″). Thus, aside from TBR2, the expression domains of eCNm markers evolve between E14.5 and E17.5. Based on combined expression patterns, the eCNm can be considered to have four subdomains at E17.5 (see Discussion, Fig. 7).

Initial segregation of eCNm and eCNi+eCNl is maintained in En1/2 CKOs but neural maturation genes are reduced

Having established a spatial molecular map of the embryonic eCNm, we next investigated whether loss of En1/2 alters the initial patterning of the eCN into medial and eCNi+eCNl populations. In scRNA-seq analysis on FACS-isolated tdT+ cells from eCN+GCP-En1/2-tdT CKO samples (n=2 pooled samples with 5 embryos each) (Fig. S10), 8606 cells passed quality control and formed 11 clusters (Fig. S10A-F, Table S4), of which Meis2-enriched cells (1925 cells) were further clustered (Fig. S10G-J, Table S5) to identify the eCNm (805 cells), which contained three Olig2-enriched clusters and one Tbr1 cluster (Fig. S10K-N, Table S6). The percentage of Tbr1+ cells of all the eCNm in the mutant (eCN+GCP-En1/2-tdT CKO) dataset (48.14%) was similar to that of the control (eCN+GCP-tdT) dataset (43.72%) and, like in control, the Olig2+ clusters in mutants were enriched for Nrp1 or Nr2f1 (Fig. S10N). To test whether the three eCN+GCP-En1/2-tdT CKO Olig2+ clusters were similar to the two eCN+GCP-tdT Olig2+ clusters (see Fig. 2C), we generated predicted identities for cells from the eCN+GCP-En1/2-tdT CKO clustering using the eCN+GCP-tdT dataset as the reference dataset. We found that most Olig2+ cells in the eCN+GCP-En1/2-tdT CKO clustering mapped to one of the two eCN+GCP-tdT Olig2+ clusters (Fig. S10O).

We then integrated our eCN+GCP-En1/2-tdT CKO samples with the eCN+GCP-tdT control samples (18,400 cells total) and performed reiterative unsupervised clustering to identify all eCN (Fig. 5A-I′, Fig. S11A-D, Tables S7-S9). The eCN from the combined genotypes (2780 cells) formed one Olig2+ and one Tbr1+ cluster of eCN and a small Mafb+ cluster that is a contaminant from outside the cerebellum (Fig. 5G-I′, Fig. S11E,F). The mutant and control cells were distributed similarly throughout the three clusters (Fig. 5H). The Olig2+ cluster had two subdomains defined by markers specific to each of the two Olig2+ control clusters (Nrp1 and Nr2f1), indicating that a further subdivision exists (Fig. S11G-O).

Fig. 5.

Initial patterning of eCN into complementary TBR1 and OLIG2 domains at E14.5 appears normal in En1/2 CKOs. (A) UMAP visualization of tdT+ cell clustering of control and eCN+GCP-En1/2-tdT CKO eCN combined at E14.5. (B,B′) Distribution of cells from each genotype across clusters. (C,C′) Expression of Meis2 and Pax6. (D) UMAP visualization of Meis2-enriched clusters reclustered. (E,E′) Distribution of cells from each genotype across clusters. (F,F′) Expression of Tbr1 and Olig2. (G) UMAP visualization of reclustering of Tbr1+ and Olig2+ clusters. (H,H′) Distribution of cells from each genotype across clusters. (I,I′) Expression of Tbr1 and Olig2. (J-N) Schematics and IF images of TBR1 and OLIG2 expression in the NTZ of E14.5 En1/2f/f (K,M) and eCN+GCP-En1/2-tdT CKO (L,N) sagittal sections of cerebella at medial and lateral levels of the cerebellum. Dashed outlines delineate the eCN-containing domain of the NTZ. (O) Quantification of MEIS2+ TBR1+ or OLIG2+ cells on nine sections across the NTZ (n=4 animals per genotype; unpaired, two-tailed Student's t-test). Data are expressed as mean±s.d. A, anterior; L, lateral; M, medial; P, posterior. Scale bars: 100 µm (K-N).

Fig. 5.

Initial patterning of eCN into complementary TBR1 and OLIG2 domains at E14.5 appears normal in En1/2 CKOs. (A) UMAP visualization of tdT+ cell clustering of control and eCN+GCP-En1/2-tdT CKO eCN combined at E14.5. (B,B′) Distribution of cells from each genotype across clusters. (C,C′) Expression of Meis2 and Pax6. (D) UMAP visualization of Meis2-enriched clusters reclustered. (E,E′) Distribution of cells from each genotype across clusters. (F,F′) Expression of Tbr1 and Olig2. (G) UMAP visualization of reclustering of Tbr1+ and Olig2+ clusters. (H,H′) Distribution of cells from each genotype across clusters. (I,I′) Expression of Tbr1 and Olig2. (J-N) Schematics and IF images of TBR1 and OLIG2 expression in the NTZ of E14.5 En1/2f/f (K,M) and eCN+GCP-En1/2-tdT CKO (L,N) sagittal sections of cerebella at medial and lateral levels of the cerebellum. Dashed outlines delineate the eCN-containing domain of the NTZ. (O) Quantification of MEIS2+ TBR1+ or OLIG2+ cells on nine sections across the NTZ (n=4 animals per genotype; unpaired, two-tailed Student's t-test). Data are expressed as mean±s.d. A, anterior; L, lateral; M, medial; P, posterior. Scale bars: 100 µm (K-N).

IF staining on littermate control and eCN+GCP-En1/2-tdT CKO E14.5 cerebella (n=4 animals per genotype) for TBR1 and OLIG2 across the CN revealed no obvious change in the mediolateral and anteroposterior distributions of the markers compared with controls (Fig. 5J-N). Additionally, quantification of TBR1+ and OLIG2+ eCN from every tenth sagittal section of the NTZ demonstrated no change in the number of cells expressing either marker between eCN+GCP-En1/2-tdT CKO and eCN+GCP-tdT embryos (Fig. 5O). Thus, En1/2 does not play a major role in the initial patterning of eCN into eCNm and eCNi+eCNl populations.

Given that there was no difference between the number or distribution of TBR1+ and OLIG2+ eCN, we then performed differential gene expression analyses between the eCN+GCP-En1/2-tdT CKO and eCN+GCP-tdT cells within the Tbr1+ or Olig2+ cluster using Libra, a pseudo-bulk differential gene expression algorithm (Squair et al., 2021) (Fig. S12, Table S10). eCN+GCP-En1/2-tdT CKO cells in both Tbr1+ and Olig2+ clusters showed altered expression of genes compared with their control counterparts (Fig. S12A,B); however, gene ontology (GO) analysis identified significant terms only for the Tbr1+ cluster (Fig. S12C, Table S11). GO analysis determined that genes downregulated in the Tbr1+ cluster (Ephb1, Gja1, Flrt2 and Pcdh17; Fig. S12D) belonged to synaptic gene and cell adhesion categories, suggesting that En1/2 have a role in promoting differentiation and maturation of the eCNm.

Molecular subpopulations of the eCNm are differentially vulnerable to the loss of En1/2

Given that cells in the eCNm of E14.5 eCN+GCP-En1/2 CKOs have altered transcriptomes related to neural maturation before some mutant cells begin to experience cell death (Willett et al., 2019), we next tested whether the six markers for molecular subpopulations of eCNm (Figs 3,4) undergo changes in their domains in mutants. IF analysis and quantification of eCN in eCNm-En1/2-tdT CKO (Fig. 6A-K, Figs S13A-I, S14A) and eCN+GCP-En1/2 CKO (Fig. S15A,B) animals revealed that at E14.5 the TBR2+ subpopulation was the main population altered. TBR2 was absent in eCNm-En1/2-tdT CKOs and greatly reduced (73.38%) in eCN+GCP-En1/2 CKOs compared with controls, despite no reduction in overall eCN number at E14.5 based on tdT or TBR1 expression (Fig. 6A-K, Figs S13A-I, S15A,B). The absolute number of eCNm cells that expressed PAX6 or TBR1 in eCNm-En1/2-tdT CKO was not altered, although there was a slight but significant increase in the percentage of tdT+ eCNm cells that expressed either marker. The other subpopulation markers appeared normal, except for a slight increase in the number of BARHL1+ eCN in eCN+GCP-En1/2 CKOs compared with controls. Furthermore, given that ∼60% of TBR2+ cells express TBR1 at E14.5 in eCNm-tdT control embryos (Fig. S5), it is likely that the lack of TBR2 expression in mutants reflects a downregulation of Tbr2 gene expression rather than loss of the cells.

Fig. 6.

En1/2 loss preferentially affects BARHL1, TBR2 and TBR1 eCNm expression domains. (A) Schematic of E14.5 cerebellum, with line indicating the anatomical level of the images in B-I. (B-I) IF images of tdT (B,C), TBR2 (D,E), TBR1 (F,G) and BARHL1 (H,I) eCNm expression in E14.5 eCNm-En1/2-tdT CKO and eCNm-tdT control cerebella. (J,J′) Quantification of tdT+ eCNm cells (from every second section) and TBR2, BARHL1 and TBR1 eCNm (from every tenth section) subpopulations in E14.5 eCNm-tdT and eCNm-En1/2-tdT CKO embryos (n=4 eCNm-tdT and 3 eCNm-En1/2-tdT CKO animals; unpaired, two-tailed Student's t-test). Error bars represent s.d. (K) Schematics of marker expression in the eCNm of eCNm-En1/2-tdT CKO embryos projected onto 3D reconstructions of the E14.5 En1/2 CKO eCNm. (L) Schematic of E17.5 cerebellum, with line indicating the anatomical level of the images in M-T. (M-T) IF images of tdT (M,N), TBR2 (O,P), TBR1 (Q,R) and BARHL1 (S,T) expression in E17.5 eCNm-En1/2-tdT CKO and eCNm-tdT cerebella. (U,U′) Quantification of tdT+ eCNm (from every second section) and TBR2, BARHL1 and TBR1 eCNm (from every tenth section) subpopulations in E17.5 eCNm-tdT and eCNm-En1/2-tdT CKO embryos (n=3 animals per genotype, unpaired, two-tailed Student's t-test). Error bars represent s.d. (V) Schematics of marker expression in eCNm of eCNm-En1/2-tdT CKO embryos projected onto 3D reconstructions of the E17.5 En1/2 CKO eCNm. Dashed outlines delineate the eCNm. A, anterior; D, dorsal; L, lateral; M, medial; P, posterior; V, ventral. Scale bars: 50 µm (B-I); 100 µm (M-T).

Fig. 6.

En1/2 loss preferentially affects BARHL1, TBR2 and TBR1 eCNm expression domains. (A) Schematic of E14.5 cerebellum, with line indicating the anatomical level of the images in B-I. (B-I) IF images of tdT (B,C), TBR2 (D,E), TBR1 (F,G) and BARHL1 (H,I) eCNm expression in E14.5 eCNm-En1/2-tdT CKO and eCNm-tdT control cerebella. (J,J′) Quantification of tdT+ eCNm cells (from every second section) and TBR2, BARHL1 and TBR1 eCNm (from every tenth section) subpopulations in E14.5 eCNm-tdT and eCNm-En1/2-tdT CKO embryos (n=4 eCNm-tdT and 3 eCNm-En1/2-tdT CKO animals; unpaired, two-tailed Student's t-test). Error bars represent s.d. (K) Schematics of marker expression in the eCNm of eCNm-En1/2-tdT CKO embryos projected onto 3D reconstructions of the E14.5 En1/2 CKO eCNm. (L) Schematic of E17.5 cerebellum, with line indicating the anatomical level of the images in M-T. (M-T) IF images of tdT (M,N), TBR2 (O,P), TBR1 (Q,R) and BARHL1 (S,T) expression in E17.5 eCNm-En1/2-tdT CKO and eCNm-tdT cerebella. (U,U′) Quantification of tdT+ eCNm (from every second section) and TBR2, BARHL1 and TBR1 eCNm (from every tenth section) subpopulations in E17.5 eCNm-tdT and eCNm-En1/2-tdT CKO embryos (n=3 animals per genotype, unpaired, two-tailed Student's t-test). Error bars represent s.d. (V) Schematics of marker expression in eCNm of eCNm-En1/2-tdT CKO embryos projected onto 3D reconstructions of the E17.5 En1/2 CKO eCNm. Dashed outlines delineate the eCNm. A, anterior; D, dorsal; L, lateral; M, medial; P, posterior; V, ventral. Scale bars: 50 µm (B-I); 100 µm (M-T).

We then examined whether specific eCNm molecular subpopulations were lost or altered in their pattern at E17.5 (Fig. 6L-V, Figs S13J-R, S14B, S15C,D), given our previous finding that ∼18% of all eCN are lost to cell death in eCN+GCP-En1/2 CKOs by this stage (Willett et al., 2019). As expected, the eCNm-En1/2-tdT CKO mutants had a significant reduction in tdT+ eCNm cells compared with controls (34%) (Fig. 6U). TBR2 was virtually absent in both E17.5 En1/2 CKOs (100% reduction in eCNm-En1/2-tdT and 98.29% in eCN+GCP-En1/2-tdT CKOs). At E17.5, the loss of TBR2+ eCNm was likely due to a failure of eCNm to express TBR2 in the absence of En1/2, as TBR1+ cell number was not reduced but significantly increased in the two En1/2 CKOs (Fig. 6U′, Fig. S15D). We also found that the BARHL1+ cells in the posterior-most eCNm were significantly reduced throughout the mediolateral axis of the E17.5 En1/2 CKOs compared with their controls (total loss: 55.11% in eCNm-En1/2-tdT CKO; 51.06% in eCN+GCP-En1/2-tdT CKO) (Fig. 6U′, Figs S13J-R, S15D). The only BARHL1+ eCNm cells remaining in the two En1/2 CKOs were scattered cells in the anterior region. The domains marked by CALB2, PAX6 and LMX1A appeared largely unaffected both in pattern and number of eCNm cells in both En1/2 CKOs at both ages compared with controls. As expected, given that the total number of cells was reduced as a result of loss of BARHL1+ cells at E17.5, CALB1+ and LMX1A+ cells (as well as TBR1) as a percentage of tdT+ cells were increased (Fig. S14B). Double labeling of TBR1/2 and BARHL1/CALB2 in mutants confirmed the location of the missing cells (Fig. S9L-O). Thus, our spatial expression results show that embryonic eCNm subpopulations are differentially vulnerable to the loss of En1/2 (summarized in Fig. 7).

Fig. 7.

Proposed adult and embryonic eCNm subdomains, and alterations seen in En1/2 conditional mutants. (A) Table listing proposed eCNm subdomains based on marker expression and spatial location in normal adult brain (based on Fujita et al., 2020) and at E17.5 in normal and En1/2 CKOs (this study). Note that most proteins are expressed in a subset of cells in the four domains. Brackets around SNCA indicate where we could not detect expression. (B) Schematics of the eCNm in the adult (left) and the NTZ (E14.5) or eCNm (E17.5) in embryos showing how the expression of some spatially defined marker proteins evolves during development and which subdomains are altered in En1/2 CKOs. Solid black lines indicate F1-F4 subdomains; dotted black lines indicate where the control subdomain would be.

Fig. 7.

Proposed adult and embryonic eCNm subdomains, and alterations seen in En1/2 conditional mutants. (A) Table listing proposed eCNm subdomains based on marker expression and spatial location in normal adult brain (based on Fujita et al., 2020) and at E17.5 in normal and En1/2 CKOs (this study). Note that most proteins are expressed in a subset of cells in the four domains. Brackets around SNCA indicate where we could not detect expression. (B) Schematics of the eCNm in the adult (left) and the NTZ (E14.5) or eCNm (E17.5) in embryos showing how the expression of some spatially defined marker proteins evolves during development and which subdomains are altered in En1/2 CKOs. Solid black lines indicate F1-F4 subdomains; dotted black lines indicate where the control subdomain would be.

TBR2 expression in GCPs and UBCs is dependent on En1/2 function

Given that one of the earliest phenotypes of En1/2 loss in the eCNm is a pronounced reduction of TBR2 expression, we examined whether the dependency of Tbr2 expression on EN1/2 is recapitulated in the other two excitatory cerebellar cell types. Interestingly, we found that at E14.5 and E17.5 TBR2 is normally expressed in a subset of GCPs located toward the inner edge of the external granule cell layer (Fig. S16). TBR2 is also expressed in all UBCs from their birth to adulthood and these cells reside in the adult internal granule cell layer (IGL), enriched in posterior vermis lobules 9/10 (Englund et al., 2006).

We first determined whether TBR2 expression was lost in En1/2 deficient UBCs at E17.5 in eCNm-En1/2-tdT CKO animals (Fig. S17), as the eCN+GCP-Cre transgene does not recombine in UBCs (Fig. S18). Strikingly, we found that TBR2 expression was greatly reduced in eCNm-En1/2-tdT CKO embryonic UBCs (Fig. S17A-F) and in the IGL of adults compared with their respective controls (Fig. 8A-D). There were also fewer CALB2+ UBCs (marker of mature Type 1 TBR2+ UBCs) in adult eCNm-En1/2-tdT CKO mice (Fig. 8E-H). Most of the tdT+ TBR2+/CALB2+ cells (UBCs) present in adult mutants were ectopically positioned outside the IGL in the white matter of lobule 10 or close to the CN, and any UBCs in the IGL appeared smaller than normal (Fig. 8I-K) and were concentrated in the ventral portion of lobule 10 and absent from lobule 9. We confirmed the tdT+ ectopic cells were not eCN by showing they did not express MEIS2 in adult mutants (Fig. S2). These results indicate that, although some UBCs can survive in the absence of En1/2, they are unable to mature normally or migrate to their proper position. The phenotype is likely associated with the loss of Tbr2 expression, because Tbr2 brain-specific conditional mutants (Nes11Cre) have very few UBCs in adults, as determined using three markers, including CALB2, and exhibit ectopic location of the developing mutant UBCs at postnatal day (P) 10 and depletion in lobule 9 (McDonough et al., 2021).

Fig. 8.

Deletion of En1/2 results in reduction of TBR2 expression and differentiation defects in RL-derived GCPs and UBCs. (A-H) IF images of TBR2 and tdT (A-D) or CALB2 and tdT (E-H) expression in lobule 10 of 11-week cerebella. Dashed rectangles (A,B,E,F) indicate regions from which images in C,D,G,H were taken. (I-K) Quantification of tdT+ TBR2+ (I) or tdT+ CALB2+ (J) UBCs in eCNm-En1/2-tdT CKOs and eCNm-tdT controls in indicated lobules (Lob) or white matter and UBC nucleus size in lobule 10 (K) (n=3 animals per genotype, unpaired, two-tailed Student's t-test). Error bars represent s.d. (L-O) IF images of TBR2+ (L,M) or PAX6+ (N,O) GCPs in the external granule cell layer (EGL; white outlines) of E14.5 En1/2f/f and eCN+GCP-En1/2-tdT CKO cerebella. (P,Q) Quantification of PAX6+ GCPs (P) and TBR2+ GCPs (Q) (n=4 per genotype, unpaired, two-tailed Student's t-test) at E14.5 in two genotypes. Error bars represent s.d. (R) UMAP visualization of reclustering of the Pax6-enriched clusters shown in Fig. 2A. (S) UMAP visualization with cells grouped by cell cycle phase. (T) Otx2 and Tbr2 expression. (U) Volcano plot showing differentially expressed genes in G0/G1-phase GCPs between two genotypes. Red dots indicate genes significantly up- or downregulated (P≤0.05); black dots indicate non-significantly enriched genes. (V) Over-representation analysis of downregulated genes in G0/G1-phase GCPs of mutants compared with controls. (W) Violin plots showing expression of synaptic genes in G0/G1-phase GCPs of both genotypes. Scale bars: 50 µm.

Fig. 8.

Deletion of En1/2 results in reduction of TBR2 expression and differentiation defects in RL-derived GCPs and UBCs. (A-H) IF images of TBR2 and tdT (A-D) or CALB2 and tdT (E-H) expression in lobule 10 of 11-week cerebella. Dashed rectangles (A,B,E,F) indicate regions from which images in C,D,G,H were taken. (I-K) Quantification of tdT+ TBR2+ (I) or tdT+ CALB2+ (J) UBCs in eCNm-En1/2-tdT CKOs and eCNm-tdT controls in indicated lobules (Lob) or white matter and UBC nucleus size in lobule 10 (K) (n=3 animals per genotype, unpaired, two-tailed Student's t-test). Error bars represent s.d. (L-O) IF images of TBR2+ (L,M) or PAX6+ (N,O) GCPs in the external granule cell layer (EGL; white outlines) of E14.5 En1/2f/f and eCN+GCP-En1/2-tdT CKO cerebella. (P,Q) Quantification of PAX6+ GCPs (P) and TBR2+ GCPs (Q) (n=4 per genotype, unpaired, two-tailed Student's t-test) at E14.5 in two genotypes. Error bars represent s.d. (R) UMAP visualization of reclustering of the Pax6-enriched clusters shown in Fig. 2A. (S) UMAP visualization with cells grouped by cell cycle phase. (T) Otx2 and Tbr2 expression. (U) Volcano plot showing differentially expressed genes in G0/G1-phase GCPs between two genotypes. Red dots indicate genes significantly up- or downregulated (P≤0.05); black dots indicate non-significantly enriched genes. (V) Over-representation analysis of downregulated genes in G0/G1-phase GCPs of mutants compared with controls. (W) Violin plots showing expression of synaptic genes in G0/G1-phase GCPs of both genotypes. Scale bars: 50 µm.

Our previous analyses of En1/2 mutant GCPs during the first postnatal week showed a mild diminution of differentiation compared with controls (mosaic analysis at P2-P8) (Willett et al., 2019). Interestingly, on sections of E14.5 and E17.5 eCN+GCP-En1/2-tdT CKO cerebella, few TBR2+ PAX6+ GCPs were detected despite PAX6 cell number being normal (Fig. 8L-Q, Fig. S19). We then combined the Pax6+ clusters of E14.5 scRNA-seq tdT+ cells from eCN+GCP-En1/2-tdT CKO and control embryos and then reclustered the cells, which yielded seven clusters that did not appear to be related to GCP location in the anterior-posterior axis because the posterior gene Otx2 was expressed in all clusters (Fig. 8R-T). When we grouped GCPs by their cell cycle stage and performed differential gene expression analysis between eCN+GCP-En1/2-tdT CKOs and controls, over-representation analyses of downregulated genes in G0/G1 phase En1/2 CKO GCP clusters revealed the En1/2 deficient GCPs have a significant downregulation of genes and GO terms related to synapse formation and cell–cell adhesion (Fig. 8U-W, Tables S12, S13). Taken together, these results show that EN1/2 positively regulate TBR2 expression in the three rhombic lip lineages. However, the severity of the consequences of En1/2 loss and subsequent downstream TBR2 loss varies by cell type. Interestingly, by analyzing conditional mutants we found that UBCs are the most dependent on Tbr2, as eCN+GCP-Tbr2 CKO mice (Atoh1-Cre/+; Tbr2lox/lox) do not have defects in eCN or granule cell survival (Fig. S20), consistent with the previous report of only a UBC defect in Tbr2 conditional mutants (McDonough et al., 2021).

In this study, we uncovered the underpinnings of embryonic eCNm subpopulation molecular diversity and defined overlapping and distinct roles for EN1/2 in all three cerebellar RL lineages. We first showed that the molecular identity of the posteriorly located subpopulation of the eCNm lost in En1/2 CKOs (Willett et al., 2019; Fig. 1, Fig. S1) is defined by Snca expression medially and SPP1 expression laterally, relating to two of the adult eCNm domains recently described (F2 and F4 in Fujita et al., 2020) (Fig. 7). scRNA-seq and spatial protein expression analysis in a transgenic mouse Cre line that labels the eCNm at E14.5 revealed that the transcriptome of the eCNm is already distinct from that of the eCNi+eCNl one day after eCN neurogenesis is complete. Additionally, we define eCNm molecular subpopulations based on expression of marker proteins restricted to spatial domains that evolve between E14.5 and E17.5 (Fig. 7). scRNA-seq analysis of En1/2 CKO eCNm indicates a role for EN1/2 in promoting neural differentiation, particularly synaptogenesis. Furthermore, eCNm subpopulations are differentially vulnerable to loss of En1/2, with a TBR2+ anterior eCNm subpopulation dependent on EN1/2 for their molecular identity, whereas a posterior BARHL1+ subpopulation relies on EN1/2 for viability (Fig. 7). Revealing global roles for EN1/2 in cerebellar excitatory neurons, we show that TBR2 expression and neural differentiation in GCPs and UBCs are also dependent on En1/2, with UBCs also requiring En1/2 for viability.

Although molecular subpopulations of adult eCN have been documented (Fujita et al., 2020; Kebschull et al., 2020), how the populations emerge during development has not been understood. Complicating an ability to perform studies to draw direct connections between the embryonic and adult eCNm subpopulations, two of the three ‘steady state’ markers of adult eCNm subpopulations are not similarly represented at embryonic stages (Fig. S4): SPP1 is not expressed in the embryo and Snca is broadly expressed. Be defining four molecular subdomains of the embryonic eCNm and determining which are lost in En1/2 CKOs, we identified two posterior E17.5 subpopulations of the eCNm labeled by BARHL1 that might relate to the adult posterior domains expressing SPP1 (F2) and Snca (F4). In addition, a distinct anterior eCNm domain marked by CALB2 in the embryo and the adult (F3) is not lost in the En1/2 CKOs, suggesting they might be similar domains. Interestingly, the anterior eCNm subdomains (F1 and F3), in which cell number is reduced in adult En1/2 CKOs, do not appear to have a similar decrease in number at E17.5 (Fig. 7). The anterior eCNm thus might become sensitive to loss of En1/2 for their viability later than the posterior domains. Our results thus reveal that acquisition of subpopulation identity and vulnerability to En1/2 loss occurs progressively soon after the eCNm cells reach the NTZ. However, the expression domains of genes that determine a particular subpopulation evolve between E14.5 and E17.5 and most embryonic genes are not expressed in the adult eCNm and, thus, the transcriptome of the eCNm must continue to be refined after birth. Inducible fate mapping with new genetic tools are needed to determine the precise relationships between the domains defined in the embryonic and adult eCNm.

Our study highlights that transcription factors can play multiple roles during the development of a cell type. We show that En1/2 are initially responsible in the E14.5 eCNm for spatial patterning (in a TBR2+ anterior domain) and enhancing synaptic gene expression, followed by a role in survival (in BARHL1+ posterior domains). Similar roles have been documented for the Engrailed genes in several species. For example, in the fly, en is responsible for directing CNS midline glia to a posterior, non-axon ensheathing fate by repressing anterior, axon-ensheathing glial genes (Watson et al., 2011). Additionally, en in the fly and En1/2 in the mouse are required for the proper development of serotonergic neurons (Fox and Deneris, 2012; Lundell et al., 1996). In both species, the Engrailed genes control expression of genes crucial for serotonin synthesis, and in the mouse loss of En1/2 results in apoptosis of 60% of dorsal raphe nucleus neurons starting at P10 (Fox and Deneris, 2012). However, whether the neurons that die belong to a molecularly defined subpopulation(s) remains to be determined. Additionally, en in the fly auditory system controls synapse formation and specificity of a subset of neurons (Pézier et al., 2014). In the mouse, En1 is required in a subset of spinal cord interneurons to make the appropriate number of recurrent inhibitory connections to motor neurons (Sapir et al., 2004). How EN1/2 mediate transcriptional control over these processes, however, remains to be elucidated.

EN1/2 are thought to primarily act as repressors (Jaynes and O′Farrell, 1991; Smith and Jaynes, 1996; Tolkunova et al., 1998) and thus absence of En1/2 would be expected to lead to upregulation of genes. Therefore, the downregulation of synaptic genes we observed in eCNm could be due to an indirect process via the repression of a repressor. However, En in the fly has been shown to act as an activator in the presence of its co-factor, Extradenticle (PBX proteins in vertebrates) (Serrano and Maschat, 1998). Moreover, chromatin immunoprecipitation data of genes regulated by both En and Gooseberry-neuro (PAX3/7 in vertebrates) in the ventral nerve cord of the fly, which act in concert to drive posterior commissure crossing, showed that most combined targets are activated, rather than repressed (Bonneaud et al., 2017). Most target genes belong to nervous system development GO categories, including axon guidance (Bonneaud et al., 2017). Thus, it is possible that EN1/2 in the mouse eCNm activate synaptic gene expression directly in the presence of unidentified partner transcription factors or co-factors.

We discovered that En1/2 are required for the activation of TBR2 expression in all three rhombic lip-derived excitatory neurons, indicating one general mechanism by which En1/2 likely controls neural differentiation. In the cerebral cortex, TBR2 and TBR1 play sequential roles in neural differentiation and acquisition of laminar fate (Bedogni et al., 2010; Mihalas et al., 2016). Similar roles in controlling neural differentiation have been suggested for the proteins in the NTZ based on their expression patterns at E13.5 (Fink et al., 2006). However, our study shows that TBR1 marks the majority (∼80%) of eCNm cells at E14.5, whereas TBR2 is only expressed in ∼20% of the eCNm with an ∼60% overlap, suggesting that TBR1 is the main player in eCNm development. Consistent with this, we found that conditional loss of Tbr2 using Atoh1-Cre does not alter eCN or GC number, and adult cerebellar morphology and size appear normal (Fig. S20; see also McDonough et al., 2021). However, these results do not rule out an important role for the TBR family in eCNm development as TBR1 is present in most TBR2+ eCNm cells by E17.5. In contrast, TBR2 plays a pivotal role in the differentiation, survival and migration of the UBCs, whereas TBR1 is not expressed (Fig. S17), and TBR2 expression is maintained from the time the UBCs are born into adulthood (Fig. 8, Fig. S17; Englund et al., 2006; McDonough et al., 2021). Strikingly En1/2-deficient adult UBCs have a similar, but possibly milder, phenotype than that reported for Tbr2 brain-specific conditional mutants (McDonough et al., 2021). TBR1 is not upregulated in En1/2-deficient UBCs as a possible means of compensation (Fig. S17). Curiously, TBR2 does not play a major role in development of GCPs, despite a lack of upregulation of TBR1 expression in En1/2 CKO GCPs (Fig. S19). Thus, EN1/2 and TBR2 have distinct roles in the three excitatory neuron types, and EN1/2 are expressed upstream of TBR2, but seemingly not Tbr1 or Pax6, another homeobox gene with differential roles in these cells (Engelkamp et al., 1999; Swanson and Goldowitz, 2011; Swanson et al., 2005).

In conclusion, our study sheds light on how molecularly defined neuron subpopulations emerge, and how transcription factors that initially pattern a large region of tissue early in development can be crucial later in defining subpopulations of cells in a lineage and then for their differentiation and/or homeostasis.

Animals

All animal experiments were performed in accordance with protocols approved and guidelines provided by the Memorial Sloan Kettering Cancer Center's Institutional Animal Care and Use Committee (IACUC). Animals were given access to food and water ad libitum and were housed on a 12 h light/dark cycle.

The following mouse lines were used in this study: En1lox (Sgaier et al., 2007), En2lox (Cheng et al., 2010), Atoh1-Cre (Matei et al., 2005), R26LSL-ntdTom (Quina et al., 2017)(Ai75D, The Jackson Laboratory, stock no: 025106), Selenow-Cre (Gerfen et al., 2013) and Tbr2lox (Intlekofer et al., 2008). Primers used for genotyping are listed in Table S14. Animals were maintained on an outbred Swiss Webster background. Both sexes were used for all analyses and no randomization was used. Noon of the day a vaginal plug was detected was designated as developmental stage E0.5.

Tissue processing

The brains of all embryonic stages were dissected in ice cold phosphate-buffered saline without calcium and magnesium (PBS) and immersion fixed in 4% paraformaldehyde (PFA) for 48 h (for E14.5) and 72 h for E17.5 at 4°C on a shaker. Adult animals were anesthetized with isoflurane and then transcardially perfused with ice-cold PBS followed by 4% PFA, and brains were post-fixed in 4% PFA overnight. Specimens for cryosectioning were placed in 30% sucrose in PBS until they sank, embedded in OCT (Tissue-Tek), frozen in dry-ice-cooled isopentane, and sectioned on a cryostat (Leica, CM3050S) in the sagittal plane at 12 μm for embryos and 14 μm for adults and collected on glass slides (Fisherbrand). Tissue was stored at −20°C.

Immunofluorescence

Primary and secondary antibodies and their related concentrations are listed in Table S14. Sections of cryosectioned tissue were air-dried for 10 min and then washed in PBS for 10 min. For IF with all antibodies, sections were subjected to antigen retrieval using sodium citrate buffer (10 mM sodium citrate with 0.05% Tween-20, pH 6.0) at 95°C for 15 min (embryonic tissue) or 20 min (adult tissue). After antigen retrieval, sections were brought to room temperature. Sodium citrate buffer was discarded and then sections underwent two washes in 0.1% PBST (1× PBS with 0.1% Triton X-100) for 5 min each. Slides were then blocked with blocking buffer [5% bovine serum albumin (Sigma-Aldrich) in 0.1% PBST] at room temperature for 1 h. Then, primary antibodies diluted in blocking buffer were placed on slides overnight at 4°C. Slides were washed three times in 0.1% PBST for 5 min each and then incubated in Alexa Fluor-conjugated secondary antibodies diluted at 1:500 in blocking buffer for 1 h at room temperature. Counterstaining was performed using Hoechst 33258 (1:1000, Invitrogen A-21422). Slides were then washed three times in PBS for 5 min each prior to cover slipping using Fluorogel mounting medium (Electron Microscopy Sciences).

RNA in situ hybridization

Probes were in vitro transcribed from PCR-amplified templates prepared from cDNA synthesized from postnatal cerebellum lysate. Primers used for PCR amplification are listed in Table S14. Primers were flanked in the 5′ with SP6 (antisense) and T7 (sense) promoters. Specimen treatment and hybridization were performed as described previously (Blaess et al., 2011).

Hematoxylin and Eosin staining

Hematoxylin and Eosin (Thermo Fisher Scientific) staining was performed according to instructions from the manufacturer for histological analysis and cerebellar area measurements of adult sections.

Nickel-enhanced DAB staining

eCN counting for adult eCN+GCP-Tbr2 CKOs was performed using nickel-enhanced 3,3'-diaminobenzidine (Ni-DAB) immunohistochemistry for NeuN (Rbfox3) as previously described (Willett et al., 2019).

TUNEL assay

Slides were first permeabilized with 0.5% Triton X-100, and pre-incubated with Tdt buffer (30 mM Tris-HCl, 140 mM sodium cacodylate and 1 mM CoCl2) for 15 min at room temperature. Slides were then incubated in a TUNEL reaction solution [containing terminal transferase (Roche, 3333574001) and Biotin-16-dUTP (Sigma-Aldrich, 11093070910)] for 1 h at 37°C. Following the TUNEL reaction, slides were incubated in a Streptavidin Alexa Fluor 647 conjugate (Invitrogen, S-32357) for 1 h. Slides were then washed in PBS twice for 5 min each followed by a 10 min incubation in Hoechst 33258 (1:1000; Invitrogen, A-21422). Prior to coverslipping using Fluorogel mounting medium (Electron Microscopy Science), slides were washed with PBS twice for 5 min each.

Microscopy

All images were collected with a DM6000 Leica fluorescent microscope or NanoZoomer Digital Pathology microscope (Hamamatsu Photonics) and processed using ImageJ (NIH) or Photoshop (Adobe) software. Image quantification, including area measurements and cell counting was performed using ImageJ (NIH). The polygon selection tool (ImageJ) was used to select areas of interest for all cerebellar area measurements, and the Cell Counter plugin (ImageJ) was used to quantify all cells (aside from adult eCN).

Embryonic eCN and GCP quantification

For all embryonic eCN and GCP quantification, every tenth section from a hemisphere section where the CN first appeared to the midline was quantified. Cells were only quantified if they expressed both the marker and tdT.

UBC quantification

For adult UBC quantification, an average of tdT+ and TBR2+ or CALB2+ cells was calculated from quantifying lobules 9 and 10 in three sections per mouse. Nuclei were measured using the ‘oval’ tool in ImageJ. From each section, 33 or 34 cells were measured, either from lobule 10 or the adjacent white matter near the CN (for eCNm-En1/2-tdT CKOs).

Adult eCN quantification

Adult eCN quantification was performed using NeuN-labeled slides (every other section for eCN+GCP-Tbr2 CKOs) using a semi-automated method described by Willett et al. (2019). Briefly, eCN from scanned images (NanoZoomer Digital Pathology) were cropped, pre-processed in ImageJ and only particles sized 100-600 um2 were quantified.

Adult cerebellar area measurements

For all adult sector and IGL areas, Hematoxylin and Eosin-stained slides were used. Three sagittal sections at midline per animal were used, and values across these sections were averaged. Areas were measured using ImageJ with the freehand selection tool.

3D reconstructions

Every other section (12 μm thick) from a consecutive sagittal series was used for 3D reconstruction from where the eCN appeared laterally to where they disappeared medially. eCNm images were cropped in Adobe Photoshop, and then imported into ImageJ. The images were registered using the StackReg plug-in in ImageJ. Minor adjustments were made manually only where needed. Registered images were imported into Adobe Illustrator and each image was assigned its own layer. Each region consisting of the eCNm, marked by tdT, was outlined in Adobe Illustrator. Aligned outlines were imported into Rhinoceros 7 and were spaced evenly by distance between sections using the SetPt command. The final 3D reconstruction was made using the Loft command (parameters: 20 control points, loose). Dorsal views of rendered images were outlined in Adobe Illustrator (using the Image Trace function) and colored for schematics.

Statistical analyses

Prism (GraphPad) was used for all statistical analyses except for genomics analyses. Statistical comparisons of two populations (genotypes) were carried out with Student's two-tailed t-test and two-way ANOVA. Post-hoc analyses of two-way ANOVA was performed using Šídák multiple comparison tests. The statistical significance cutoff was set at P≤0.05. Population statistics are represented either as mean±s.d. or as mean±s.e.m. No statistical methods were used to predetermine the sample size, but sample sizes were similar to those generally employed in the field. At least three mice were used for each experiment and the numbers for animals used for each experiment are stated in the figure legends.

Sample preparation for scRNA-seq

Animals (numbers of embryos pooled for each genotype are listed in relevant figures) were dissected in ice-cold Hank's balanced salt solution (HBSS) (Gibco) and were pooled together for downstream steps. Two replicate experiments were performed. On ice, cerebella were detached from the rest of the brain using forceps and dissociated in Accutase (Innovative Cell Technologies) at 37°C for 7-10 min. Following dissociation, Accutase was washed out using neural stem cell (NSC) media (Neurobasal, supplemented with N2, B27 and non-essential amino acids, Gibco). Samples were triturated by pipetting up and down using a P1000, filtered through a 35 μm cell strainer (Falcon) and then centrifuged at 500 g at 4°C for 5 min. Cells were washed again in NSC media and were finally filtered into polypropylene round-bottom tubes (Falcon) prior to sorting for tdT+ cells. tdT+ cells were sorted into polypropylene round-bottom tubes, spun down at 500 g at 4°C for 5 min and resuspended in NSC media prior to single-cell library preparation.

Library preparation and sequencing

Single cell libraries were prepared using 10x Genomics' Chromium Next GEM Single Cell 3′ Reagent Kits, v3.1, according to the manufacturer's instructions. Prior to loading the 10x Genomics chip, the limited sample preparation protocol was used. Cell viability was quantified using 0.2% (w/v) Trypan Blue on a hemocytometer, and 7000 viable cells were targeted per pooled sample.

After PicoGreen quantification and quality control using an Agilent TapeStation, libraries were pooled at equimolar concentrations and run on a NovaSeq 6000 in a PE28/91 run, using the NovaSeq 6000 S1 Reagent Kit (100 Cycles) (Illumina).

scRNA-seq analyses

The Cell Ranger Single Cell software suite from 10x Genomics was used to align reads and generate feature-barcode matrices. The reference genome used was the Genome Reference Consortium Mouse Build 38 (mm10). Raw reads were processed using the Cell Ranger count program using default parameters. Note that the number of mutant cells that passed quality control was lower than for control cells. This might be because some of the mutant cells are more susceptible to dying during the cell isolation and the first step in the 10x process, as soon after this stage the mutant cells began dying in vivo.

Seurat v4.1.1 was used to generate a UMI (unique molecular identifier) count matrix from the Cell Ranger output (Butler et al., 2018; Macosko et al., 2015). Genes expressed in fewer than ten cells were removed from further analyses. Cells that had fewer than 500 UMIs, 250 genes or had over 30% of their UMIs mapped to mitochondrial genes were considered low quality/outliers and discarded from the datasets. Normalization was performed on individual samples using the scTransform function, with regression of Xist. Samples were then integrated using the PrepSCTIntegration, FindIntegrationAnchors and the IntegrateData functions. For all analyses, the first 25 dimensions were used for the FindNeighbors function, and clusters were identified using the FindClusters function with a resolution of 0.5. Data were projected into the 2D space using the FindUMAP function with 25 dimensions. Cluster markers and further differential gene expression analyses were all performed on the RNA assay, following normalization using the NormalizeData functions. Cluster markers were identified using the FindAllMarkers function (for eCN+GCP-tdT or eCN+GCP-En1/2-tdT CKO only analyses) or FindConserved markers (for joint control and eCN+GCP-En1/2-tdT CKO analyses) and comparing markers generated to existing literature on the embryonic cerebellum and analysis of the Allen Developing Brain Expression Atlas. To refine clustering further, the SubsetData function was used to create a new Seurat object, and the above clustering was reiterated. For the eCN+GCP-En1/2-tdT CKO only clustering, selection of clusters for downstream reclustering as well as final mapping of the eCN clusters was performed using the FindTransferAnchors and TransferData functions with the corresponding eCN+GCP-tdT dataset as the reference dataset.

Differential gene expression analysis between controls and En1/2 CKOs was performed using Libra (Squair et al., 2021) using the edgeR-LRT pseudobulk method. Genes that showed an adjusted P≤0.05 were considered significantly up- or downregulated. Results were visualized using the EnhancedVolcano package on Bioconductor. GO analyses were performed separately on up- and downregulated genes using the clusterProfiler (Wu et al., 2021; Yu et al., 2012) package in Bioconductor.

We thank past and present members of the Joyner lab for helpful comments throughout the work. We also thank the Memorial Sloan Kettering Cancer Center Flow Cytometry Core, Integrated Genomics Operations Core Facility and the Center for Comparative Medicine and Pathology for technical support. We acknowledge the use of the Integrated Genomics Operation Core, funded by an NCI Cancer Center Support Grant (CCSG, P30 CA08748), Cycle for Survival, and the Marie-Josée and Henry R. Kravis Center for Molecular Oncology. We are grateful to Natalia De Marco Garcia for the Selenow-Cre mice and to Joseph Sun for the Tbr2lox mice. We thank Thomas Vierbuchen for valuable feedback and Daniel Medina Cano for assistance with using the 10x Chromium Controller. We thank Armita R. Manafzadeh for assistance with 3D reconstruction of the eCNm.

Author contributions

Conceptualization: A.K., A.L.J.; Methodology: A.K., N.S.B., A.L.J.; Validation: A.K., A.S.L., A.L.J.; Formal analysis: A.K., A.S.L.; Investigation: A.K., A.S.L., D.N.S., O.N., Z.L.; Resources: A.S.L., A.L.J.; Data curation: A.K., A.L.J.; Writing - original draft: A.K., A.L.J.; Writing - review & editing: A.K., A.S.L., N.S.B., A.L.J.; Visualization: A.K., A.L.J.; Supervision: A.L.J.; Project administration: A.L.J.; Funding acquisition: A.L.J.

Funding

This work was supported by grants from the National Institute of Neurological Disorders and Stroke (R01NS092096 to A.L.J.), the National Institute of Mental Health (R37MH085726 to A.L.J.), the National Institute of Child Health and Human Development (T32HD060600 to A.S.L.) and a National Cancer Institute Cancer Center Support Grant (P30 CA008748-48). The work was also supported by Cycle for Survival funds, in conjunction with the Memorial Sloan-Kettering Cancer Center. A.K. was supported by a Dorris J. Hutchison Pre-doctoral Fellowship in conjunction with the Memorial Sloan-Kettering Cancer Center. A.S.L. was also supported by a Weill Cornell Medicine Clinical & Translational Science Center Predoctoral Training Award (TL1TR002386) from the National Center for Advancing Translational Sciences. N.S.B. was supported by postdoctoral fellowships from New York State Stem Cell Science (NYSTEM) (C32599GG) and the National Institute of Neurological Disorders and Stroke (K99/R00 NS112605-01). O.N. was supported by the McNulty Scholars Program at Hunter College. Deposited in PMC for release after 12 months.

Data availability

scRNA-seq data are available in the Gene Expression Omnibus database under accession number GSE246894. Original 3D reconstruction files can be accessed at 10.6084/m9.figshare.24650286. We did not generate any custom code for the analysis.

Altieri
,
S. C.
,
Jalabi
,
W.
,
Zhao
,
T.
,
Romito-DiGiacomo
,
R. R.
and
Maricich
,
S. M.
(
2015
).
En1 directs superior olivary complex neuron positioning, survival, and expression of FoxP1
.
Dev. Biol.
408
,
99
-
108
.
Altman
,
J.
and
Bayer
,
S. A.
(
1985
).
Embryonic development of the rat cerebellum. II. Translocation and regional distribution of the deep neurons
.
J. Comp. Neurol.
231
,
27
-
41
.
Apps
,
R.
,
Hawkes
,
R.
,
Aoki
,
S.
,
Bengtsson
,
F.
,
Brown
,
A. M.
,
Chen
,
G.
,
Ebner
,
T. J.
,
Isope
,
P.
,
Jörntell
,
H.
,
Lackey
,
E. P.
et al.
(
2018
).
Cerebellar modules and their role as operational cerebellar processing units
.
Cerebellum
17
,
654
-
682
.
Bedogni
,
F.
,
Hodge
,
R. D.
,
Elsen
,
G. E.
,
Nelson
,
B. R.
,
Daza
,
R. A. M.
,
Beyer
,
R. P.
,
Bammler
,
T. K.
,
Rubenstein
,
J. L. R.
and
Hevner
,
R. F.
(
2010
).
Tbr1 regulates regional and laminar identity of postmitotic neurons in developing neocortex
.
Proc. Natl Acad. Sci. USA
107
,
13129
-
13134
.
Blaess
,
S.
,
Bodea
,
G. O.
,
Kabanova
,
A.
,
Chanet
,
S.
,
Mugniery
,
E.
,
Derouiche
,
A.
,
Stephen
,
D.
and
Joyner
,
A. L.
(
2011
).
Temporal-spatial changes in Sonic Hedgehog expression and signaling reveal different potentials of ventral mesencephalic progenitors to populate distinct ventral midbrain nuclei
.
Neural Dev.
6
,
29
.
Bonneaud
,
N.
,
Layalle
,
S.
,
Colomb
,
S.
,
Jourdan
,
C.
,
Ghysen
,
A.
,
Severac
,
D.
,
Dantec
,
C.
,
Nègre
,
N.
and
Maschat
,
F.
(
2017
).
Control of nerve cord formation by Engrailed and Gooseberry-Neuro: a multi-step, coordinated process
.
Dev. Biol.
432
,
273
-
285
.
Butler
,
A.
,
Hoffman
,
P.
,
Smibert
,
P.
,
Papalexi
,
E.
and
Satija
,
R.
(
2018
).
Integrating single-cell transcriptomic data across different conditions, technologies, and species
.
Nat. Biotechnol.
36
,
411
-
420
.
Cheng
,
Y.
,
Sudarov
,
A.
,
Szulc
,
K. U.
,
Sgaier
,
S. K.
,
Stephen
,
D.
,
Turnbull
,
D. H.
and
Joyner
,
A. L.
(
2010
).
The Engrailed homeobox genes determine the different foliation patterns in the vermis and hemispheres of the mammalian cerebellum
.
Development
137
,
519
-
529
.
Chizhikov
,
V. V.
,
Lindgren
,
A. G.
,
Currle
,
D. S.
,
Rose
,
M. F.
,
Monuki
,
E. S.
and
Millen
,
K. J.
(
2006
).
The roof plate regulates cerebellar cell-type specification and proliferation
.
Development
133
,
2793
-
2804
.
Corrales
,
J. D.
,
Rocco
,
G. L.
,
Blaess
,
S.
,
Guo
,
Q.
and
Joyner
,
A. L.
(
2004
).
Spatial pattern of sonic hedgehog signaling through Gli genes during cerebellum development
.
Development
131
,
5581
-
5590
.
Engelkamp
,
D.
,
Rashbass
,
P.
,
Seawright
,
A.
and
van Heyningen
,
V.
(
1999
).
Role of Pax6 in development of the cerebellar system
.
Development
126
,
3585
-
3596
.
Englund
,
C.
,
Kowalczyk
,
T.
,
Daza
,
R. A. M.
,
Dagan
,
A.
,
Lau
,
C.
,
Rose
,
M. F.
and
Hevner
,
R. F.
(
2006
).
Unipolar brush cells of the cerebellum are produced in the rhombic lip and migrate through developing white matter
.
J. Neurosci.
26
,
9184
-
9195
.
Fleming
,
J. T.
,
He
,
W.
,
Hao
,
C.
,
Ketova
,
T.
,
Pan
,
F. C.
,
Wright
,
C. C. V.
,
Litingtung
,
Y.
and
Chiang
,
C.
(
2013
).
The Purkinje neuron acts as a central regulator of spatially and functionally distinct cerebellar precursors
.
Dev. Cell
27
,
278
-
292
.
Fink
,
A. J.
,
Englund
,
C.
,
Daza
,
R. A. M.
,
Pham
,
D.
,
Lau
,
C.
,
Nivison
,
M.
,
Kowalczyk
,
T.
and
Hevner
,
R. F.
(
2006
).
Development of the deep cerebellar nuclei: transcription factors and cell migration from the rhombic lip
.
J. Neurosci.
26
,
3066
-
3076
.
Fox
,
S. R.
and
Deneris
,
E. S.
(
2012
).
Engrailed is required in maturing serotonin neurons to regulate the cytoarchitecture and survival of the dorsal raphe nucleus
.
J. Neurosci.
32
,
7832
-
7842
.
Fujita
,
H.
,
Kodama
,
T.
and
du Lac
,
S.
(
2020
).
Modular output circuits of the fastigial nucleus for diverse motor and nonmotor functions of the cerebellar vermis
.
eLife
9
,
e58613
.
Gerfen
,
C. R.
,
Paletzki
,
R.
and
Heintz
,
N.
(
2013
).
GENSAT BAC cre-recombinase driver lines to study the functional organization of cerebral cortical and basal ganglia circuits
.
Neuron
80
,
1368
-
1383
.
Glickstein
,
M.
and
Voogd
,
J.
(
1995
).
Lodewijk Bolk and the comparative anatomy of the cerebellum
.
Trends Neurosci.
18
,
206
-
210
.
Intlekofer
,
A. M.
,
Banerjee
,
A.
,
Takemoto
,
N.
,
Gordon
,
S. M.
,
DeJong
,
C. S.
,
Shin
,
H.
,
Hunter
,
C. A.
,
Wherry
,
E. J.
,
Lindsten
,
T.
and
Reiner
,
S. L.
(
2008
).
Anomalous type 17 response to viral infection by CD8+ T cells lacking T-bet and eomesodermin
.
Science
321
,
408
-
411
.
Jaynes
,
J. B.
and
O'Farrell
,
P. H.
(
1991
).
Active repression of transcription by the engrailed homeodomain protein
.
EMBO J.
10
,
1427
-
1433
.
Joyner
,
A. L.
(
1996
).
Engrailed, Wnt and Pax genes regulate midbrain-hindbrain development
.
Trends Genet.
12
,
15
-
20
.
Joyner
,
A. L.
and
Bayin
,
N. S.
(
2022
).
Cerebellum lineage allocation, morphogenesis and repair: impact of interplay amongst cells
.
Development
149
,
dev185587
.
Joyner
,
A. L.
,
Herrup
,
K.
,
Auerbach
,
B. A.
,
Davis
,
C. A.
and
Rossant
,
J.
(
1991
).
Subtle cerebellar phenotype in mice homozygous for a targeted deletion of the En-2 homeobox
.
Science
251
,
1239
-
1243
.
Ju
,
J.
,
Liu
,
Q.
,
Zhang
,
Y.
,
Liu
,
Y.
,
Jiang
,
M.
,
Zhang
,
L.
,
He
,
X.
,
Peng
,
C.
,
Zheng
,
T.
,
Lu
,
Q. R.
et al. (
2016
).
Olig2 regulates Purkinje cell generation in the early developing mouse cerebellum
.
Sci. Rep.
6
,
30711
.
Judd
,
E. N.
,
Lewis
,
S. M.
and
Person
,
A. L.
(
2021
).
Diverse inhibitory projections from the cerebellar interposed nucleus
.
eLife
10
,
e66231
.
Kebschull
,
J. M.
,
Richman
,
E. B.
,
Ringach
,
N.
,
Friedmann
,
D.
,
Albarran
,
E.
,
Kolluru
,
S. S.
,
Jones
,
R. C.
,
Allen
,
W. E.
,
Wang
,
Y.
,
Cho
,
S. W.
et al.
(
2020
).
Cerebellar nuclei evolved by repeatedly duplicating a conserved cell-type set
.
Science
370
,
eabd5059
.
Kebschull
,
J. M.
,
Casoni
,
F.
,
Consalez
,
G. G.
,
Goldowitz
,
D.
,
Hawkes
,
R.
,
Ruigrok
,
T. J. H.
,
Schilling
,
K.
,
Wingate
,
R.
,
Wu
,
J.
,
Yeung
,
J.
et al.
(
2023
).
Cerebellum lecture: the cerebellar nuclei-core of the cerebellum
.
Cerebellum
23
,
620
-
677
.
Kuemerle
,
B.
,
Zanjani
,
H.
,
Joyner
,
A.
and
Herrup
,
K.
(
1997
).
Pattern deformities and cell loss in Engrailed-2 mutant mice suggest two separate patterning events during cerebellar development
.
J. Neurosci.
17
,
7881
-
7889
.
Lee
,
A. S.
,
Arefin
,
T. M.
,
Gubanova
,
A.
,
Stephen
,
D. N.
,
Liu
,
Y.
,
Lao
,
Z.
,
Krishnamurthy
,
A.
,
De Marco García
,
N. V.
,
Heck
,
D. H.
,
Zhang
,
J.
et al.
(
2024
).
Cerebellar output neurons impair non-motor behaviors by altering development of extracerebellar connectivity
.
bioRxiv
.
Leto
,
K.
,
Arancillo
,
M.
,
Becker
,
E. B. E.
,
Buffo
,
A.
,
Chiang
,
C.
,
Ding
,
B.
,
Dobyns
,
W. B.
,
Dusart
,
I.
,
Haldipur
,
P.
,
Hatten
,
M. E.
et al.
(
2016
).
Consensus paper: cerebellar development
.
Cerebellum
15
,
789
-
828
.
Lundell
,
M. J.
,
Chu-LaGraff
,
Q.
,
Doe
,
C. Q.
and
Hirsh
,
J.
(
1996
).
The engrailed and huckebein genes are essential for development of serotonin neurons in the Drosophila CNS
.
Mol. Cell. Neurosci.
7
,
46
-
61
.
Machold
,
R.
and
Fishell
,
G.
(
2005
).
Math1 is expressed in temporally discrete pools of cerebellar rhombic-lip neural progenitors
.
Neuron
48
,
17
-
24
.
Macosko
,
E. Z.
,
Basu
,
A.
,
Satija
,
R.
,
Nemesh
,
J.
,
Shekhar
,
K.
,
Goldman
,
M.
,
Tirosh
,
I.
,
Bialas
,
A. R.
,
Kamitaki
,
N.
,
Martersteck
,
E. M.
et al.
(
2015
).
Highly parallel genome-wide expression profiling of individual cells using nanoliter droplets
.
Cell
161
,
1202
-
1214
.
Matei
,
V.
,
Pauley
,
S.
,
Kaing
,
S.
,
Rowitch
,
D.
,
Beisel
,
K. W.
,
Morris
,
K.
,
Feng
,
F.
,
Jones
,
K.
,
Lee
,
J.
and
Fritzsch
,
B.
(
2005
).
Smaller inner ear sensory epithelia in Neurog1 null mice are related to earlier hair cell cycle exit
.
Dev. Dyn.
234
,
633
-
650
.
McDonough
,
A.
,
Elsen
,
G. E.
,
Daza
,
R. M.
,
Bachleda
,
A. R.
,
Pizzo
,
D.
,
DelleTorri
,
O. M.
and
Hevner
,
R. F.
(
2021
).
Unipolar (dendritic) brush cells are morphologically complex and require Tbr2 for differentiation and migration
.
Front. Neurosci.
14
,
598548
.
Mihalas
,
A. B.
,
Elsen
,
G. E.
,
Bedogni
,
F.
,
Daza
,
R. A. M.
,
Ramos-Laguna
,
K. A.
,
Arnold
,
S. J.
and
Hevner
,
R. F.
(
2016
).
Intermediate progenitor cohorts differentially generate cortical layers and require Tbr2 for timely acquisition of neuronal subtype identity
.
Cell Rep.
16
,
92
-
105
.
Millen
,
K. J.
,
Wurst
,
W.
,
Herrup
,
K.
and
Joyner
,
A. L.
(
1994
).
Abnormal embryonic cerebellar development and patterning of postnatal foliation in two mouse Engrailed-2 mutants
.
Development
120
,
695
-
706
.
Millen
,
K. J.
,
Hui
,
C.-C.
and
Joyner
,
A. L.
(
1995
).
A role for En-2 and other murine homologues of Drosophila segment polarity genes in regulating positional information in the developing cerebellum
.
Development
121
,
3935
-
3945
.
Morales
,
D.
and
Hatten
,
M. E.
(
2006
).
Molecular markers of neuronal progenitors in the embryonic cerebellar anlage
.
J. Neurosci.
26
,
12226
-
12236
.
Orvis
,
G. D.
,
Hartzell
,
A. L.
,
Smith
,
J. B.
,
Barraza
,
L. H.
,
Wilson
,
S. L.
,
Szulc
,
K. U.
,
Turnbull
,
D. H.
and
Joyner
,
A. L.
(
2012
).
The engrailed homeobox genes are required in multiple cell lineages to coordinate sequential formation of fissures and growth of the cerebellum
.
Dev. Biol.
367
,
25
-
39
.
Paxinos
,
G.
and
Watson
,
C.
(
2006
).
The Rat Brain in Stereotaxic Coordinates: Hard Cover Edition
.
Elsevier
.
Pézier
,
A.
,
Jezzini
,
S. H.
,
Marie
,
B.
and
Blagburn
,
J. M.
(
2014
).
Engrailed alters the specificity of synaptic connections of Drosophila auditory neurons with the giant fiber
.
J. Neurosci.
34
,
11691
-
11704
.
Quina
,
L. A.
,
Harris
,
J.
,
Zeng
,
H.
and
Turner
,
E. E.
(
2017
).
Specific connections of the interpeduncular subnuclei reveal distinct components of the habenulopeduncular pathway
.
J. Comp. Neurol.
525
,
2632
-
2656
.
Rose
,
M. F.
,
Ahmad
,
K. A.
,
Thaller
,
C.
and
Zoghbi
,
H. Y.
(
2009
).
Excitatory neurons of the proprioceptive, interoceptive, and arousal hindbrain networks share a developmental requirement for Math1
.
Proc. Natl Acad. Sci. USA
106
,
22462
-
22467
.
Sapir
,
T.
,
Geiman
,
E. J.
,
Wang
,
Z.
,
Velasquez
,
T.
,
Mitsui
,
S.
,
Yoshihara
,
Y.
,
Frank
,
E.
,
Alvarez
,
F. J.
and
Goulding
,
M.
(
2004
).
Pax6 and engrailed 1 regulate two distinct aspects of Renshaw cell development
.
J. Neurosci.
24
,
1255
-
1264
.
Serrano
,
N.
and
Maschat
,
F.
(
1998
).
Molecular mechanism of polyhomeotic activation by Engrailed
.
EMBO J.
17
,
3704
-
3713
.
Seto
,
Y.
,
Ishiwata
,
S.
and
Hoshino
,
M.
(
2014
).
Characterization of Olig2 expression during cerebellar development
.
Gene Expr Patterns
15
,
1
-
7
.
Sgadò
,
P.
,
Albéri
,
L.
,
Gherbassi
,
D.
,
Galasso
,
S. L.
,
Ramakers
,
G. M. J.
,
Alavian
,
K. N.
,
Smidt
,
M. P.
,
Dyck
,
R. H.
and
Simon
,
H. H.
(
2006
).
Slow progressive degeneration of nigral dopaminergic neurons in postnatal Engrailed mutant mice
.
Proc. Natl. Acad. Sci. USA
103
,
15242
-
15247
.
Sgaier
,
S. K.
,
Lao
,
Z.
,
Villanueva
,
M. P.
,
Berenshteyn
,
F.
,
Stephen
,
D.
,
Turnbull
,
R. K.
and
Joyner
,
A. L.
(
2007
).
Genetic subdivision of the tectum and cerebellum into functionally related regions based on differential sensitivity to engrailed proteins
.
Development
134
,
2325
-
2335
.
Sillitoe
,
R. V.
,
Stephen
,
D.
,
Lao
,
Z.
and
Joyner
,
A. L.
(
2008
).
Engrailed homeobox genes determine the organization of purkinje cell sagittal stripe gene expression in the adult cerebellum
.
J. Neurosci.
28
,
12150
-
12162
.
Sillitoe
,
R. V.
,
Vogel
,
M. W.
and
Joyner
,
A. L.
(
2010
).
Engrailed homeobox genes regulate establishment of the cerebellar afferent circuit map
.
J. Neurosci.
30
,
10015
-
10024
.
Simon
,
H. H.
,
Saueressig
,
H.
,
Wurst
,
W.
,
Goulding
,
M. D.
and
O'Leary
,
D. D.
(
2001
).
Fate of midbrain dopaminergic neurons controlled by the engrailed genes
.
J. Neurosci.
21
,
3126
-
3134
.
Smith
,
S. T.
and
Jaynes
,
J. B.
(
1996
).
A conserved region of engrailed, shared among all en-, gsc-, Nk1-, Nk2- and msh-class homeoproteins, mediates active transcriptional repression in vivo
.
Development
122
,
3141
-
3150
.
Squair
,
J. W.
,
Gautier
,
M.
,
Kathe
,
C.
,
Anderson
,
M. A.
,
James
,
N. D.
,
Hutson
,
T. H.
,
Hudelle
,
R.
,
Qaiser
,
T.
,
Matson
,
K. J. E.
,
Barraud
,
Q.
et al.
(
2021
).
Confronting false discoveries in single-cell differential expression
.
Nat. Commun.
12
,
5692
.
Swanson
,
D. J.
and
Goldowitz
,
D.
(
2011
).
Experimental Sey mouse chimeras reveal the developmental deficiencies of Pax6-null granule cells in the postnatal cerebellum
.
Dev. Biol.
351
,
1
-
12
.
Swanson
,
D. J.
,
Tong
,
Y.
and
Goldowitz
,
D.
(
2005
).
Disruption of cerebellar granule cell development in the Pax6 mutant, Sey mouse
.
Dev. Brain Res.
160
,
176
-
193
.
Tolkunova
,
E. N.
,
Fujioka
,
M.
,
Kobayashi
,
M.
,
Deka
,
D.
and
Jaynes
,
J. B.
(
1998
).
Two distinct types of repression domain in engrailed: one interacts with the groucho corepressor and is preferentially active on integrated target genes
.
Mol. Cell. Biol.
18
,
2804
-
2814
.
Voogd
,
J.
and
Glickstein
,
M.
(
1998
).
The anatomy of the cerebellum
.
Trends Cogn. Sci.
2
,
307
-
313
.
Wang
,
V. Y.
,
Rose
,
M. F.
and
Zoghbi
,
H. Y.
(
2005
).
Math1 expression redefines the rhombic lip derivatives and reveals novel lineages within the brainstem and cerebellum
.
Neuron
48
,
31
-
43
.
Watson
,
J. D.
,
Wheeler
,
S. R.
,
Stagg
,
S. B.
and
Crews
,
S. T.
(
2011
).
Drosophila hedgehog signaling and engrailed-runt mutual repression direct midline glia to alternative ensheathing and non-ensheathing fates
.
Development
138
,
1285
-
1295
.
Willett
,
R. T.
,
Bayin
,
N. S.
,
Lee
,
A. S.
,
Krishnamurthy
,
A.
,
Wojcinski
,
A.
,
Lao
,
Z.
,
Stephen
,
D.
,
Rosello-Diez
,
A.
,
Dauber-Decker
,
K. L.
,
Orvis
,
G. D.
et al.
(
2019
).
Cerebellar nuclei excitatory neurons regulate developmental scaling of presynaptic Purkinje cell number and organ growth
.
eLife
8
,
e50617
.
Wilson
,
S. L.
,
Kalinovsky
,
A.
,
Orvis
,
G. D.
and
Joyner
,
A. L.
(
2011
).
Spatially restricted and developmentally dynamic expression of engrailed genes in multiple cerebellar cell types
.
Cerebellum
10
,
356
-
372
.
Wizeman
,
J. W.
,
Guo
,
Q.
,
Wilion
,
E. M.
and
Li
,
J. Y.
(
2019
).
Specification of diverse cell types during early neurogenesis of the mouse cerebellum
.
eLife
8
,
e42388
.
Wu
,
T.
,
Hu
,
E.
,
Xu
,
S.
,
Chen
,
M.
,
Guo
,
P.
,
Dai
,
Z.
,
Feng
,
T.
,
Zhou
,
L.
,
Tang
,
W.
,
Zhan
,
L.
et al.
(
2021
).
clusterProfiler 4.0: a universal enrichment tool for interpreting omics data
.
Innovation
2
,
100141
.
Wurst
,
W.
,
Auerbach
,
A. B.
and
Joyner
,
A. L.
(
1994
).
Multiple developmental defects in Engrailed-1 mutant mice: an early mid-hindbrain deletion and patterning defects in forelimbs and sternum
.
Development
120
,
2065
-
2075
.
Yeung
,
J.
,
Ha
,
T. J.
,
Swanson
,
D. J.
and
Goldowitz
,
D.
(
2016
).
A novel and multivalent role of Pax6 in cerebellar development
.
J. Neurosci.
36
,
9057
-
9069
.
Yu
,
G.
,
Wang
,
L.-G.
,
Han
,
Y.
and
He
,
Q.-Y.
(
2012
).
clusterProfiler: an R package for comparing biological themes among gene clusters
.
OMICS
16
,
284
-
287
.

Competing interests

The authors declare no competing or financial interests.

Supplementary information