Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract We have identified active enhancers in the mouse cerebellum at embryonic and postnatal stages which provides a view of novel enhancers active during cerebellar development. The majority of cerebellar enhancers have dynamic activity between embryonic and postnatal development. Cerebellar enhancers were enriched for neural transcription factor binding sites with temporally specific expression. Putative gene targets displayed spatially restricted expression patterns, indicating cell-type specific expression regulation. Functional analysis of target genes indicated that enhancers regulate processes spanning several developmental epochs such as specification, differentiation and maturation. We use these analyses to discover one novel regulator and one novel marker of cerebellar development: Bhlhe22 and Pax3, respectively. We identified an enrichment of de novo mutations and variants associated with autism spectrum disorder in cerebellar enhancers. Furthermore, by comparing our data with relevant brain development ENCODE histone profiles and cerebellar single-cell datasets we have been able to generalize and expand on the presented analyses, respectively. We have made the results of our analyses available online in the Developing Mouse Cerebellum Enhancer Atlas, where our dataset can be efficiently queried, curated and exported by the scientific community to facilitate future research efforts. Our study provides a valuable resource for studying the dynamics of gene expression regulation by enhancers in the developing cerebellum and delivers a rich dataset of novel gene-enhancer associations providing a basis for future in-depth studies in the cerebellum. Editor's evaluation This manuscript is a valuable study to understand regulatory elements in the early developing mouse cerebellum. The authors have generated important ChIP-seq datasets from embryonic and early postnatal mouse cerebellum. The authors use these data to convincingly identify enhancers in the developing cerebellum. The authors have also made their data and analyses available online in the "Developing Mouse Cerebellum Enhancer Atlas." https://doi.org/10.7554/eLife.74207.sa0 Decision letter eLife's review process Introduction Neuronal development is a complex and dynamic process that involves the coordinated generation and maturation of countless cell types. For cerebellar granule cells, the most numerous neurons in the brain, neuronal differentiation consists of several steps beginning with the commitment of neural stem cells to become specified neural precursors, followed by multiple migratory stages to reach and mature at its final destination (Consalez et al., 2020). Underpinning these events is the expression of gene regulatory networks that drive dynamic molecular processes required for proper brain formation (Ziats et al., 2015). However, the transcriptional mechanisms that precisely regulate these gene expression programs have not been fully described. Gene expression is typically activated when transcription factors (TFs) bind to non-coding regulatory elements and recruit the necessary components to begin transcription. Among the several classes of non-coding sequences that regulate gene expression, enhancers are the most common, with thousands predicted to coordinate transcriptional regulation during development (Heinz et al., 2015). Enhancers are regulatory sequences that serve as binding sites for TFs and activate distal target gene expression. In the brain, enhancers help to ensure that gene expression is spatially- and temporally-specific, defining what genes will be active during distinct stages of development (Nord and West, 2020). Transcriptional regulation by enhancers has been shown to be critical for cellular identity, maturation during central nervous system (CNS) development, and activity-dependent responses in mature neurons (Frank et al., 2015; Pattabiraman et al., 2014). A detailed understanding of the enhancers that govern changes in gene expression during embryonic and early postnatal cerebellum development remains limited. Profiling genome-wide enhancer activity at different time points and identifying their gene regulatory targets can provide insight into developmental processes regulated by enhancer elements. Several molecular properties have been associated with enhancer activity, and the advancement of sequencing technology has facilitated their identification genome-wide in several developing brain structures (Carullo and Day, 2019). Enhancers are marked with histone post-translational modifications H3K4me1 and H3K27ac, both of which contribute to opening chromatin for TF binding (Calo and Wysocka, 2013). H3K27ac delineates active from poised elements and has been a reliable marker for enhancer activity genome-wide (Creyghton et al., 2010). Analysis of these marks, in conjunction with transcriptomic and epigenomic datasets, has revealed that the vast majority of non-coding variants associated with neurological and psychiatric disorders are found within these regulatory elements, highlighting their importance in functional readout in the brain (Barešić et al., 2020). Thus, profiling enhancer-associated histone modifications in the brain across time provides a comprehensive understanding of gene-regulatory principles, disease-associated variants, and the genetics of brain development (Nott et al., 2019). The cerebellum has been a long-standing model to study the developmental genetics of the brain. This is, in part, due to the limited number of cell types, well-defined epochs of development for these cell types and a simple trilaminar structure in which these cells are organized, making for an enhanced resolution of events in time and space (Wang and Zoghbi, 2001). Our lab has previously generated transcriptomic time course datasets spanning embryonic and postnatal development using microarrays and Cap Analysis of Gene Expression followed by sequencing (CAGE-seq; Forrest et al., 2014; Ha et al., 2019), which led to the discovery of novel TFs critical for proper development. More recently, the developing cerebellum has served as an ideal setting for pioneering mouse and human single-cell RNA-seq time courses (Aldinger et al., 2021; Carter et al., 2018; Peng et al., 2019; Wizeman et al., 2019). These studies have provided an unbiased classification of cerebellar populations using their associated gene expression profiles and insight into the cellular origins of distinct postmitotic cell subtypes from neural progenitors. When considering the non-coding regulatory elements, such as enhancers, that fine tune the spatial and temporal expression of these genes, there are only a few studies that have profiled these sequences genome-wide and specific to the developing cerebellum. Enhancer-associated histone marks (H3K27ac and H3K4me1) have been identified in postnatal and adult mouse cerebella (Frank et al., 2015). During embryonic development, these marks have been profiled in the mouse hindbrain, but not specifically in the cerebellum (Gorkin et al., 2020). Recently, the open chromatin landscape has been examined during embryonic and postnatal cerebellum development using single-nuclear ATAC-seq (snATAC-seq) (Sarropoulos et al., 2021). While this analysis has provided a comprehensive atlas of predicted regulatory sequences at cell-type resolution, additional signals associated with enhancer activity, such as H3K4me1 and H3K27ac, have yet to be quantified during embryonic and early postnatal development. Profiling these chromatin marks can provide further evidence of regulatory activity and is an important step in establishing a catalog of active enhancers involved in cerebellar development. Identifying enhancers and predicting their target genes would also provide a valuable resource for the research community and would facilitate the discovery of novel genetic drivers of the precisely-timed and cell-specific molecular events in the developing cerebellum. We utilize chromatin immunoprecipitation followed by sequencing (ChIP-seq) of enhancer associated histone marks H3K4me1 and H3K27ac at 3 stages of embryonic and early postnatal cerebellar development. We identify temporally specific enhancers using a differential peak analysis comparing postnatal and embryonic timepoints. TF motif enrichment and prediction of gene targets led to the elucidation of molecular processes regulated by enhancers during these stages. As examples of the use of this data for discovery, we identify two novel regulators of cerebellar development, Pax3 and Bhlhe22: a novel marker of GABAergic progenitors and a regulator of postnatal granule cell migration, respectively. Finally, we also demonstrate how this data may be used to explore the role of enhancers in neurodevelopmental disabilities by identifying an enrichment of autism spectrum disorder (ASD) associated SNPs and de novo variants found in ASD-affected individuals in cerebellar enhancers. We have made the results of our analyses available online in the Developing Mouse Cerebellum Enhancer Atlas which can be easily queried, curated and exported by the research community. This provides a rich resource that can be utilized to discover novel genetic regulators of cerebellar development and contribute to understanding the impact of genetic variation associated with neurodevelopmental disorders. Results Enhancer identification during cerebellar development To identify enhancers active during embryonic and postnatal cerebellar development, we generated genome-wide H3K27ac and H3K4me1 ChIP-seq profiles from mouse cerebella dissected at embryonic day 12 (E12), postnatal day 0 (P0) and postnatal day 9 (P9) (Figure 1A). These stages represent a window of time that spans important neurodevelopmental events such as cell specification, emergence from the cell cycle, differentiation, migration, and maturation for all of the major neuronal types of the cerebellum including the cerebellar granule cells, Purkinje cells, cerebellar interneurons, and cerebellar nuclear neurons (Goldowitz and Hamre, 1998; Hatten and Heintz, 1995; Wang and Zoghbi, 2001). H3K27ac and H3K4me1 signals were reproducible between biological replicates as exemplified in a region on chromosome 14 (Figure 1B). There was a high correlation between replicates for both marks at each age (Figure 1—figure supplement 1A). Therefore, we had confidence in using our H3K27ac and H3K4me1 data in downstream analyses. Robust cerebellar enhancers were identified by the presence of overlapping peaks between the two enhancer-associated histone marks at each age. This highlighted a total of 9,622 peaks; 5,859, 474, and 3,289 peaks that were in both the H3K27ac and H3K4me1 datasets at E12, P0, and P9, respectively (Figure 1C). Duplicate peaks between ages were removed, producing a list of 7024 active cerebellar enhancers derived from overlapping H3K27ac and H3K4me1 signals (Supplementary file 1). Figure 1 with 1 supplement see all Download asset Open asset Enhancer identification during cerebellar development. (A) An overview of the stages of cerebellar development profiled in this study. The datasets collected at these ages and the downstream analyses are shown in the flow chart. Labels: NE: Neuroepithelium, RL: Rhombic lip, EGL: External granular layer, PL: Purkinje layer, IGL: Inner granular layer, CN: Cerebellar nuclei (B) A region of the mouse genome chr14:122,715,876–122,899,964 (mm9) in the Integrative Genomics Viewer (IGV) showing H3K27ac and H3K4me1 profiles across biological replicates of E12, P0, P9 cerebella. Active cerebellar enhancers are highlighted (gray box). (C) Venn diagrams displaying overlap between H3K27ac and H3K4me1 peaks at E12, P0 and P9. (D–E) An example of a cerebellar enhancer identified from the E12 cerebellum. Shown is normalized H3K27ac and H3K4me1 signal at the enhancer (gray box), as well as (E) normalized CAGE-seq expression of the nearest gene, Ascl1, across developmental time, at E12, P0, P9. TPM, Transcripts Per Million. (F–G) An example of a cerebellar enhancer identified from the P9 cerebellum. Shown is normalized H3K27ac and H3K4me1 signal at the enhancer (gray box), as well as (G) normalized (TPM) CAGE-seq expression of the nearest gene, Pax6, across developmental time, at E12, P0, P9. The relationship between enhancer activity and genes relevant to cerebellar development is shown in genomic regions flanking Ascl1 and Pax6, two genes critical to cerebellar development (Kim et al., 2008; Yeung et al., 2016). We identified an enhancer active at E12 located in close proximity to Ascl1 (Figure 1D). A decrease in the H3K27ac ChIP-seq signal at this enhancer corresponded to a decrease in Ascl1 gene expression (Figure 1E). We identified two active enhancers at P9 located near Pax6 (Figure 1F). H3K27ac ChIP-seq signal also showed a pattern of activity similar to Pax6 expression, increasing from embryonic to postnatal ages (Figure 1G). These results provide validation for the enhancers identified in our dataset relative to genes critical to cerebellar development. We compared our list of robust cerebellar enhancers to four previously published cerebellar enhancer datasets. First, P7 H3K27ac ChIP-seq and DNase-seq profiles previously generated by Frank et al., 2015 were overlapped with robust cerebellar enhancers. Greater than 90% of our reported robust cerebellar enhancers are replicated by H3K27ac and DNAse-seq peaks from this study (Figure 1—figure supplement 1B-C). Second, we compared the number of robust cerebellar enhancers and our histone profiles identified at P9 to H3K27ac and H3K4me1 ChIP-seq datasets from the adult cerebellum (Gorkin et al., 2020). We found that 73% (7037/9545) of H3K27ac peaks, 59% (20674/35010) H3K4me1peaks and 60% (1974/3289) of robust cerebellar enhancers overlapped with peaks identified in the adult cerebellum (Figure 1—figure supplement 1D). This indicates that a subset of the enhancers identified at P9 are active specifically during postnatal development when compared to adult stages. Third, enhancers retrieved from the enhancer database EnhancerAtlas 2.0, reporting enhancer activity in the mouse cerebellum at P0-P14 (Gao and Qian, 2020), were compared to robust cerebellar enhancers. We found that 73%, and 80% of our enhancers overlapped with the postnatal cerebellum enhancer dataset at P0, and P9, respectively (Figure 1—figure supplement 1E). Fourth, mouse enhancers that had experimentally validated hindbrain activity at E11.5 from the VISTA Enhancer Browser (Visel et al., 2007) were compared to the cerebellar enhancers reported here. We found that 56% of VISTA enhancers overlap with our cerebellar enhancer sequences at E12 (Figure 1—figure supplement 1F). Additionally, we found that a higher proportion of H3K27ac peaks overlapped with VISTA enhancers compared to open chromatin regions identified in the developing cerebellum (Sarropoulos et al., 2021) and the developing hindbrain (Gorkin et al., 2020; Figure 1—figure supplement 1G). This indicates that our identified H3K27ac peaks may be more predictive of active enhancers than previously generated chromatin accessibility datasets. Taken together, these confirmative findings indicate our approach was effective in capturing active cerebellar enhancers. Enhancer dynamics during cerebellar development The dynamics of enhancer activity over cerebellar development were examined through a differential peak analysis of H3K27ac signal. The majority, 89% (6238/7023), of cerebellar enhancers had significant differences in peak signal (adjusted P-value ≤0.05) throughout cerebellar development (Figure 2A, Supplementary file 2). At P9, 1273 cerebellar enhancers were significantly active compared to either P0, E12 or both. At E12, 4432 active enhancers were differentially active compared to either P9, P0 or both. At P0, in contrast, only a small number of enhancers with differential signal was identified (403 and 154 showed significant changes when compared to E12 and P9, respectively). However, none of these P0 cerebellar enhancers were differentially active when compared to both E12 and P9, indicating that enhancer activity did not spike at birth. Quality control metrics evaluating sensitivity of the H3K27ac and H3K4me1 ChIPs (Supplementary file 3), such as the fraction of reads in peaks (FRiP) and relative strand correlation (RSC), all met ENCODE standards (Landt et al., 2012). The P0 samples had slightly lower scores, on average, in both metrics than the E12 and P9 samples, suggesting a minor decrease in sensitivity compared to the other stages. We acknowledge that the difference in sensitivity may have masked P0-specific enhancer activity in our analysis and that future experiments may reveal temporally-specific enhancers at this stage. Taken together, this analysis highlights two temporally specific windows of enhancer activity at Early (embryonic) and Late (postnatal) stages (Figure 2B). Figure 2 with 4 supplements see all Download asset Open asset Enhancer activity is dynamic throughout cerebellar development. (A) Volcano plots showing robust cerebellar enhancers with differential H3K27ac peak signal for three comparisons: E12 vs P9, E12 vs P0, and P0 vs P9. Differential signal strength was identified for 4433 and 4355 robust cerebellar enhancers when comparing E12 to P9 and to P0, respectively. At P9, 1275 and 403 robust cerebellar enhancers had differential signal when compared to E12 and P0, respectively. Enhancers with significant differential activity are colored at a cutoff of an adjusted P-value <0.05. Displayed on the y-axis is the negative log10 adjusted p-value and on the x-axis is the difference in ChIP-seq signal between to the ages for a given peak. (B) Diagram displays how Early and Late active cerebellar enhancers are classified based on differential peak analysis results. (C) Boxplot shows mean ChIP-seq signal (y-axis) for all Early (upper) and Late (lower) active enhancers. Error bars represent the standard error of the mean. (D) Mean profile and heatmaps of H3K27ac signal at the midpoint of our predicted cerebellar enhancers (rows ±3 kb) in Early and Late groups at E12, P0, and P9. Distinct patterns of enhancer activity were observed for temporally classified enhancers. For Early active enhancers, there was a loss of mean H3K27ac signal over time, with a steep decline after E12 (Figure 2C). Late active enhancers exhibited a gain in activity over time, with mean H3K27ac signal increasing steadily through development. These patterns are seen when looking at the changes in signal flanking the summits of our peaks across time (Figure 2D). These results indicate that the majority of cerebellar enhancers are dynamic throughout time and exhibit temporally specific activity. A subset of cerebellar enhancers is active in other developing regions of the brain To examine cerebellar enhancer activity in the context of other developing brain regions, we overlapped our histone profiles and robust cerebellar enhancer coordinates with H3K4me1 and H3K27ac profiles previously generated from either the developing mouse hindbrain, midbrain or forebrain at E12 and P0 (Gorkin et al., 2020). At E12, we found that the majority of cerebellar H3K27ac and H3K4me1 peaks overlapped with profiles generated in either the hindbrain, midbrain, or forebrain (Figure 2—figure supplement 1A). Similar results were observed at P0, with the majority of cerebellar H3K27ac peaks identified in other brain regions; however, only a modest overlap was found for H3K4me1 profiles at this stage (Figure 2—figure supplement 1B). We found that most robust cerebellar enhancers overlapped with H3K27ac peaks detected in the hindbrain (89.1%), midbrain (78.6%), or forebrain (78.6%; Figure 2—figure supplement 1C). When looking at multiple brain regions, 69.8% of robust cerebellar enhancers were active in all three developing brain regions (Figure 2—figure supplement 2). Interestingly, the majority of the enhancers active in these brain regions (73.9%) were classified as Early active enhancers in the cerebellum which display temporally specific activity to embryonic development. Importantly, we identified 467 (6.7%) robust cerebellar enhancers specifically active in the cerebellum and not the other brain regions, indicating our analysis has uncovered novel enhancer sequences potentially critical for cerebellar formation (Figure 2—figure supplement 2). Over half of the cerebellar specific enhancers (55.03%) were classified as Late active enhancers with activity peaking during postnatal development, while only 15.92% of these enhancers were Early active enhancers and the remaining 29.05% of enhancers were not differentially active (labeled as ‘Neither’). We then examined whether Early and Late and enhancers displayed similar patterns of activity in other brain regions from E12 to P0. Early enhancers displayed a decrease in average activity over time in the hindbrain, midbrain and forebrain (Figure 2—figure supplement 3). Late enhancers showed an increase in hindbrain samples, but minimal changes in the developing midbrain and forebrain (Figure 2—figure supplement 3). These collective results indicate that Early enhancers may also be active in other developing regions of the brain while Late active enhancers may be more likely to be active specifically in the cerebellum. Comparison of robust cerebellar enhancers with chromatin accessibility identified in single cells To gain a more granular view of the spatial activity of robust cerebellar enhancers, we overlapped their coordinates with cis-regulatory elements (CREs) active in the developing cerebellum previously identified by single-nuclei ATAC-seq (snATAC-seq; Sarropoulos et al., 2021). We found that 6342/7024 (90.1%) robust cerebellar enhancers overlapped with CREs with open chromatin conformation. Cell-types were then assigned to robust cerebellar enhancers based on the activity of an overlapping CRE. CREs were previously aggregated into 26 clusters based on activity using an iterative clustering procedure (Sarropoulos et al., 2021). Using these clusters, we were able to assign a predicted cell type to the 6342 robust cerebellar enhancers that overlapped with CREs (Figure 2—figure supplement 4A). The predicted cell types with the largest proportion of robust cerebellar enhancers were granule cells (19.32%), progenitor cells (18.35%), Purkinje cells (11.48%) and multiple early-born neuron types (9.69%; Figure 2—figure supplement 4A). When splitting cerebellar enhancers into Early and Late active groups, we found that the majority of Early enhancers were predicted to be active in progenitor cells, and multiple early born neuron types such as Purkinje cell precursors while Late enhancers were predicted to be active in developing granule cells, which is the dominant population of cells produced during postnatal stages, and other late-born cell types (Figure 2—figure supplement 4B). We then examined the average open chromatin signal at our established Early and Late robust cerebellar enhancers in progenitor cells and the predominant neuron types: granule cells, Purkinje cells, and interneurons. Early enhancers peaked in accessibility at embryonic stages and steadily declined throughout developmental time for all cell types (Figure 2—figure supplement 4C). Late enhancers gradually increased in accessibility throughout time peaking at postnatal stages for progenitor cells, granule cells and interneurons. In Purkinje cells, Late active enhancers showed a minimal change in accessibility throughout time (Figure 2—figure supplement 4C). These findings indicate that Early and Late enhancers display their respective activity patterns in the context of chromatin accessibility in individual cell-types. Cerebellar enhancers are enriched for neural transcription factor binding sites in an age-dependent manner We then sought to identify transcription factors whose activity is dictated by the availability of robust cerebellar enhancers, as many neural lineage-defining factors drive cell commitment in the developing brain through enhancer binding (Elsen et al., 2018; Lindtner et al., 2019). We used HOMER to search for enriched motifs (adjusted p-value <1E-11) in Late and Early active cerebellar enhancers; which were then matched to known transcription factor motifs in the JASPAR database (Heinz et al., 2010). This analysis revealed a distinct set of significantly enriched motifs for Early and Late enhancers (Figure 3A). The TFs with the best matching DNA binding motifs belonged to protein families with known regulatory roles in cerebellar development, serving as validation for our analysis (Figure 3A). Since TF motifs are similar between protein family members, it is possible that Early and Late cerebellar enhancers are bound by TFs in the same protein family as the predicted best match indicated in Supplementary file 4. TFs enriched in the Early active enhancer group show a decrease in expression over time while TFs enriched in the Late active enhancer group show an increase in expression over time. This correspondence between enriched TF expression and enhancer activity provides validation for our findings and indicates the timing of enhancer activity may be dictated by the expression and binding of these enriched TFs. Figure 3 with 2 supplements see all Download asset Open asset Neural transcription factors with known and novel function in the developing cerebellum are enriched in dynamic cerebellar enhancers. (A) Dot plot displaying significantly enriched (adjusted p-value <1E-11) motifs and the predicted matching transcription factor (TF). Displayed are the results for Early (top) and Late (bottom) active enhancers. TFs with an unknown functional role in cerebellar development are indicated with a red arrow. Size of the dots indicate the negative log10 adjusted p-value for a given motif and the color scale displays the z-score normalized expression throughout the cerebellar developmental time course. (B) Top: Immunofluorescent staining of Pax3 in the mouse cerebellum at E12, E15, P0, and P3. Bottom: Pax3 and Ptf1a immunofluorescent co-staining of the E12 mouse cerebellum. Immunofluorescent co-staining of Pax3 and Pax2 in the mouse cerebellum at E15 and P0. Labels: CP: Cerebellar parenchyma, EGL: External granular layer, NGL: Nascent granular layer, RL: Rhombic lip, VZ: Ventricular zone, Scalebars = 100 µm. The top three enriched TF motifs for Early active enhancers were Ascl1, Meis2, and Atoh1 (Figure 3A). These TFs have established roles in cerebellar development, acting as markers of GABAergic or glutamatergic cell types and regulators of differentiation (Ben-Arie et al., 1997; Kim et al., 2008; Wizeman et al., 2019). Importantly, many of the motifs enriched in the Early group matched with TFs which have received little to no attention in the cerebellum, including Sox4, Lhx2, Rfx4, Pou3f1, and Pax3 (Figure 3A). These TFs have been previously associated, however, with the development of other brain areas (Frantz et al., 1994; Porter et al., 1997; Su et al., 2016; Zhang et al., 2006). In contrast, the TFs matching the motifs enriched in the Late active enhancers have a previously identified role in cerebellar development; but not necessarily in the same cellular processes (Figure 3A). For example, the top 3 enriched motifs matched with Neurod1, Nfia/b/x, and NF1, which have all been associated with granule cell differentiation (Miyata et al., 1999; Sanchez-Ortiz et al., 2014; Wang et al., 2007). However, two other TFs with enriched binding sites, Pax6 and Smad4 have been found to be critical for granule cell precursor proliferation, a process preceding differentiation (Swanson and Goldowitz, 2011). These results suggest a dynamic role for the majority of our Early and Late active enhancers, driven by TFs involved in distinct stages of neuron development. Early active enhancers are enriched for Pax3 binding sites, a novel marker for GABAergic cells The TF motif enrichment analysis of Early enhancers led to the discovery of several novel TFs in the context of embryonic cerebellar development; potentially involved in seminal aspects of development such as cellular specification or commitment. As a case study, we focused on Pax3, as other members of the Pax protein family have been shown to play key roles in the developing cerebellum (Leto et al., 2009; Urbánek et al., 1997; Yeung et al., 2016). Immunofluorescent staining was conducted to profile Pax3 expression in the embryonic and early postnatal mouse cerebellum. We observed robust expression in the ventricular zone (VZ); a neural progenitor region for GABAergic cells in the cerebellum (Leto et al., 2006; Figure 3B). A molecular marker of the VZ is Ptf1a, a GABAergic lineage-defining molecule in the cerebellum (Hoshino et al., 2005), and we examined if there was co-expression between Pax3 and Ptf1a. Colocalization between 50% of Pax3 positive cells in the VZ and Ptf1a (Figure 3—figure supplement 1A), (Hoshino et al., 2005) confirmed Pax3 expression within GABAergic neural progenitors. At E15, Pax3 + cells are seen in the region just dorsal to the VZ, which consist of post-proliferative cells such as Purkinje cells and interneurons (Hoshino et al., 2005; Leto et al., 2006; Figure 4B). We examined Pax3 co-labeling with markers for these cell types using Foxp2 (marker of Purkinje cells) and Pax2 (a general marker of GABAergic interneurons; Fujita et al., 2008; Maricich and Herrup, 1999). While colocalization between 91% of Pax3 + ce
