ACSA‐2 and GLAST classify subpopulations of multipotent and glial‐restricted cerebellar precursors

Abstract The formation of the cerebellum is highly coordinated to obtain its characteristic morphology and all cerebellar cell types. During mouse postnatal development, cerebellar progenitors with astroglial‐like characteristics generate mainly astrocytes and oligodendrocytes. However, a subset of astroglial‐like progenitors found in the prospective white matter (PWM) produces astroglia and interneurons. Characterizing these cerebellar astroglia‐like progenitors and distinguishing their developmental fates is still elusive. Here, we reveal that astrocyte cell surface antigen‐2 (ACSA‐2), lately identified as ATPase, Na+/K+ transporting, beta 2 polypeptide, is expressed by glial precursors throughout postnatal cerebellar development. In contrast to common astrocyte markers, ACSA‐2 appears on PWM cells but is absent on Bergmann glia (BG) precursors. In the adult cerebellum, ACSA‐2 is broadly expressed extending to velate astrocytes in the granular layer, white matter astrocytes, and to a lesser extent to BG. Cell transplantation and transcriptomic analysis revealed that marker staining discriminates two postnatal progenitor pools. One subset is defined by the co‐expression of ACSA‐2 and GLAST and the expression of markers typical of parenchymal astrocytes. These are PWM precursors that are exclusively gliogenic. They produce predominantly white matter and granular layer astrocytes. Another subset is constituted by GLAST positive/ACSA‐2 negative precursors that express neurogenic and BG‐like progenitor genes. This population displays multipotency and gives rise to interneurons besides all glial types, including BG. In conclusion, this work reports about ACSA‐2, a marker that in combination with GLAST enables for the discrimination and isolation of multipotent and glia‐committed progenitors, which generate different types of cerebellar astrocytes.

Here we describe the expression of ACSA-2 in defined astrogliallike progenitors throughout the development of the cerebellum.
Immunophenotyping, transcriptomic data, and transplantation assays demonstrate that ACSA-2 marks a subpopulation of GLAST + cells and thus discriminates two distinct populations of multipotent and glia-committed progenitors, which generate distinct types of cerebellar astrocytes.

Significance
The developing cerebellum comprises a variety of glial precursor populations. Distinguishing these populations will be beneficial to better understand the role of glial cells during cerebellar development and their impact on developmental diseases. By using astrocyte cell surface antigen-2 (ACSA-2) in combination with GLAST we developed a cell surface marker code that allows for the separation of multipotent versus parenchymal astrocytes-restricted precursors from the neonatal murine cerebellum. Consequently, this makes ACSA-2 a powerful tool to study differences in cerebellar glia not only in homeostasis but also during development and disease.
trypsin (Miltenyi Biotec) in combination with the gentleMACS™ Octo Dissociator with heaters. Dissociated tissue was passed through a 100 μm cell strainer (BD) and pelleted by 10 min centrifugation at 300 g at room temperature (RT). The cell pellet was resuspended in D-PBS buffer (Lonza) and processed immediately.

| Flow cytometry analysis
Murine single-cell suspensions, of either sex, were obtained from cerebellar tissue, incubated with FcR Blocking Reagent mouse (Miltenyi Biotec) to prevent unspecific antibody binding and then stained with fluorochrome-conjugated antibodies (compare also Table 1) Anti-S100beta Mouse Clone SH-B1 RRID:AB_477499 Anti-Ter119 MicroBeads Rat Ter-119

| Immunohistology
Adult and neonatal mice, of either sex, were anesthetized with keta-

| Imaging
Histological analyses of the grafts were performed using an E-800 Nikon microscope. Confocal images were taken on a Zeiss (LSM) or a Leica (TCS SP-5) confocal microscope. Images were processed using NIH ImageJ (RRID:SCR_003070) software.

| Analysis of sc/snRNA-seq published data sets
To better understand the nature of the A + /G + and A − /G + cells and identify candidate marker genes for these two populations, we analyzed two recently published data sets of sc/snRNA-seq in the embryonic/postnatal and adult mouse cerebellum (Kozareva et al., 2020;Vladoiu et al., 2019). The sequence data of these two data sets are available in the Gene Expression Omnibus (Boulay et al., 2017) repository under accession GSE11 8068 and at the Neuroscience Multi-omics (NeMO) Archive (https://singl ecell. broad insti tute.org/single_cell/study/ SCP79 5/a-trans cript omic-atlas -of-the-mouse -cereb ellum), respectively. For both data sets, the RNA-seq count matrices were downloaded and processed with the Seurat R package (v 3.6.2).
Specifically, for the first data set (Vladoiu et al., 2019), we focused on the data obtained at P0 and P7. scRNAseq data were subjected to quality control, filtering of low-quality cells, log-normalization, scaling, and linear dimensional reduction (PCA) with the top 20 principal components. Afterwards, a graph-based clustering approach was used to identify, for each developmental stage, the astrocyte population based on known marker genes (Slc1a3, Fabp7, Aldh1l1, and S100b). The count matrix of the astrocyte population was therefore extrapolated and further subjected to the same clustering approach to identify distinct subpopulations, as described in Figures S1-S3 (Cerrato et al., 2018;Farmer et al., 2016;Salvi et al., 2019).
Specifically, the enriched expression of Gdf10, Gria1, and Gria4 allowed identifying the BG/BG progenitor subpopulations, while other parenchymal astrocytes/astrocyte progenitors were classified as those cells showing higher expression of Aqp4. The expression of Ptf1a, Dcx, and Ascl1, typically high in neuronal progenitors (Gleeson et al., 1999;Grimaldi et al., 2009;Hoshino et al., 2005) allowed the classification of one of the clusters observed at P0 as a neurogenic progenitors cluster.
For the adult data set (Kozareva et al., 2020), the original classification of the distinct cerebellar cell types provided by the authors ( Figure 4) was used to identify the astrocyte population within the data set. The count matrix of this population was then extracted and the data obtained from the distinct biological replicates (n = 6) were integrated according to the SCTransform workflow in Seurat (Hafemeister & Satija, 2019), to correct for batch effects.
Subsequently, cells were subjected to clustering to identify the distinct astrocyte subpopulations ( Figures S1-S3).
The Seurat FindAllMarkers function was applied for both data sets to identify the differentially expressed genes of each cluster, selecting only significantly upregulated genes (p < 0.01) at least 1.4-fold overexpressed in more than 25% of the cells belonging to the subpopulation of interest (when compared to all other astroglial cells in the data set under analysis). For the different developmental stages, the lists of the top 100 differentially expressed genes of the clusters classified as BG/BG progenitors, parenchymal astrocytes/ astrocyte progenitors, or neurogenic progenitors were used.

| Statistics
Statistics were calculated with GraphPad Prism (GraphPad Software Inc., California USA) (RRID:SCR_002798). Error bars in the graphs represent average values ± standard deviation (SD).

| ACSA-2 is expressed by subpopulations of astroglial cells in the adult and developing cerebellum
As others and we demonstrated previously, the ACSA-2 reveals a broad expression in the murine adult cerebellum. Initial studies with the general astrocyte marker GLAST disclosed a non-overlapping pattern in distinct astroglial populations of the adult cerebellum (Kantzer et al., 2017). To address this in more detail, high-resolution analyses were performed and the dynamic and localization of ACSA-2 + astrocytes were investigated using brain lipid-binding protein (BLBP), besides GLAST; two markers that are known to be expressed by cerebellar astrocytes (Anthony et al., 2004;Cerrato et al., 2018). In line with former studies (Holmseth et al., 2012;Jungblut et al., 2012;Storck et al., 1992), GLAST was broadly expressed by BG, an astrocyte cell type closely associated with Purkinje neurons in the ML As cerebellar astrocytes are mainly generated postnatally (Cerrato et al., 2018), we addressed the ontogenesis of ACSA-2 + astrocytes early after birth using immunostainings ( Figure 2). To assess TA B L E 5 Test of independence for the cell populations revealed upon grafting

Main effect (p)
BG versus astrocytes: A + /G + (n = 7); A − /G + (n = 9) Chi-squared test 21.98 <0.0001 (****) Glia versus interneurons: A + /G + (n = 7); A − /G + (n = 9) Chi-squared test 19.08 <0.0001 (****) the specificity of ACSA-2 labeling to astroglial cells, we co-stained ACSA-2 with the neural/glial antigen (NG2), which is expressed by oligodendrocyte precursor cells (OPCs) (Polito & Reynolds, 2005) ( Figure S4a). No co-expression of the two markers was detected ( Figure S4a) and this result was further confirmed by cytofluorimetric assessment of positivity for ACSA-2 and the other OPC marker platelet-derived growth factor receptor alpha (PDGFRα) ( Figure S4b). Thus, we concluded that ACSA-2 only marks the astroglial lineage. Quantification of BLBP + astroglial progenitors at P3 showed that they are present throughout the cerebellar layers: they represent a small fraction of cells in the PWM and GL while they are the major population in the PCL/ML ( Figure S4c). Here, bona fide BG progenitors in the PCL and GL astrocytes, marked by BLBP and a very rare population of ACSA-2 + /GLAST − cells (not included in F I G U R E 1 ACSA-2 presents a unique expression pattern in the adult cerebellum. Coronal (a) and sagittal (b,c) adult mouse (younger than 2 months) cerebellar sections were used to investigate ACSA-2 expression. Confocal stacks identified ACSA-2, unlike the common astrocyte markers GLAST (a,c) and BLBP (b), to be distinct in the three cerebellar cortical layers (a,c). GLAST + BLBP + BG processes in the ML (a″,b″,c″; arrowheads in a″,b″) revealed weak ACSA-2 expression (a′-c′; arrowheads in a′,b′). High ACSA-2 expression was found on velate protoplasmic astrocytes in the GL (a′-c′). ACSA-2 is further expressed by a third type of cerebellar glia, the fibrous astrocytes in the WM (c). the quantifications, Figure 2d-d″″, empty arrows). These observations suggested that ACSA-2 and GLAST define two major astroglial cell subsets in the developing cerebellum: ACSA-2 + expressing cells that co-express GLAST + (A + /G + ) and cells that are devoid of ACSA-2 but express GLAST (A − /G + ). The occurrence of these subsets was later confirmed by flow cytometry (Figure 3).
Since the PWM hosts proliferating progenitors with astroglial traits as well as post-mitotic astrocytes, we inquired whether ACSA-2 + astroglial cells could represent early post-mitotic elements ( Figure   S5a-d: close up in b and orthogonal projection in c). Co-labeling with the proliferation marker Ki67 showed that at P3 8.4% ± 1.55% ( Figure   S5d) of the ACSA-2 + cells in the PWM proliferate ( Figure  In summary, ACSA-2 expression starts during the first postnatal days in astroglial-like cells in the PWM. It then gradually expands throughout the cerebellum to label astrocytes in the GL and WM, respectively, and to a lesser extent BG in the ML. F I G U R E 2 ACSA-2 defines a subpopulation of astrocyte progenitors in the neonatal cerebellum. The ontogeny of ACSA-2 + cells was investigated using confocal stacks of P3 (a-d), P7 (e) and P12 (f) sagittal sections. At P3, ACSA-2 was not detected in the nascent ML (a′,c′) and neither in the PCL (a′,c′). BG progenitors, which are marked by BLBP and GLAST (a″,c″,c″′), are not expressing ACSA-2. ACSA-2 is present in the PWM (a′,b′,c′,d″) but it is not as broadly expressed as the common markers GLAST and BLBP (a″,c″,c″′,d′,d″′). Within the GLAST + population (of the PWM) ACSA-2 positive (white arrowheads in d) as well as ACSA-2 negative cells (filled white arrows in d) and a very rare population of ACSA-2 + /GLAST − cells (empty arrows in d) were detected (d-d″). At P7, the labeling of ACSA-2 appeared in the PWM and the GL and overlapped with the reporter expression of the human glial fibrillary acidic protein (hGFAP) in the PWM and the GL (e). At P12, ACSA-2 expression has expanded to include BLBP + and GLAST + cells in both inner compartments: the GL and the WM (f). DAPI, nuclear stain. Scale bars: 100 μm ( paired t test; Table 3). We also tested the ganglioside marker A2B5 (Fredman et al., 1984;Lee et al., 2000). At P3, A2B5 was significantly higher on A + /G + compared to A − /G + (Figure 3c) (*p = 0.032; paired t test; n = 3; Table 4). In conclusion, these data show that markers of gliogenic progenitors (A2B5 and CD15) are particularly enriched in To perform comparative analyses on separated cell fractions we established magnetic isolation protocols (see Methods). Using these protocols, viable A + /G + cells (Figure 3e)
The top A − /G + 10 hits included, instead, three markers, namely myosin-binding protein C (Mybpc1), neuropeptide Y gene (Npy), and the glutamate ionotropic receptor AMPA type subunit 1 (Gria), previously described in BG or BG subsets (Kozareva et al., 2020;Reeber et al., 2018;Rodriques et al., 2019;Saab et al., 2012;Zeisel et al., 2018). Of note, also the transporter Slco4a1, actin beta-like protein (Actbl2), and clarin (Clrn1) have been reported in this astroglial type (Zeisel et al., 2018), while nephrin (Nphs1) was shown to be expressed by radial glia cells of the neonatal cerebellum (Putaala et al., 2001). Beta-1,3-galactosyltransferase 5 (B3galt5) is specifically expressed in certain lobules of the cerebellum (Rodriques et al., 2019), while no known association with astroglial cells or with a cerebellar origin was found for the growth/differentiation factor 2 (Gdf2)-also known as bone morphogenic protein 9 (BMP9)-and for the disintegrin and metalloproteinase with thrombospondin motifs 19 (Adamts19). Furthermore, within the top 50 differentially expressed genes in the A − /G + population we found Ptf1a, a bHLH F I G U R E 4 Transcriptional analysis identifies major differences between A − /G + and A + /G + precursors. (a) Volcano plot of microarraybased gene expression data. Probes with an at least fourfold median up-or downregulation, for both comparisons, and a Benjamini-Hochberg corrected p value (two-tailed Student's t test) of less than or equal to 0.05 are highlighted in red. The top 10 genes showing the highest upregulation in the A + /G + (left) or A − /G + (right) fraction are labeled. (b) Heatmap depicts the top 100 hits of the microarray probes found in the A + /G + fraction (light blue header bar) compared to the A − /G + fraction (magenta header bar). All probes show at least fourfold upregulation and present a Benjamini-Hochberg corrected p value (two-tailed Student's t test) of less than or equal to 0.05. Probes without gene assignments were excluded. Assignments of the microarray probes to the Uniprot keywords "Signal," "Developmental protein," "Cell adhesion," and "Differentiation" are indicated on the right hand side of the heatmap. (c) Heatmap depicts the top 93 median-centered log2 values of microarray probes with an at least fourfold median upregulation in the A − /G + fraction (magenta header bar) compared to the A + /G + fraction (light blue header bar) and with a Benjamini-Hochberg corrected p value (two-tailed Student's t test) of less than or equal to 0.05. Probes without gene assignments were excluded. Assignments of the microarray probes to the Uniprot keywords "Signal," "Developmental protein," "Cell adhesion," and "Differentiation" are indicated on the right hand side of the heatmap [Color figure can be viewed at wileyonlinelibrary.com] transcriptional gene that defines cerebellar GABAergic interneurons (Hoshino et al., 2005), known to derive from PWM astroglial-like progenitors (Parmigiani et al., 2015;Vladoiu et al., 2019).
To obtain further functional information from the cell transcriptomes we expanded the differentially expressed gene list to 100 genes for the A + /G + sample and to 93 for A − /G + cells (Figure 4b,c, see Methods) and used these lists as input files for the annotation enrichment tool DAVID (da Huang et al., 2009a(da Huang et al., , 2009b Overall, the transcriptomic analysis revealed unexpected differences in the gene expression profile of A − /G + progenitors, displaying markers of specific astroglia subtypes (BG and PWM precursors), and A + /G + progenitors that instead exhibit more generic astroglial traits.

| A − /G + and A + /G + populations comprise distinct cerebellar astroglial precursor subsets associated with different lineages
To gain a better understanding of the identity of A − /G + and A + /G + precursors in the context of cerebellar development and cell type differentiation, we compared their gene expression profiles to those of cerebellar astroglial cells at two different developmental stages (i.e., P0 ( Figure S1), and P7 ( Figure S2)). Data were extracted from a recently published single-cell (sc) RNA-seq data set of the developing mouse cerebellum (Vladoiu et al., 2019). Namely, at both stages, we first identified and extrapolated the cells with astroglial features based on the expression of genes for typical astrocyte markers such as Slc1a3, Fabp7, Aldh1l1, and S100b (see Methods), and then performed an unbiased cluster analysis. This allowed, at both P0 and P7, to distinguish two main cell subpopulations, classified as BG progenitors or parenchymal astrocyte progenitors, as determined by the expression of well-known markers (Figures S1 and S2, see Methods). Furthermore, a subpopulation of cells expressing typical markers of cerebellar neurogenic progenitors, such as Ptf1a, Dcx, and Ascl1 (Gleeson et al., 1999;Grimaldi et al., 2009;Hoshino et al., 2005), was identified at P0 ( Figure S1), but disappeared at P7, as expected ( Figure S2; Leto et al., 2016). Next, we compared the gene expression profiles of these three subpopulations with the gene lists of A − /G + and A + /G + (Figure 4b,c). At both stages, A − /G + cells shared several genes with BG/BG progenitors and none with parenchymal astrocytes and their precursors (Figure 5a,b). Among these genes, beside some already highlighted above, we found growth/differentiation factor 10 (Gdf10) which was implicated in BG development (Mecklenburg et al., 2014) and the WNT inhibitory factor 1 (Wif1), recently associated with a BG subset (Kozareva et al., 2020). By contrast, A + /G + cells reflected the opposite pattern (Figure 5a

| Grafting experiments uncover distinct differentiation potentials between A + /G + and A − / G + precursors
To directly assess the actual developmental potential of A + /G + and A − /G + cells, we performed grafting experiments. We isolated A + / G + and A − /G + cells from P1-P3 β-actin-GFP cerebella, a stage of intense genesis of interneurons and glia cells .
A + /G + or A − /G + cells were injected into the cerebellar vermis of F I G U R E 5 Comparisons with available sc/snRNA-seq cerebellar data sets clarifies the nature of A + /G + and A − /G + precursors. Heatmaps show the average gene expression for Atp1b2, Npy and genes found to be shared by A + /G + or A − /G + cells and distinct subpopulations of astrocytes/ astrocyte progenitors at P0 (a), P7 (b) or P60 (Adult, c) (values were obtained from the sc/snRNA-seq data sets). A − /G + cells shared genes with the BG subpopulations and with the P0 neurogenic subpopulation, while A + /G + cells showed genes in common with the sole parenchymal astrocyte subpopulation at all stages analyzed. P3 cerebellar sections were stained with candidate markers specific for A − /G + (d-g) and A + /G + (h,i) cells. GRIA1 (d), NPY (e), and MYBPC1 (f) were confirmed to be enriched in A − /G + BG progenitors in the PCL and in the PWM (arrow in g) compared to A + /G + cells (arrowheads in g), while AGT (h) and SLC6A11 (i) were specifically expressed in A + /G + progenitors in the PWM. The astroglial identity was validated using an intrinsic GFP expressed under the control of the hGFAP promoter (d-g,i) or by GFAP staining ( For the analyses, the cellular phenotype was defined according to co-expression of glial or interneuronal markers, as well as by cell morphology and layering. The different cell populations grafted showed remarkable differences in frequencies of astrocytes and BG (p < 0.0001; Pearson's Chi-squared test; Table 5) as well as significantly different amounts of glia and interneurons (p < 0.0001; Pearson's Chi-squared test; Table 5) (Figures 6a,b and 7a,b). In both conditions, the grafted cells differentiated into glial cells, including a fraction of oligodendrocytes, as defined by OLIG2 and MBP expression and morphology (A + /G + : 26.56% ± 11.1%; A − /G + : 13.75% ± 7.1%; Figures 6g and 7e-g) and a major proportion of GFAP + astrocytes (Figures 6e,f and 7c,d). Further analysis showed that astrocytes from A + /G + and A − /G + (tot astrocytes A + / G + : 71.61% ± 11.2% versus. A − /G + : 18.0% ± 5.4%; Figures 6b and 7b) were similarly distributed in the GL and the WM. However, and very interestingly, only astrocytes derived from A − /G + cells were found in the ML (Figure 6b) where they differentiated into astrocytes including BG (Figure 6a,b,e) which were not seen in the A + /G + transplants (A − /G + : 21.9% ± 4.8% vs. A + /G + 0.7% ± 0.3%) (Figure 7a,b). A further striking difference in the progenies of the grafted cells is the yield of interneurons-these were exclusively found in the A − /G + grafts (Figures 6a-d and 7a,b). The ML comprises two populations of PARVALBUMIN + (PV) interneurons: basket cells and stellate cells. Both cell types are marked by the expression of PV and are generated at the early postnatal period, precisely at the age we chose for the grafting (Leto et al., 2006).
Basket and stellate cells were found in the ML of the A − /G + grafting sides (arrowheads in Figure 6b, stellate cells: Figure 6c, basket cells: Figure 6d). Overall, the transplantation experiment revealed a clearly distinct differentiation potential of A − /G + and A + /G + progenitor populations: A − /G + cells differentiated into interneurons and all major cerebellar astrocyte phenotypes including BG while A + /G + were exclusively gliogenic and, among astrocytes, differentiated only into GL and WM parenchymal astrocytes.
To expand further on the distinct differentiation capacities of the precursor populations we assessed their potential in a nonneurogenic environment. After heterochronic grafting of neonatal A − /G + or A + /G + cells into P60 cerebella, we detected PV positive interneurons exclusively in the A − /G + grafts ( Figure S6a). By contrast, A + /G + grafted cells differentiated exclusively into glial cells ( Figure S6b). These results confirm the multipotential of A − /G + cells and the distinct developmental potential for A − /G + and A + / G + cells.
In summary, early (GLAST + ) cerebellar progenitors that do not express ACSA-2 retain a remarkably broad differentiation potential, as highlighted by the production of interneurons, and by the ability to differentiate into all major cerebellar astrocyte types including BG. However, a subpopulation of GLAST + precursors that co-expresses ACSA-2 defines a population of essentially glialrestricted progenitors that differentiate into defined astrocyte types.
Here, using surface marker profiling, gene expression analyses and cell , we present the subclassification of GLAST + cerebellar precursors. The dual presence of ACSA-2 and GLAST was associated with a non-neurogenic phenotype giving rise to parenchymal astrocytes and oligodendrocytes whereas the absence of ACSA-2 was coupled with a multipotent potential leading to BG, parenchymal astrocyte, oligodendrocytes, and interneuron cell fates.

| BG precursors do not express ACSA-2
In the developing cerebellum, GLAST is detectable on radial glia cells as early as E12 and shows a broad expression on all astrocytes of the neonatal and adult cerebellum (Mori et al., 2006). In contrast to GLAST and other pan-astrocytic markers, we found ACSA-2 to be restricted in the PWM and not expressed in BG precursors of the PCL in the early neonatal cerebellum (Figure 2).
In the adult cerebellum, ACSA-2 is co-expressed with GLAST on a proportion of WM astrocytes and velate astrocytes of the GL ( Figure 1). Since ACSA-2 showed areas of bright and low intensities local circuits might control the protein expression. Interestingly, BG, a third type of astrocytes, which differentiates from precursors of the PCL, showed low levels of ACSA-2 and no labeling of the typical lamellate processes. Moreover, BG are only generated upon transplantation of A − /G + precursors that share a gene expression profile previously described for BG progenitors (Kozareva et al., 2020;Vladoiu et al., 2019). For example, we identified and validated three of the top 10 differentially expressed genes: Mybpc1, Npy, and Gria1. As discussed in the literature, all three proteins are important for BG. We showed that these proteins are expressed on A − /G + BG precursors ( Figure 5): the first candidate, MYBPC1 was previously associated with muscle tissue and only recently connected with the BG lineage (Kozareva et al., 2020;Rodriques et al., 2019). MYBPC1 + cells were also detected in the PMW where it is enriched in A − /G + cells, supporting an association between BG and multipotent precursors. Neuropeptide Y (NPY) is another protein we validated on BG precursors. NPY was demonstrated to be present on BG before (Reeber et al., 2018).
It serves as a trophic factor and thereby potentially modulates adult neurogenesis (Decressac et al., 2009;Geloso et al., 2015;Lattanzi & Geloso, 2015). Finally, we validated the presence of the glutamate ionotropic receptor AMPA type subunit 1 (Gria1) on A − / G + cells. GRIA1, also known as GluA1, is one of the predominant AMPA receptors expressed by BG and was shown to be perquisite for proper motor coordination (Saab et al., 2012). Nephrin is also known as being expressed by BG precursors in the PCL (Putaala et al., 2001). In addition, GDF10, involved in BG development (Gupta et al., 2018;Mecklenburg et al., 2014) and GLI1, expressed by BG during development and adulthood suggesting activation of Shh signaling (Corrales et al., 2004;Fleming et al., 2013), were further upregulated in the A − /G + population. Interestingly, it has been shown that the specification of BG is driven by extrinsic signals that lead to intrinsic changes (De Luca et al., 2016;He et al., 2018;Leto et al., 2016). Our data therefore support this idea, suggesting that several genes specific for the BG lineage might be suppressed in A + /G + precursors, and further stresses the fact that BG precursors are unlikely to be A + /G + . However, a full demonstration would need further investigation.

| Cerebellar A − /G + precursors are multipotent while A + /G + precursors are glia restricted
The PWM of the developing cerebellum is composed of distinct progenitor subsets namely multipotent cells and fate-restricted precursors (Buffo & Rossi, 2013;Leto et al., 2010;Milosevic & Goldman, 2002). Several studies have used lineage tracing or marker gradients to address the heterogeneity of these precursor populations (Cerrato et al., 2018;Fleming et al., 2013;Parmigiani et al., 2015). Remarkable, not all GLAST + cells in the PWM are ACSA-2 + , thus arguing that ACSA-2 marks a subpopulation of GLAST + cells. Comparable to previous studies describing GLAST + precursors as bipotent and thus capable of developing glia cells and interneurons (Parmigiani et al., 2015), we show that A − /G + precursors differentiate into interneurons and all astrocyte types including BG. Accordingly, only A − /G + , but not A + /G + , precursors, express genes, such as Ptf1a and Kirrel2, found specifically in a cluster of neurogenic astroglial progenitors that is transcriptionally different from BG and parenchymal astrocytes (Kozareva et al., 2020;Vladoiu et al., 2019). By contrast, progenitors that co-express GLAST along with ACSA-2 (A + /G + ) showed a limited differentiation potential in vivo. With the notable exception of BG, A + /G + cells generated exclusively glia and did not give rise to interneurons. These findings highlight that A + /G + progenitors have a purely gliogenic character. Indeed, A + /G + progenitors display a gene signature of parenchymal astrocytes while they do not express any gene of neurogenic or BG progenitor clusters. We identified Slc6a11 and Agt as representative genes and disclosed their protein expression pattern. Interestingly, SLC6A11 recapitulated the expression pattern of ACSA-2 to a very high degree being expressed by A + /G + PWM precursors at P3. The expression of the Agt gene has been previously described to discriminate astrocytes in non-telencephalic and caudal regions (Zeisel et al., 2018). We could confirm and further expand this finding by showing that the AGT protein is enriched in A + /G + precursors in the PWM. Thus, our data reveal that ACSA-2 expression defines two transcriptionally and functionally distinct progenitor populations in the developing cerebellum.

| The relationship between A + /G + and A − / G + precursors
The A − /G + population might serve as a progenitor pool giving rise to a common precursor lineage, which then develops interneurons and all astroglial subtypes. However, there is currently no clonal study that proves the existence of such lineage in the postnatal cerebellum. As a summary of this and previous studies (Cerrato et al., 2018;Parmigiani et al., 2015) we propose that GLAST + precursors constitute a heterogeneous precursor pool.
The first progenitor population (A − /G + ) is multipotent and less committed. This population resides in the PWM and includes progenitors generating both interneurons, BG and WM astrocytes (Parmigiani et al., 2015). A + /G + precursors instead contain astrocyte progenitors already committed to becoming astrocytes that generate WM/GL cells. Interestingly, however, both A + /G + and A − /G + cells give rise to oligodendrocytes.
Earlier fate-mapping studies showed that oligodendrocytes are generated from GLAST + progenitors preferentially in the embryo (Parmigiani et al., 2015). However, we obtained oligodendrocytes in our grafts upon the transplantation of early postnatal GLAST + cells.
Apart from the frequencies of oligodendrocytes, the proportional contribution of BG, interneurons, and astrocytes in the A − / G + grafts is comparable with the frequencies seen in Parmigiani et al. (2015). Since ACSA-2 + and GLAST + cells did not co-express PDGFRα or NG2, we consider the possibility of contaminating oligodendrocyte precursors within the grafts as rather unlikely. In this study we used an antibody affinity-based approach, whereas the former study used an inducible GLAST::Cre ERT line and lentiviral constructs to label GLAST + precursors. Thus, the variances might be explainable by the different methods used, as the promoter activity might not reflect the intrinsic level of the protein.
Despite their dissimilarity we showed in this study that both populations-A + /G + and A − /G + -express GLAST, thereby suggesting a lineage relationship. It might be that astroglial-like A − /G + cells of the PWM give rise to astroglial-committed A + /G + progenitors. Future experiments will disclose how these subsets are interrelated and whether A + /G + cells are ontogenically related to A − /G + cells or form a separate class.

| Outlook
Na + /K + ATPases maintain sodium and potassium concentrations, balance osmosis, and preserve the electrochemical gradient (Lecuona et al., 1996). The target of the anti-ACSA-2 antibody, the beta-2 subunit of the sodium/potassium ATPase (ATP1B2), is highly enriched in the murine cerebellum and is stably expressed during stab wound injury (Batiuk et al., 2017). This makes ACSA-2 an ideal candidate to not only study its function during cerebellar development but also the biological role upon neuro-inflammatory diseases and degenerative disorders.
Recently, a structural variation in exon 2 of the ATP1B2 gene has been linked to cerebellar ataxia in Belgian Shepherds (Mauri et al., 2017). Their publication described the same defects in motor coordination as seen in ATP1B2 (0/0) mice accompanied with death in the second postnatal week (Antonicek & Schachner, 1988;Magyar et al., 1994). Furthermore, aberrant expression of ATP1B2 has been linked to glioblastoma multiforme (GBM). It was shown that patients with GBM have changes in the isoforms of ATP1B2 (Rotoli et al., 2017) and that targeting ATP1B2 induced glioblastoma cells apoptosis (Li et al., 2019). These examples emphasize the importance to further investigate the function of ACSA-2/ATP1B2 expressing astrocytes that was beyond the scope of this study.

D ECL A R ATI O N O F TR A N S PA R EN C Y
The authors, reviewers and editors affirm that in accordance to the policies set by the Journal of Neuroscience Research, this manuscript presents an accurate and transparent account of the study being reported and that all critical details describing the methods and results are present.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1002/jnr.24842.

DATA AVA I L A B I L I T Y S TAT E M E N T
The gene expression profiling data that support the findings of this study are available in NCBI's Gene Expression Omnibus   Figures S2 and S3) confirmed the classification as a good fit. At P0, cluster n.1 expressed genes typically associated with a neuronal progenitor fate, such as Ptf1a, Dcx, and Ascl1 (Hoshino et al., 2005;Gleeson et al., 1999, Grimaldi et al., 2009, and was therefore classified as a cluster of neurogenic progenitors.
Further clusters did not present a clear identity-based on known marker genes-and were therefore not further considered for the