scRNA-seq of NEPs during adaptive reprogramming reveals increased cellular stress, ROS signaling and DNA damage

To investigate the molecular changes that NEP subpopulations undergo upon injury to the EGL, in particular an increase in the signaling pathways associated with injury induced cellular stress and ROS, we performed multiplexed scRNA-seq of CFP+ cells isolated by fluorescence-activated cell sorting (FACS) of cerebella from Nes-Cfp/+ transgenic neonates either irradiated at P1 (IR; P2, P3, P5) or non-irradiated (nonIR; P1, P2, P3, P5) (Figure 1A, Supplementary Figure 1A, B). Following the filtering out of poor quality cells and integration of replicates and the two conditions, the clustering of 11,878 cells (6,978 nonIR and 4,900 IR) was performed (Hao et al., 2021). The analysis revealed the expected three distinct groups of cells: gliogenic-NEPs and astrocytes (Hopx+ clusters 2, 3, 6, 10), neurogenic-NEPs (Ascl1+ clusters 5, 8, 11) and GCP (Atoh1+ clusters 1, 4, 7, 12, 14) that were present at each stage and in both conditions (Figure 1B-E, Supplementary Figure 1C, Table S2). These groups of cells were further subdivided into molecularly distinct clusters based on their cell cycle profiles or developmental stages (Table S2). In addition, oligodendrocyte progenitors (cluster 15), microglia (clusters 17, 21), ependymal cells (clusters 13, 18, 19) and meninges (cluster 16) were detected (Figure 1B, D, Table S2). Cluster 20 represented low-quality cells and was omitted from downstream analyses. As expected, the GCP clusters were enriched in the cells from irradiated mice and at P5 (Supplementary Figure 1C).

Open in new tabFigure 1: Injury induces ROS and cell stress signaling reflected by changes in the transcriptome and chromatin landscape of gliogenic-NEPs.

(A) Schematic summarizing the experimental plan.

(B-C) UMAP projections of 11,878 cells (6,978 nonIR and 4,900 IR) showing cluster annotations (B), treatment (black: nonIR, red: IR) and the age of the samples (red: P1, green: P2, blue: P3, purple: P5) (C).

(D) Dot plot showing the expression levels of key marker genes used for cluster annotation (gliogenic-NEPs: Hopx, Gdf10, Slc1a3, neurogenic-NEPs: Ascl1, Ptf1a, immature neurons: Pax2, GCPs: Atoh1, postmitotic neurons: Tubb3, microglia: Cx3cr1, oligodendrocyte progenitors: Olig2, oligodendrocytes: Gpr17, Ependymal cells: Foxj1).

(E) Feature plots showing Hopx (gliogenic-NEPs), Ascl1 (neurogenic-NEPs) and Atoh1 (GCPs) expression highlighting the three main populations of interest. Clusters containing Hopx-NEPs (clusters 2, 3, 6, 10), Ascl1-NEPs (clusters 5, 8, 11), or GCPs (clusters 1, 4, 7, 12, 14) were subsetted from the original data set and were divided according to age (P2 or P3+P5) for the downstream differential expression analyses.

(F) Volcano plot showing differentially expressed genes in the P2 gliogenic-NEPs (red: adjusted p-value≤0.05, log₂fold-change>|1|).

(G) Violin plots showing some of the top differentially expressed genes in P2 gliogenic-NEPs and how their expression changes over time with respect to their expression in control cells.

(H) Top GO terms associated with differentially expressed genes in P2 gliogenic-NEPs that were either upregulated in nonIR (top panel) or IR (bottom panel) cells (adjusted p-value≤0.05, Table S3).

(I) Heatmap showing differentially open chromatin regions in P2 nonIR and IR NEPs, identified by bulk ATAC-seq (1168 differentially open regions, adjusted p-value<0.05, Table S4).

(J) Tracks highlighting the injury-induced open chromatin regions around Cdkn1a and Phlda3, the top differentially expressed genes identified in (F).

(K) Gene network analysis of ATAC-seq data revealed an active transcriptional network involved in regulating cell death and apoptosis. Genes colored in red (Ppara, Egln3, Foxo3, Jun and Nos1ap) have been implicated as upregulated with increased ROS levels or involved in ROS signaling.

Our further analyses focused on changes in the signaling pathways associated with injury induced cellular stress and ROS. We previously showed that a subset of the Hopx+ gliogenic-NEPs that are present in the BgL are the ones that undergo adaptive reprograming following GCP death (Wojcinski et al., 2017, Bayin N. S., 2021). We therefore assessed the immediate and later effects of GCP death on Hopx+ gliogenic-NEPs by performing differential expression analyses between nonIR and IR gliogenic-NEPs (Hopx+, clusters 2, 3, 6, 10) at P2, or at P3 and P5 (P3+5). 24 hr after injury at P1, 132 genes in gliogenic-NEP clusters were significantly upregulated in IR compared to 34 genes that were upregulated in nonIR P2 cells (adjusted p-value≤0.05, Figure 1F, Table S3). The significantly increased genes included Cdkn1a, Phlda3, Ass1 and Bax, all of which have been implicated as increased in response to DNA damage and in ROS signaling, as well as in anti-apoptotic functions (Figure 1G) (Masgras et al., 2012, Bensellam et al., 2019, Qiu et al., 2014, Jiang et al., 2008). Indeed, some of the top gene ontology (GO) terms associated with the genes upregulated in gliogenic-NEPs with injury were related to response to irradiation, DNA damage, the P53 pathway and ROS metabolic processes, whereas the top GO terms associated with the genes upregulated in the nonIR cells at P2 were related to metabolic processes (p-value≤0.05, Figure 1H, Table S4). Interestingly, the transcriptional changes observed at P2 were less pronounced at later time points, where although some of the top differentially expressed genes at P2 were still significantly upregulated at P3+5 stages combined (e.g. Cdkn1a, Phlda3, Ass1, Bax), the increase in expression levels of these genes upon injury and/or the number of cells expressing them gradually declined after P2 in IR gliogenic-NEPs when compared to their nonIR counterparts (Figure 1G). Genes upregulated in IR P3+5 gliogenic NEPs were associated with similar GO terms to those at P2, such as response to irradiation and the P53 pathway, however, “ROS metabolic processes” was no longer a significantly enriched GO term (p-value≤0.05, Supplementary Figure 2A, F, Table S3-4).

To assess whether the injury induced transcriptional signatures are specific to gliogenic-NEPs, we performed the same differential expression analysis on nonIR and IR neurogenic-NEPs (Ascl1+, clusters 5, 8, 11) and GCPs (Atoh1+, clusters 1,4,7,12,14) at P2, or at P3+P5 to identify the immediate and later changes upon injury at birth. Similar to the gliogenic-NEPs, P2 IR neurogenic-NEPs showed significant upregulation of genes associated with GO terms related to stress response and apoptosis following injury, although the ROS related GO term was not prominent (Supplementary Figure 2B, G, Table S3-4). Interestingly, although IR neurogenic-NEPs at P3+5 had only 27 genes that were significantly upregulated following injury (adjusted p-value≤0.05, Table S3, Supplementary Figure 2C), the genes included ones associated with neural stem cells and BgL-NEPs (e.g. Id1, Apoe, Ednrb). The latter genes could represent the transitory Ascl1+ BgL-NEP population that induces Ascl1 expression during adaptive reprogramming and would be included in the Ascl1+ neurogenic clusters or reflect the delayed neurogenesis of neurogenic-NEPs previously demonstrated (Bayin N. S., 2021). Consistent with the latter, the nonIR P3+5 neurogenic-NEPs showed an increase in mature neuron markers compared to IR cells (adjusted p-value<0.05, Supplementary Figure 2C, H, Tables S3-4).

In contrast to the gliogenic- and neurogenic-NEP subtypes, P2 IR GCPs showed upregulation of genes enriched in GO terms such as neural differentiation, axonogenesis and nonIR P2 GCPs showed increased expression of genes involved in cell cycle and mitosis (Supplementary Figure 2D, I). This result could reflect the death of highly proliferative GCPs after irradiation and sparing of only postmitotic granule cells upon irradiation at P1. P3+5 IR GCPs showed increased expression of genes associated with BgL-NEPs (Id1 and Gdf10, adjusted p-value ≤ 0.05) and genes associated with GO terms such as cell cycle and cell fate commitment whereas P3+5 nonIR GCPs showed enrichment for GO terms related to cell migration and neurogenesis (adjusted p-value ≤0.05, Supplementary Figure 2D, E, I, J, Tables S3-4). This result could reflect the delayed neurogenesis of GCPs following injury. Interestingly, we did not observe significant enrichment for GO terms associated with cellular stress response in the GCPs that survived the irradiation compared to controls (Table S4). Collectively, these results indicate that the gliogenic- and neurogenic-NEP subtypes transiently upregulate stress response genes upon GCP death, but only gliogenic-NEPs have a strong upregulation of ROS signaling.

Injury induces changes in NEP chromatin landscape at P2

We next tested whether GCP death at birth induces changes to the chromatin landscape of NEPs that reflect the altered gene expression observed with scRNA-seq, by performing bulk ATAC-seq on FACS-isolated CFP+ cells from P2 control and injured Nes-Cfp/+ pup cerebella (Figure 1A). P2 was chosen as it is the stage when GCPs contribute the least to the total Nes-CFP+ FACS population and to identify the immediate effects of the injury on NEP chromatin. Analysis of differentially open chromatin showed that injury induces major changes to the chromatin landscape of the NEPs (1168 differentially open regions, adjusted p-value<0.05, Figure 1I. Table S5). Of interest, Cdkn1a and Phlda3, two genes stimulated by ROS and injury (Bensellam et al., 2019, Masgras et al., 2012) and that were upregulated in gliogenic-NEPs after irradiation (Figure 1F-G, Supplementary Figure 2B) exhibited new accessible regions around their gene bodies compared to the nonIR P2 Nes-CFP+ cells (Figure 1J). Known motif analysis in the regions with increased accessibility upon injury showed enrichment for binding motifs of the JUN/AP1 transcriptional complex (p-value = 10^-14, % of target sequences with motif = 15%, Table S6) which is known to act in response to cellular stress and be activated by ROS. In addition, the DNA binding site for FOXO3, a transcription factor that regulates ROS levels, had increased chromatin accessibility (p-value = 0.001, % of target sequences with motif = 22.41%) (Table S6) (Filosto et al., 2003, Auten and Davis, 2009, Hagenbuchner et al., 2012). Finally, gene network analysis of ATAC-seq data revealed an active transcriptional network involved in regulating cell death and apoptosis (Figure 1K). Furthermore, some of the genes involved in this response, such as Ppara, Egln3, Foxo3, Jun and Nos1ap, have been implicated as upregulated with increased ROS levels or involved in ROS signaling (Devchand et al., 2004, Kaelin, 2005, Hagenbuchner et al., 2012, Filosto et al., 2003). In summary, our ATAC-seq data analysis along with the scRNA-seq provide strong evidence that irradiation causes increased ROS signaling predominantly in the BgL-NEPs upon GCP death by inducing transcriptional and epigenetic changes within 1 day after injury (P2). In addition, genes related to cell survival and death, cellular stress and DNA damage are upregulated in NEPs shortly upon injury, possibly as a means to overcome the cellular effects of injury and induce adaptive reprogramming of NEPs to replenish the lost cells.

Transient increase in cerebellar ROS during apoptosis of granule cell precursors

To validate that the transient increase in expression of genes associated with cellular stress and ROS signaling in NEPs is due to an increase in ROS upon cerebellar injury, we quantified ROS levels in whole cerebellum samples using a mitochondrial superoxide indicator (MitoSOX) via flow cytometry. A significant increase in ROS (cells present in the top 90% MitoSOX+ intensity) was observed specifically at P2 (p=0.0005, n≥10) and not at P3 or P5 in IR pups compared to nonIR (Figure 2A, B and Supplementary Figure 3A, B). Furthermore, quantification of mitochondrial mass with MitoTracker revealed a reduction only at P2 (p=0.0005, n≥10) in IR pups (Figure 2C and Supplementary Figure 3C-E). Additionally, quantification of TUNEL staining in midline sagittal sections of the cerebella showed a large increase in cell death in the EGL of injured cerebella at P2 (p=0.0040, n=4), but not at P3 compared to the controls (Figure 2D, E; see also Figure 3K, L). Outside the EGL there was a small increase in cell death after injury that was only significant at P3 (p=0.0379, n≥3) (Figure 2F, G). Thus, a transient increase in ROS in the cerebellum one day after irradiation at P1 correlates with the timing of the major death of GCPs.

Open in new tabFigure 2: Cerebellar injury at P1 results in increased superoxide production, a reduction in mitochondria and increased cell death in the EGL that peaks 24h after injury.

(A) Examples of flow cytometry analysis of mitochondrial ROS at P2 from nonIR and IR cerebella using MitoSOX dye. Gating determined the top 90% MitoSOX signal (MitoSOX high cells).

(B, C) Quantification of MitoSOX high (B) and MitoTtracker high (C) expression in nonIR and IR cerebella at P2.

(D, E) Quantification of TUNEL+ cell density in the EGL at P2 (D) and P3 (E) in lobules 3-5 of nonIR and IR mice.

(F, G) Quantification of TUNEL+ cell density outside the EGL at P2 (F) and P3 (G) in lobules 3-5 of nonIR and IR mice.

EGL, External granular layer; SSC, side scatter; P, postnatal day; nonIR, non-irradiated; IR, irradiated. All statistical significance was determined using an unpaired t-test and data are represented as mean ± SEM.

Open in new tabFigure 3: Cerebellar injury at P1 induces transient microglial recruitment to the EGL and prolonged astroglial microenvironment changes in the cerebellum.

(A-H) Immunohistochemical (IHC) staining of medial sagittal cerebellar sections for GFAP (green) in lobules 4/5 of nonIR and IR cerebellum at the stages indicated. Nuclei were counterstained with Hoechst. (B’) and (F’) show high-power images of white dashed line boxes in (B) and (F), respectively.

(I, J) Quantification of IBA1+ cell density in the WM at P2 (I) and P3 (J) in lobules 3-5 of nonIR and IR mice.

(K, L) IHC staining of medial sagittal cerebellar sections for IBA1 and TUNEL in lobule 3 of nonIR and IR cerebellum at P2. Nuclei were counterstained with Hoechst. White matter (WM) and external granular layer (EGL) are delineated by white dotted lines and dashed lines, respectively. High-power image in (L) of the area indicated by the white dashed line represents an IBA1+ cell present in the EGL. White arrowheads indicate additional IBA1+ cells in the EGL.

(M, N) Quantification of IBA1+ cell density in the EGL at P2 (M) and P3 (N) in lobules 3-5 of nonIR and IR mice. EGL, External granular layer; WM, White matter; P, postnatal day; nonIR, non-irradiated; IR, irradiated. Scale bar: A and E 100μm, B, C, D, E, F, G and H: 250μm, B’ and F’: 50μm, I and J: 50μm. All statistical significance was determined using an unpaired t-test and data are represented as mean ± SEM.

Altered glial microenvironment following death of granule cell precursors

Given the potential importance of glial cells to regenerative cellular responses after injury, we next asked whether the glial microenvironment of the cerebellum changes during early postnatal development in response to irradiation at P1. We first analyzed the astrocyte marker GFAP, since it is generally upregulated in soon after brain injury (Burda and Sofroniew, 2014). Most astrocytes, including the specialized Bg, express GFAP in the adult cerebellum, but the cells are generated by gliogenic-NEPs during the first two weeks after birth. We therefore determined the normal location and timing of initiation of GFAP expression in these cells during postnatal cerebellum development. GFAP was first detected in rare astrocytes at P2 located in the white matter (WM) below the lobules, with strong expression in all WM astrocytes at P5 and later stages (P8 and P12) (Figure 3A-D, observed in n=4 mice/stage). In contrast, GFAP expression in astrocytes in the developing IGL and in the glial processes of Bg that project through the molecular layer (ML) and EGL was not detectable until P5 and was much stronger at P8 and P12 (Figure 3A-D). Interestingly, after injury at birth, GFAP expression was prematurely upregulated in the WM astrocytes below the lobules at P2 and in the remaining astrocytes at P5, including in the fibers of Bg that extend through the ML and EGL, compared to nonIR cerebella (Figure 3A-H, observed in n=4 mice/stage). GFAP expression was similar in all glia in both conditions at P8 and P12. These results reveal that astrocytes and Bg react to EGL injury caused by irradiation at P1 by initiating GFAP expression earlier than normal in the deep WM, and then the IGL and in Bg.

The macrophages of the brain, microglia, are generated in the early embryo but their main increase in cell number occurs during neonatal development in mice (Hammond et al., 2018). We found that most microglia were located in the WM of the cerebellum, and that the density of IBA1+ microglia increased in the WM between P2 (235.1 ± 18.3 cells/mm²) and P3 (344.3 ± 53.1 cells/mm²) and then was maintained at P5 (350.4 ± 40.2 cells/mm²) (Fig. 3I-K, Supplementary Figure 3F). Since the area of the cerebellum is increasing between P2 and P5, active microglia production and/or infiltration must continue to occur between P2 and P5. As expected, after irradiation the density of microglia in the injured EGL was significantly increased (∼3 fold) at P2 but not at P3 or P5 compared to controls (p=0.0331, n=4) (Fig. 3L-N, Supplementary Figure 3G). No difference in the microglial density in the WM was detected between conditions at both time points (Fig. 3I-L; Supplementary Figure 3F). Thus, as expected microglia density was increased in the EGL one day after injury when the maximum GCP cell death occurs.

Decreasing ROS impairs cerebellar repair

Given the transient increase in ROS signaling in gliogenic-NEPs during peak GCP death and recruitment of microglia to the EGL, and the later astroglial response, we tested whether an increase in ROS is necessary for adaptive reprogramming and cerebellar repair following EGL injury at P1. To reduce the level of ROS we utilized an mCAT transgenic mouse line that expresses the human mitochondrial catalase ubiquitously from a CMV promoter (Schriner et al., 2005). The transgene is expected to reduce mitochondrial ROS levels in all cells by catalyzing the breakdown of hydrogen peroxide into water and oxygen, hence protecting the cells from oxidative damage. We confirmed that mCAT protein is expressed throughout the cerebellum using immunohistochemical (IHC) staining of cerebellar sections (Figure 4A, B). MitoSOX flow cytometry revealed a significant decrease in ROS in the cerebella of mCAT/+ mice 1 day after irradiation (P2) (p=0.0163, n≥8) and not at P3 or P5 compared to control IR mice and no baseline decrease in ROS at any stage in nonIR mice (Figure 4C and Supplementary Figure 4A, B). Mitochondrial mass, as measured by MitoTracker flow cytometry, was reduced at P2 in mCAT/+ IR compared to nonIR mCAT/+ pups with no significant difference observed between mCAT/+ and control IR mice (Fig. 4D, Supplementary Figure 4C, D). The level of cell death in the EGL and density of IBA1+ microglia in the EGL and elsewhere in the cerebellum were similar between mCAT/+ and littermate control IR mice at P2 and P3 (Fig. 4E, F, Supplementary Figure 4G, J). There was a slight increase in cell death outside the EGL at P2 in IR mCAT/+ compared to nonIR mCAT/+ mice but not compared to IR controls (Supplementary Figure 4F, E). Thus, the mCAT transgene counteracts the transient increase in cerebellar ROS following EGL injury but does have a major effect on GCP death or the infiltration of microglia to the injured EGL (Supplementary Figure 4 F-J).

Open in new tabFigure 4: Reduction of ROS impairs adaptive reprogramming and cerebellar repair.

(A, B) IHC staining of medial sagittal cerebellar sections for human catalase in control (A) AND mCAT/+ mice (B) at P7. Nuclei were counterstained with Hoechst (blue). Similar staining was seen in four mCAT/+ mice.

(C) Quantification of MitoSOX high expression at P2 in control and mCAT/+ cerebella, with and without irradiation at P1 (Two-way ANOVA, F_(1,34)=6.768, p=0.0136).

(D) Quantification of MitoTracker high expression at P2 in control and mCAT/+ cerebella, with and without irradiation at P1 (Two-way ANOVA, F_(1,31)=25.06, p<0.0001).

(E) Quantification of TUNEL+ cell density in the EGL at P2 in control and mCAT/+ cerebella, with and without irradiation at P1 (Two-way ANOVA, F_(1,14)=87.56, p<0.0001).

(F) Quantification of IBA1+ cell density in the EGL at P2 in control and mCAT/+ cerebella, with and without irradiation at P1 (Two-way ANOVA, F(1,14)=15.58, p=0.0015).

(G-J) Hematoxylin and eosin staining on mid-sagittal sections of P30 control and mCAT/+ cerebellum with or without irradiation.

(K) Quantification of P30 cerebellar mid-sagittal section area in controls and mCAT/+ nonIR and IR mice (Two-way ANOVA, F_(1,20)=11.82, p=0.0026).

(L) Graph showing the average area of mid-sagittal cerebellar sections at P3, P5, P8, P12 and P30 in control and mCAT/+ non-irradiated and irradiated mice. Detailed statistics are shown in Supplementary Figure 4. EGL, External granular layer; P, postnatal day; nonIR, non-irradiated; IR, irradiated. Scale bar: A and B: 100μm, F-I: 1mm. Significant Tukey’s post hoc multiple comparison tests are shown in the figures and data are represented as mean ± SEM.

We next determined the regenerative efficiency of mCAT/+ mice by analyzing the area of sections of cerebella from nonIR and IR P30 mCAT/+ mice compared to littermate controls. Strikingly, the cross-sectional area of the medial cerebellum (vermis) of IR mCAT/+ adult mice was significantly reduced compared to IR controls (p=0.0040, n≥6) (Fig. 4G-K). Analysis of cerebellar area across ages (P3, 5, 8 and 12) revealed that the vermis sectional area of the IR mCAT/+ cerebella was only significantly reduced at P30 compared to IR controls, however at P12 it was reduced in mCAT/+ IR cerebella compared to mCAT/+ nonIR mice, whereas it was not significantly different between control IR and control nonIR mice (Fig. 4L, Supplementary Figure 4 K-N). These results indicate that cerebellar growth begins to be reduced at P12 in mCAT/+ mice following injury. Thus, a reduction in ROS at the time of cell death in the EGL leads to a diminution of cerebellar recovery.

Reduced regeneration in mCAT mice is associated with reduced adaptive reprogramming at P5

Given that a decrease in ROS following EGL injury reduces regeneration of the neonatal cerebellum, we determined whether specific stages of the adaptive reprogramming process are altered in mCAT/+ mice compared to controls. First, we analyzed the replenishment of the EGL by BgL-NEPs in vermis lobules 3-5. Interestingly, we found that although the thickness of the EGL in IR mice of both genotypes was similarly reduced compared to nonIR mice at P5 (p=0.0041 control and p=0.0005 mCAT/+; n≥4) by P8 the control EGL was a similar thickness to the control nonIR whereas the thickness of the mCAT/+ IR EGL was significantly reduced compared to mCAT/+ nonIR mice (p=0.035, n=4)(Figure 5A, B). A key regenerative process that contributes to the expansion of the EGL following injury is the migration of BgL-NEPs to the EGL. Strikingly, the density of CFP+ cells in the EGL (mainly BgL-derived NEPs) was significantly decreased in mCAT/+ IR mice compared to control IR mice at P5 (p=0.0002, n=4) but not P3 (Figure 5C-G, Supplementary Figure 5A). Furthermore, the density of NEPs (CFP+ or SOX2+ cells) in the BgL was significantly decreased at P5 in mCAT/+ IR cerebella compared to controls (p=0.0010, n≥5) but not at other stages (Figure 5H, Supplementary Figure 5B-D). These results indicate that BgL-NEPs have a blunted response to EGL injury, and therefore do not fully expand and contribute to the replenishment of GCPs in the EGL after irradiation.

Open in new tabFigure 5: Reduced ROS impairs expansion of BgL-NEPs and their recruitment to the EGL after injury.

(A, B) Quantification of EGL thickness at P5 (Two-way ANOVA, F_(1,21)=36.64, p<0.0001)

(A) and P8 (Two-way ANOVA, F_(1,12)=11.34, p=0.0056)(B) in lobules 3-5 of Nes-Cfp/+ control and Nes-Cfp/+; mCAT/+ mutant mice with and without irradiation at P1

(C-F) IHC staining of medial sagittal cerebellar sections showing expression of CFP (green) in the lobules 4/5 of Nes-Cfp/+ control and Nes-Cfp/+; mCAT/+ mutant mice at P5. Nuclei were counterstained with Hoechst (blue). (D’) and (F’) show a high-power images of the yellow boxed area in the single channel CFP. EGL is delineated by the dashed white lines.

(G, H) Quantification of CFP+ cell density in the EGL (Two-way ANOVA, F_(1,19)=5.192, p=0.0359) (G) and BgL (Two-way ANOVA, F_(1,17)=6.191, p=0.0223) (H) at P5 in Nes-Cfp/+ control or Nes-Cfp/+; mCAT/+ mutant non-irradiated and irradiated mice.

EGL, External granular layer; BgL, Bergmann glia Layer; P, postnatal day; nonIR, non-irradiated; IR, irradiated. Scale bar: D-F: 100μm. Significant Tukey’s post hoc multiple comparison tests are shown in the figures and data are represented as mean ± SEM.

Microglia contribute to one aspect of adaptive reprogramming

Given that several steps in adaptive reprogramming were decreased specifically at P5 in mCAT/+ IR cerebella, we asked whether microglia/macrophages could be involved in any of the processes. We first determine the density of IBA1+ cells in the EGL and WM of vermis lobules 3-5 at P5 in nonIR and IR mice of both genotypes. As expected, the density of microglia in the EGL was very low in the nonIR control and mCAT/+ cerebella (Figure 6A-E). Interestingly, whereas control IR mice had a similar number of IBA1+ cells in the WM as nonIR mice of both genotypes at P5, the mCAT/+ IR mice had a lower density of microglia/macrophages in the WM compared to control IR mice at P5 (p=0.0012, n≥3) (Figure 6A-D, F). This result raised the question of whether macrophages/microglia play a role in adaptive reprogramming. We, therefore, tested whether reducing the density of IBA1+ microglia/macrophages after birth would alter adaptive reprogramming at P5 or cerebellar regeneration at later stages. Since macrophages and cerebellar microglia are dependent on colony stimulating factor 1 (CSF1) for their survival, we administered PLX5622, a small molecule inhibitor of CSF receptor 1 (CSFR1), to pups every day from P0-5 (PLX treatment) (Kana et al., 2019, Tan et al., 2021). As expected, IBA1+ cells were significantly decreased in the cerebellum of PLX-treated mice at P5 compared to their controls, both nonIR and IR (p=0.0015 and p=0.0059, n=3 and n=5, respectively) (Figure 6G, Supplementary Figure 6B-E). The thickness of the EGL was not significantly altered at P5 in PLX-treated IR mice compared to IR controls (Figure 6H). Interestingly, similar to mCAT/+ mice, the density of Nes-CFP+ cells in the EGL was significantly decreased in PLX-treated IR mice compared to IR controls at P5 (p=0.0008, n=5) (Figure 6I-M). In contrast, the density of SOX2+ cells in the BgL, corresponding to the gliogenic BgL-NEPs, were unchanged in the PLX-treated and control mice, whether irradiated or not, suggesting that the decrease in expansion of BgL-NEPs caused by ROS is not mediated by microglia (Supplementary Figure 6F). When mice were treated with PLX from P0-8 and allowed to age to P30, we found the cerebellar vermis section area was not decreased in PLX-treated IR mice compared to IR controls (Figure 6N). Thus, reducing the density of IBA1+ microglia/macrophages in neonatal mice reduces the recruitment of Nes-CFP+ cells to the EGL at P5, but does not have a long-term significant impact on regeneration of the cerebellum.

Open in new tabFigure 6: Microglia promote recruitment of NEPs to the EGL during cerebellar adaptive reprogramming after injury.

(A-D) IHC staining of medial sagittal cerebellar sections for IBA1 (red) in control and mCAT/+ mice at P5. Nuclei were counterstained with Hoechst (blue).

(E, F) Quantification of IBA1+ cell density in the external granular layer (E) and white matter (Two-way ANOVA, F_(1,13)=24.74, p=0.0003) (F) at P5 on midsagittal sections of lobules 3-5 in the cerebellum of control and mCAT/+ animals, with or without irradiation.

(G-J) IHC staining of medial sagittal cerebellar sections at P5 for CFP (green) in lobules 4/5 of Nes-Cfp/+ mice treated with PLX5622 or control DMSO with or without irradiation. Nuclei were counterstained with Hoechst (blue). (H’) and (J’) show a high-power image of area indicated by yellow boxes. EGL is delineated by the white dashed lines.

(K) Quantification of IBA1+ cell density in the white matter at P5 on mid-sagittal sections in lobules 3-5 of Nes-Cfp/+ mice treated with PLX5622 or control DMSO, with or without irradiation (Two-way ANOVA, F_(1,12)=42.40, p<0.001).

(L) Quantification of EGL thickness at P5 in lobules 3-5 of Nes-Cfp/+ mice treated with PLX5622 or control DMSO with or without irradiation (Two-way ANOVA, F_(1,12)=109.5, p<0.001).

(M) Quantification of CFP+ cells density in the EGL at P5 on mid-sagittal sections in lobules 3-5 of Nes-Cfp/+ mice treated with PLX5622 or control DMSO with or without irradiation (Two-way ANOVA, F_(1,12)=10.62, p=0.0068).

(N) Measurement of cerebellar mid-sagittal section area at P30 in controls or mice treated with PLX, with or without irradiation at P1 (Two-way ANOVA, F_(1,12)=13.29, p=0.0034).

EGL, External granular layer; WM, White matter; P, postnatal day; nonIR, non-irradiated; IR, irradiated. Scale bar: A-D and G-J: 250 μm. Significant Tukey’s post hoc multiple comparison tests are shown in the figures and data are represented as mean ± SEM.