Autism-associated CHD8 controls reactive gliosis and neuroinflammation via remodeling chromatin in astrocytes

Abstract

Reactive changes of glial cells during neuroinflammation impact brain disorders and disease progression. Elucidating the mechanisms that control reactive gliosis may help us to understand brain pathophysiology and improve outcomes. Here, we report that adult ablation of autism spectrum disorder (ASD)-associated CHD8 in astrocytes attenuates reactive gliosis via remodeling chromatin accessibility, changing gene expression. Conditional Chd8 deletion in astrocytes, but not microglia, suppresses reactive gliosis by impeding astrocyte proliferation and morphological elaboration. Astrocyte Chd8 ablation alleviates lipopolysaccharide-induced neuroinflammation and septic-associated hypothermia in mice. Astrocytic CHD8 plays an important role in neuroinflammation by altering the chromatin landscape, regulating metabolic and lipid-associated pathways, and astrocyte-microglia crosstalk. Moreover, we show that reactive gliosis can be directly mitigated in vivo using an adeno-associated virus (AAV)-mediated Chd8 gene editing strategy. These findings uncover a role of ASD-associated CHD8 in the adult brain, which may warrant future exploration of targeting chromatin remodelers in reactive gliosis and neuroinflammation in injury and neurological diseases.

Keywords: AAV; ASD-associated gene; CHD8; CP: Neuroscience; CRISPR gene editing; astrocyte; brain injury; chromatin remodeling; neuroinflammation; reactive gliosis.

Figure 1.. Reduced reactive gliosis in the global Chd8-cKO mice in the stab-wound injury model

(A) Genetic strategy targeting exon 4 of the Chd8 gene. The exon was flanked with loxP sites to excise the loci, resulting in a frameshift mutation that disrupts the production of the CHD8 protein after Cre excision. (B and C) Representative images of CHD8 in the CA1 regions of adult mouse brains. CHD8 is expressed in neurons and astrocytes in controls (B). After tamoxifen administration, Chd8^fx/fx: CAGGS-CreER^+/− mouse brains showed non-detectable levels of CHD8 protein (C). In both (B) and (C), arrows point to astrocytes expressing CHD8 and its knockout thereafter. (D) Schematic diagram for tamoxifen-induced Chd8 cKO, the stab-wound injury model, and the analysis of reactive gliosis. (E and F) Response of GFAP⁺ astrocytes and Iba1⁺ microglia after stab-wound injury. Control mice (E) exhibit astrocytic and microglial response as expected, while global Chd8 cKO mice (F) show reduced staining for both GFAP⁺ and Iba1⁺ along the needle track. (G and H) High-magnification images from the injury site from the corresponding genotypes. Note the reduction in cell body size, process elongation, and polarity in astrocytes from global Chd8 cKO mice (H). (I and J) Decreased area occupied by GFAP⁺ astrocytes (I) and Iba1⁺ microglia (J) in global Chd8 cKO mice compared to controls. In (B) and (C), scale bars indicate 20 μm; in (G) and (H), scale bars indicate 50 μm; in (E) and (F), scale bars indicate 500 μm. The dashed rectangle indicates the ROIs that were quantified. In (E)–(H), the dashed lines indicate the needle track of the injury. In (I) and (J), data points illustrate the quantified area from the six brain slices most proximal to the injury epicenter. Data are normalized to the means of the ipsilateral site in control mice. ****p < 0.0001, ns, not significant; statistical analysis was performed with two-way ANOVA; on the violin plots, dashed lines indicate the 25%, mean, and 75% percentile, respectively, from bottom to top; n = 4 mice per genotype.

Figure 2.. Reduced reactive gliosis in astrocyte Chd8-cKO mice in the stab-wound injury model

(A) Strategy for tamoxifen-induced, astrocyte-specific Chd8 cKO utilizing the Aldh1l1-CreERT2 line. Mice were crossed with the Ai14-tdTomato reporter line to visualize recombined cells. (B and C) Astrocyte-specific Chd8-cKO mice show non-detectable CHD8 protein expression in cortical slices. Controls are mice without the Chd8 floxed alleles but expressing Aldh1l1-CreERT2 to turn on the expression of tdTomato reporter. Arrows indicate CHD8 expression in tdTomato⁺ astrocytes in control but not astrocyte Chd8 cKO. (D and E) GFAP and Iba1 staining of astrocytes and microglia, respectively, after stab-wound injury in control (D) and astrocyte cKO mice (E). (F and G) High-magnification images from the injury sites from the corresponding genotypes showing reduction in cell body size, process elongation, and polarity in astrocytes and reduced microglia numbers in astrocyte cKO mice. (H and I) Decreased area occupied by GFAP⁺ astrocytes and their numbers in astrocyte cKO mice compared to controls. (J and K) Decreased area occupied by Iba1⁺ microglia and their numbers in astrocyte cKO mice compared to controls. In (H)–(K), data points represent the quantified area from the six brain slices most proximal to the injury epicenter. Data are normalized to the means of the ipsilateral site in control mice. In (B) and (C), scale bars indicate 20 μm; in (F) and (G), scale bars indicate 50 μm; in (D) and (E), scale bars indicate 500 μm. The dashed rectangle indicates the ROIs that were quantified. In (D)–(G), the dashed lines in merged images indicate the needle track of the injury. In (H)–(K), ***p < 0.001; ****p < 0.0001; ns, not significant; statistical comparisons were analyzed with two-way ANOVA; on the violin plots, dashed lines indicate the 25%, mean, and 75% percentile, respectively, from bottom to top; n = 6 mice per genotype.

Figure 3.. Reduced proliferation and impaired morphological changes of reactive astrocytes in astrocyte Chd8-cKO mice

(A and B) Representative images show reduced staining for proliferation markers (BrdU and Ki67) after stab-wound injury in astrocyte cKO mice (B) as compared to control mice (A). The dashed rectangle indicates the ROI used to quantify BrdU⁺ and Ki67⁺ cells in (C)–(F). (C) Decreased BrdU⁺ nuclei in the astrocyte cKO mice as shown by high-magnification images and quantification. (D) Reduced Ki67⁺ nuclei in the astrocyte cKO mice. (E) Decreased proliferation of tdTomato⁺ astrocytes from astrocyte cKO mice as shown by reduced tdTomato⁺/BrdU⁺ colocalized cells. (F) Decreased proliferation of tdTomato⁺ astrocytes from astrocyte cKO mice as shown by reduced tdTomato⁺/BrdU⁺ colocalized cells. (G and H) Representative skeletonized images of two astrocytes from control (G) and astrocyte cKO mice (H). (I) Sholl analysis of astrocytes localized within 300 μm of the needle track revealed reduced intersections in astrocyte cKO mice at distances between 18 and 35 μm from the soma, indicating reduced cell elongation. Conversely, astrocyte cKO mice featured astrocytes with processes concentrated closer to the cell soma (n = 24 cells from 4 mice per genotype). (J) Decreased territory volume of astrocytes from astrocyte cKO mice, indicating reduced hypertrophy and elongation toward the injury site. (K) Smaller cellular volume in astrocytes from astrocyte cKO mice. (L and M) Fewer endpoints (L) and reduced number of branchpoints (M) in astrocytes from astrocyte cKO mice. In (A) and (B), scale bars indicate 200 μm; in (C)–(F), scale bars indicate 50 μm. The dashed rectangles indicate the areas that were quantified. *p < 0.05; **p < 0.01; ****p < 0.0001; ns, not significant. Statistical comparisons were analyzed with the two-tailed Welch’s t test; error bars indicate the SEMs. In (I), asterisks indicate the range where the number of intersections is significantly different, analyzed through the mixed-effects model. In (C)–(F) and (J)–(M), n = 4 mice per genotype.

Figure 4.. Impaired microglial response and alleviated hypothermia in LPS-treated astrocyte cKO mice

(A–C) Representative images from the cortex of saline-treated control mice (A), LPS-treated control mice (B), and LPS-treated astrocyte cKO mice (C). GFAP⁺ staining does not show a difference between genotypes after LPS stimulation. In LPS-treated astrocyte cKO mice (C), microglia reactivity, as indicated by the Iba1⁺ signal, is reduced as compared to controls (B). (D and E) High-magnification images of microglia from LPS-treated control (D) and LPS-treated astrocyte cKO mice (E) in the cortex, indicating the different reactive characteristics of Iba1⁺ microglia. (F and G) Representative skeletonized images of two microglia from control (F) and astrocyte cKO mice (G). (H) Sholl analysis of microglia from both genotypes reveals a smaller number of intersections in microglia from astrocyte cKO mice (n = 40 cells from 3 mice per genotype). (I) Decreased microglial cell territory volume after LPS treatment in astrocyte cKO mice. (J) Reduced microglial cell volume in astrocyte cKO mice. (K–M) Microglia from astrocyte cKO mice feature a reduced number of branchpoints (K) and endpoints (L) and maximum branch lengths (M). (N) Typical temperature drop after LPS injection in control mice compared to saline-treated animals. (O) Astrocyte cKO animals experience no drop in temperature compared to the saline-treated group after LPS administration. (P) Survival curve depicting the observed mortality of controls after LPS injections. On the contrary, astrocyte cKO animals did not experience mortality (p = 0.0412, logrank [Mantel-Cox] test, n = 14 mice for control group; n = 16 mice for astrocyte cKO group). In (A)–(C), scale bars indicate 200 μm; in (D) and (E), scale bars indicate 50 μm. In (H), asterisks indicate the range where the number of intersections is different, statistical analysis was performed with the mixed-effects model. In (I)–(M), *p < 0.05; **p < 0.01; ***p < 0.001; statistical analysis was carried out with the two-tailed Welch’s t test; error bars indicate the SEM; data points represent n = 3 mice per genotype. Data points are normalized to the mean of each control group.

Figure 5.. Transcriptional analysis of cortical tissues from control and astrocyte Chd8 cKO mice after LPS administration

(A) Heatmap of DEGs identified through RNA-seq between control and astrocyte Chd8 cKO mice after LPS treatment (n = 5 mice per genotype). A total of 109 DEGs were identified, 76 of which were upregulated and 33 were downregulated (FDR < 0.05). (B) Volcano plot depicting the distribution of upregulated and downregulated genes, relative to their quantified fold change and their corresponding p values. The threshold was set at p (adjusted) < 0.05. (C) Venn diagram of the detected DEGs, depicting a subset of DEGs that correspond to genes whose expression is specific to astrocytes, neurons, and microglia. Of those, many DEGs (41) were determined to be astrocyte specific, while fewer were deemed to be specific in neurons (4) and microglia (1). (D) Bar plot showing the fold enrichment of the detected DEGs in our dataset, indicating significant enrichment of DEGs for astrocyte marker genes. Notably, no enrichment was detected when comparing neuronal or microglial genes to the cell-type markers from previous studies, as cited. ***p < 0.001; ****p < 0.0001. (E) GO terms analysis reveals changes associated with many cellular processes, including lipid and metabolic pathways in astrocyte cKO mice in response to LPS stimulation. (F) Heatmap of the top 20 DEGs identified through RNA-seq. Of these, 13 were upregulated and 7 were downregulated in astrocyte cKO mice treated with LPS vs. control mice treated with LPS. (G–O) qPCR analysis of Gstt3 (G), Acsl3 (H), Etnppl (I), Phykpl (J), Gjb6 (K), Slc9a8 (L), Agt (M), Tnfrsf25 (N), and Lcat (O) mRNA confirms the altered expression shown in (F) (n = 4 mice per group). (P) Representative western blotting of TNFRSF25 from cortices of control and astrocyte cKO mice after LPS administration, showing reduced TNFRSF25 protein in the cortex of astrocyte cKO mice (n = 7 mice per group). (Q) Representative western blotting of LCAT from cortices of control and astrocyte cKO mice after LPS administration showing reduced LCAT protein in astrocyte cKO mice (n = 7). (R) Representative images of TNFRSF25 staining in the cortex of control and astrocyte cKO mice after LPS administration. Signal intensity quantification of TNFRSF25 staining (n = 4 mice per genotype). *p < 0.05; scale bars indicate 200 μm. Data points are normalized to the mean of the control group. Error bars depict the SEM. Statistical comparisons were performed with the one-tailed Welch’s t test.

Figure 6.. CHD8 mediates chromatin accessibility changes during LPS-induced neuroinflammation

(A) Flowchart of the experimental pipeline used to obtain astrocyte-enriched mouse brain samples for the subsequent ATAC-seq processing. (B) Genomic annotation enrichment for altered (increased and decreased) chromatin accessibility between saline-treated control mice and LPS-treated control mice. (C) Genomic annotation enrichment for altered (increased and decreased) chromatin accessibility between LPS-treated astrocyte cKO mice and LPS-treated control mice. (D) Heatmap display of ATAC-seq deviations in chromatin accessibility across the three conditions. (E) Heatmap representation of changes in ATAC-seq peaks near DEGs identified in our bulk RNA-seq experiments in Figure 5A. (F and G) Representative genome tracks showing loss of accessibility proximal to transcription start sites of the Basp1 gene identified in a previous study (F) and of the Lcat gene identified in our RNA-seq (G), in the astrocyte cKO samples.

Figure 7.. CRISPR-SaCas9-mediated Chd8 editing through AAV in astrocytes mitigates reactive gliosis in the stab-wound injury model

(A) Schematic diagram illustrating the elements required for the designed AAV for astrocyte-specific Chd8 editing in vivo via CRISPR-SaCas9. (B) Diagram for the simultaneous AAV injection and stab-wound injury with analysis of reactive gliosis performed at 7 days post-injection. (C) Representative images near the needle track from control mice (Scramble-AAV injected). CHD8 is detectable in astrocytes (SOX9⁺) expressing SaCas9 (HA tag⁺). (D) Representative images near the needle track from mice injected with the Chd8-cKO AAVs. CHD8 is undetectable in the majority of HA⁺ and SOX9⁺ astrocytes (white arrows), while fewer HA⁺ and SOX9⁺ astrocytes still show CHD8 expression (yellow arrows). (E and F) GFAP and Iba1 staining of astrocytes and microglia, respectively, after stab-wound injury and AAV injection in the Scramble-AAV (E) and Chd8-cKO-AAV groups (F). (G) Decreased area occupied by GFAP⁺ astrocytes in the Chd8-cKO-AAV mice. (H) Quantification of the area occupied by Iba1⁺ microglia between control and Chd8-cKO-AAV mice. In (C) and (D), scale bars indicate 20 μm; in (E) and (F), scale bars indicate 500 μm. The dashed rectangle indicates the ROIs that were quantified in (G) and (H). The dashed lines in merged images indicate the needle track of the injury. In (G) and (H), statistical comparisons were performed with two-way ANOVA; **p < 0.01. Data points indicate n = 3 mice per group. Data are normalized to the means of the ipsilateral site in the control group.