Abstract
The chromatin remodeler CHD8 represents a high-confidence risk factor in autism, a multistage progressive neurologic disorder, however the underlying stage-specific functions remain elusive. In this study, by analyzing Chd8 conditional knock-out mice (male and female), we find that CHD8 controls cortical neural stem/progenitor cell (NSC) proliferation and survival in a stage-dependent manner. Strikingly, inducible genetic deletion reveals that CHD8 is required for the production and fitness of transit-amplifying intermediate progenitors (IPCs) essential for upper-layer neuron expansion in the embryonic cortex. p53 loss of function partially rescues apoptosis and neurogenesis defects in the Chd8-deficient brain. Further, transcriptomic and epigenomic profiling indicates that CHD8 regulates the chromatin accessibility landscape to activate neurogenesis-promoting factors including TBR2, a key regulator of IPC neurogenesis, while repressing DNA damage- and p53-induced apoptotic programs. In the adult brain, CHD8 depletion impairs forebrain neurogenesis by impeding IPC differentiation from NSCs in both subventricular and subgranular zones; however, unlike in embryos, it does not affect NSC proliferation and survival. Treatment with an antidepressant approved by the Federal Drug Administration (FDA), fluoxetine, partially restores adult hippocampal neurogenesis in Chd8-ablated mice. Together, our multistage functional studies identify temporally specific roles for CHD8 in developmental and adult neurogenesis, pointing to a potential strategy to enhance neurogenesis in the CHD8-deficient brain.
SIGNIFICANCE STATEMENT The role of the high-confidence autism gene CHD8 in neurogenesis remains incompletely understood. Here, we identify a stage-specific function of CHD8 in development of NSCs in developing and adult brains by conserved, yet spatiotemporally distinct, mechanisms. In embryonic cortex, CHD8 is critical for the proliferation, survival, and differentiation of both NSC and IPCs during cortical neurogenesis. In adult brain, CHD8 is required for IPC generation but not the proliferation and survival of adult NSCs. Treatment with FDA-approved antidepressant fluoxetine partially rescues the adult neurogenesis defects in CHD8 mutants. Thus, our findings help resolve CHD8 functions throughout life during embryonic and adult neurogenesis and point to a potential avenue to promote neurogenesis in CHD8 deficiency.
- ATP-dependent chromatin remodeler
- autism spectrum disorders
- CHD8
- embryonic and adult neurogenesis
- neural stem/progenitors
- transit-amplifying intermediate progenitor
Introduction
Autism spectrum disorder (ASD) is a multistage progressive disorder with significant genotypic and phenotypic complexity (An and Claudianos, 2016; Courchesne et al., 2019). Deficits in early neural development are commonly believe to underly autism etiology (Packer, 2016; Varghese et al., 2017). However, the clinical symptoms of ASD persist into adulthood (Bernier et al., 2014; Courchesne et al., 2019); currently, the mechanisms that link autism-risk factors to stage-specific defects during developmental and adult neurogenesis remain incompletely understood.
Neurogenesis is essential for the production of the proper number of neurons and glial cells from neural stem cells (NSCs) and neural progenitor cells (NPCs) in the brain and occurs throughout life (Hsieh and Gage, 2004; Delgado et al., 2020). In the developing cortex, stem-like radial glial cells (RGCs) differentiate into neurons either directly from NPCs or indirectly via secondary basal progenitors such as intermediate progenitor cells (IPCs) marked by TBR2 [also known as EOMES (Eomesodermin); Noctor et al., 2004; Kriegstein and Alvarez-Buylla, 2009; Mihalas et al., 2016]. In contrast to direct neurogenesis, which only produces limited neurons per cell cycle, indirect neurogenesis enables phenomenal production and amplification of cortical neurons, particularly those destined to form the superficial layers, and leading to expansion of the mammalian cortex during evolution (Cardenas et al., 2018; Borrell, 2019). Transit-amplifying IPCs are vital for generating appropriate numbers of upper-layer excitatory neurons (Sessa et al., 2008; Kriegstein and Alvarez-Buylla, 2009), a process known to be disrupted in autism (Velmeshev et al., 2019).
In the adult brain, neurogenesis occurs in two specific regions, the subventricular zone (SVZ) and subgranular zone (SGZ). Newborn neurons differentiated from a population of IPCs in the SVZ migrate via the rostral migratory stream to the olfactory bulb and differentiate further into interneurons, whereas new dentate granule cells are generated from the SGZ neural progenitor cells in the hippocampus (Ming and Song, 2011; Hsieh and Zhao, 2016; Obernier et al., 2018). Adult neurogenesis has been shown to be critical for normal neuronal functions and neural circuitry (Aimone et al., 2006; Deng et al., 2010; Song et al., 2012).
Previous studies showed that loss-of-function disruptive mutations in CHD8, encoding the ATP-dependent-chromodomain-helicase-DNA-binding-protein-8, can cause ASD (Bernier et al., 2014; Cotney et al., 2015; Stolerman et al., 2016). Haploinsufficient mutation or knockdown animal models share several clinical features with ASD patients including impaired cortical development and plasticity, macrocephaly, social behavioral abnormalities, and intellectual impairment (Durak et al., 2016; Katayama et al., 2016; Gompers et al., 2017; Platt et al., 2017; Jung et al., 2018; Suetterlin et al., 2018; Xu et al., 2018; Jiménez et al., 2020; Ellingford et al., 2021; Hurley et al., 2021). Gene knock-out studies indicate that Chd8 is critical for cortical progenitor development and survival as well as neural specification from embryonic stem cells (ESCs; Sood et al., 2020; Hurley et al., 2021; Kweon et al., 2021). However, at present, the stage- and cell-type-specific effects of the CHD8 loss of function on neural progenitor survival and neurogenic processes in developing and adult brains remain elusive.
In this study, we use developmentally regulated and neural progenitor-specific Chd8 ablation approaches to investigate the effect of CHD8 on embryonic and adult neurogenesis and find that CHD8 ablation in NSCs impairs their proliferation and survival in a stage-dependent manner. Strikingly, we identify a critical requirement for CHD8 in the generation and survival of transit-amplifying IPCs for indirect neurogenesis. Transcriptomics and epigenomic analyses further indicate that CHD8 maintains proper chromatin landscape and regulates neurogenesis-related programs including TBR2, a key regulator of IPC neurogenesis. In the adult brain, CHD8 ablation in NSCs inhibits neurogenesis via IPCs in both forebrain SVZ and hippocampal SGZ; however, it does not affect NSC proliferation and survival. Furthermore, pharmacological enhancement of NSC differentiation by the antidepressant fluoxetine, approved by the Federal Drug Administration (FDA), partially enhances neurogenesis in CHD8-depleted adult mice. Thus, our observations reveal that CHD8 exhibits conserved, yet spatiotemporally distinct, functions in developmental and adult neurogenesis, pointing to a potential therapeutic avenue of augmenting neurogenesis in the CHD8-deficient brain.
Materials and Methods
Animals
Mice homozygous for Chd8 floxed alleles (Chd8fl/fl; Zhao et al., 2018) were crossed with Emx1-Cre+/− mice (Gorski et al., 2002; from Kenneth Campbell) to generate Chd8 conditional KO (cKO; Chd8fl/fl;Emx1-cre+/−), Chd8 cHet (Chd8fl/+;Emx1-cre+/−), and their littermate wild-type control. Nestin-CreER+/− mice (from Ryo Kageyama) were crossed with Chd8-flox mice to generate Chd8 inducible KO (iKO; nestin-CreERT2+/−; Chd8fl/fl) and its control (nestin-CreERT2+/−). Nestin-Cre mice (stock #003771) and the reporter mice CAG-CAT-EGFP (ccGFP; stock #024636) were obtained from The Jackson Laboratory. We bred the ccGFP reporter line with Chd8fl/fl mice to obtain Chd8fl/fl; ccGFP+/− mice, which then crossed with Chd8 cHet mice to produce Chd8 cKO, ccGFP (Chd8fl/fl;Emx1-cre+/−;ccGFP+/−) and Chd8 cHet, ccGFP (Chd8fl/+;Emx1-cre+/−; ccGFP+/−) mice for fluorescence-activated cell sorting (FACS) and assay for transposase-accessible chromatin using sequencing (ATAC-seq) in relation to wild-type or Chd8fl/fl controls. Both male and female animals were used in the developmental and behaviors study. All mice used in experiments were maintained on a mixed C57/Bl6;129Sv background and housed in a pathogen-free vivarium with a 12 h light/dark cycle and normal chow food and water ad libitum. Mouse embryos harvested at different ages with embryonic day (E)0.5 considered the time of vaginal plug. All animal experiments were approved by the Institutional Animal Care and Use Committee of Cincinnati Children's Hospital Medical Center.
Tamoxifen and fluoxetine administration
Tamoxifen (TAM; catalog #T5648, Sigma-Aldrich) was dissolved in a vehicle of ethanol and sunflower seed oil (1:9 v/v; catalog #S5007, Sigma-Aldrich) as a concentration of 20 mg/ml. For adult neurogenesis experiments, mice were intraperitoneally injected with 100 µl tamoxifen once daily for 5 consecutive days. For embryonic-stage-specific experiments, pregnant mice on certain embryonic days were intraperitoneally injected with tamoxifen (100 mg/kg) once daily for 3 consecutive days. Fluoxetine (FLX) was obtained from Spectrum Chemical. Mice in adult neurogenesis rescue experiments were intraperitoneally injected with FLX 15 mg/kg daily for 21 d (Ohira et al., 2013). Treatments were administrated to both control and mutant mice. Mice were analyzed at different days after tamoxifen administration as designed.
Tissue processing, immunohistochemistry, and immunoblotting
Mice at defined ages were anesthetized with isoflurane and transcardially perfused with PBS, pH 7.4, followed by 4% PFA. Embryonic and postnatal brains were dissected and fixed in 4% PFA for 45 min (embryonic) and 2 h (postnatal) at 4°C. After washing in PBS, brain samples were dehydrated in 25% sucrose over night at 4°C, embedded in optimal cutting temperature medium, and cryosectioned at 20 µm. For those paraffin sectioned samples, dissected brains were fixed in 4% PFA overnight, embedded in paraffin, and sectioned at 5 µm. Both male and female animals were used for the study.
Cryosection slides were washed in PBS three times followed by permeabilization with 0.2% Triton X-100 in PBS for 20 min and blocking with blocking buffer (5% normal donkey serum in PBS) at room temperature. Primary antibodies were incubated overnight at 4°C. Antibodies used in the study were the following: rabbit anti-CHD8 (catalog #ab84527, Abcam), rabbit anti-Cux1 (catalog #sc-13024, Santa Cruz Biotechnology), rat anti-CTIP2 (catalog #ab18465, Abcam), mouse anti-TUJ1 (catalog #801201, BioLegend), goat anti-Sox2 (catalog #sc-17320, Santa Cruz Biotechnology), mouse anti-p53 (catalog #2524, Cell Signaling Technology), rabbit anti-cleaved caspase 3 (catalog #9661, Cell Signaling Technology), rabbit anti-TBR2 (catalog #ab23345, Abcam), rabbit anti-phospho-Histone H3, pH 3 (catalog #PA5-17869, Thermo Fisher Scientific), mouse anti-BrdU (5-Bromo-2′-deoxyuridine; catalog #555627, BD Pharmingen), rat anti-BrdU (catalog #ab6326, Abcam), mouse anti-neuronal nuclei (NeuN; catalog #MAB377, Millipore), goat anti-doublecortin (DCX; catalog #sc-8066, Santa Cruz Biotechnology), rabbit anti-PAX6 (catalog #ab5790, Abcam), rabbit anti-GFAP (catalog #13-0300, Cell Signaling Technology), rabbit anti-Iba1 (catalog #019-19741, Wako Pure,), mouse anti-γH2A.X (catalog #05-636, Millipore), rabbit anti-Calbindin (CB; catalog #13176, Cell Signaling Technology), rabbit anti-Tyrosine Hydroxylase (TH; catalog #13106, Cell Signaling Technology), mouse anti-Nestin (catalog #ab6142, Abcam), rabbit anti-OLIG2 (catalog #AB9610, Millipore), and rat anti-PDGFRα (catalog #558774, BD Biosciences). After washing in PBS, sections were incubated with secondary antibodies conjugated to Cy2, Cy3, or Cy5 (1:500; Jackson ImmunoResearch) and DAPI at room temperature for 2 h, washed in PBS, and mounted with Fluoromount-G (SouthernBiotech). For BrdU immunostaining, after normal staining for other required antibodies, sections were subjected to DNA denaturation with 2N hydrochloric acid at 37°C for 40 min, then neutralized with 0.1 m sodium borate, pH 8.5, for 2 × 8 min and normal immunostaining protocol with anti-BrdU antibody. For Nissl staining, paraffin section slides were deparaffinized, rehydrated, and subjected to cresyl violet. All immunofluorescence-labeled images were acquired using a Nikon C2+ confocal microscope. Nissl staining images were acquired by Olympus BX53 Bright Field microscope.
The brains were sectioned at 20 μm by taking 1 μm serial optical thin sections for the quantification of labeled cells. Somata were defined as the counting unit. Every six sections were spaced 120 μm apart throughout the entire ventricular zone (VZ) and SVZ, and the hippocampus extent was used to assess the number of labeled cells and avoid any potential of double counting or tissue processing artifact on consecutive sections and ensure that the area counted for each brain is constant. A sequence of at least eight sections per animal (at least three animals per group) were obtained for 3D reconstruction with Imaris image software (Bitplane) to estimate the number of stained cells representing the entire volume of the brain tissue. Counts were also normalized to the length of VZ, SGZ, or cortical strip from the dorsolateral cortex as indicated in the figures. The cell counts were quantified by personnel blinded to the experimental condition of each sample. The quantification method is similar to those generally used for neurogenesis analysis (Feng et al., 2013; Insolera et al., 2014).
Behavior tests
For all behavior tests, mice we used were group housed (three or four animals per cage) in a temperature-controlled room with a 12 h light/dark cycle. Both male and female mice, 8 weeks old, were weaned in mixed genotype home cages, and experimenters blinded to genotype during testing and analysis were used according to a previous study (Durak et al., 2016). No previous analyses were performed on animals used for behavioral testing. All tests were conducted during the light cycle. Each apparatus was cleaned with dilute sodium hypochlorite solution before testing of each animal to prevent bias because of olfactory cues. Open field test was performed first as it is less stressful followed by the three chamber tests.
Open field test
Spontaneous exploratory and anxiety behavior was assessed following previously described procedures (Gould et al., 2009). Each animal was placed in the corner of an open-field apparatus (40 × 40 × 60 cm; Shanghai-Jiliang Software Technology), Total distance traveled and time spent outside of the central area (25% of total area) were automatically recorded over 15 min. The testing room was illuminated at 100 lux.
Three-chamber test
The three-chamber tests were performed in a black acrylic box with three chambers (20 × 30 cm). All mice were isolated for 3 d before the test. On the test day, the subject mice were first habituated in the center chamber for 10 min. In the sociability test, a novel (C57/BL strain) mouse (same sex as the subject mouse) was placed in a small metal container located in the center of the left-side chamber, and the object was placed in the other side. The wall between the compartments was removed to allow free access for the subject mouse to explore each of the three chambers. The duration of the time the subject mouse spent in each compartment was monitored. In the social novelty test, the first mouse (novel mouse) remained in one side of the chamber and now became a familiar social stimulus. The subject was tested in all three chambers (60 × 30 cm) for 10 min for each experiment. Time in each chamber was recorded automatically by EthoVision XT 14 software.
5-Iodo-2′-deoxyuridine/BrdU pulse labeling
Pregnant mice were sequentially injected with 100 mg/kg body weight with 5-Iodo-2′-deoxyuridine (IdU; catalog #I7125, Sigma-Aldrich) and BrdU (catalog #B9285, Sigma-Aldrich) intraperitoneally 2 h and 0.5 h, respectively, before being killed. Immunostaining was performed as described above with the BrdU antibody mouse anti-BrdU (catalog #555627, BD Pharmingen) which can recognize both IdU and BrdU, whereas the rat anti-BrdU antibody (catalog #Ab6326, Abcam) only recognizes BrdU.
Neural progenitor cell isolation and culture
Cortices of (E)14.5 control or Chd8 cKO mouse embryos were dissected in cold PBS. Each culture was derived from an independent embryo. Cells were dissociated with Collagenase IV, followed with FACS sorting for reporter GFP+ cells. All sorted cells grew in NPC maintenance medium composed of DMEM/F-12 (catalog #11320-033, Thermo Fisher Scientific), Neurobasal-A (catalog #10888-022, Thermo Fisher Scientific), B27 (catalog #12587-010, Thermo Fisher Scientific), 20 ng/ml FGF2, and 20 ng/ml EGF at 37°C, 5% CO2. After 3 d, neurospheres were dissociated with StemPro Accutase (catalog #A11105-01, Thermo Fisher Scientific) into single-cell suspensions and plated at 50,000 cells/ml in NPC maintenance medium for neurosphere cultures. The primary sorted cells are considered first passage NPCs.
Assay for ATAC-Seq and analysis
ATAC-seq assays were performed as previously described (Buenrostro et al., 2015). We isolated ∼50,000 GFP+ cells using FACS from the cortices of E14.5 embryos, Chd8 cKO, cHet, and control animals carrying a ccGFP reporter. Cells were lysed in cold lysis buffer, and their nuclei were isolated by spinning down at 500 × g for 10 min at 4°C. Immediately after the nuclei preparation, we performed the transposase reaction for 30 min at 37°C. Using a Qiagen MinElute kit, we purified DNA from cells. After DNA purification, libraries were generated using the Ad1_noMX and Ad2.1–2.4 barcoded primers and were amplified for 11 total cycles by NEBNext PCR master mix. The libraries were purified using a Qiagen MinElute kit to remove contaminating primer dimers. All libraries were then run on the Illumina Sequencer HiSeq 4000.
CHD8 chromatin immunoprecipitation followed by sequencing (ChIP-seq) data were downloaded from the previous study on the DRA database (Katayama et al., 2016). ATAC-seq data together with the downloaded ChIP-seq data files were then mapped to the mm10 genome assembly. Peak calling was performed by Model-Based Analysis of ChIP-seq (MACS; version 1.4.2, http://liulab.dfci.harvard.edu/MACS) with default parameters to get primary binding regions. To ensure that the data were of high quality and reproducibility, we called peaks with enrichment ≥10-fold over control (p ≤ 10−9) and compared the peak sets using the Encyclopedia of DNA Elements (ENCODE) overlap rules. The identified primary regions were further filtered using the following criteria, to define a more stringent protein-DNA interactome: (1) The p value cutoff was set to ≤10−9, and (2) we required an enrichment of six-fold and peak height >5. Two samples of ATAC-seq or ChIP-seq data were combined by normalizing with reads number to generate heat maps. ChIP-seq data of histone marks were obtained from ENCODE database from mouse E14.5 whole brain (H3K4me3, https://www.encodeproject.org/experiments/ENCSR172XOZ/; H3K27ac and H3K27me3, https://www.encodeproject.org/experiments/ENCSR831YAX/). The heat maps of ChIP-seq data were drawn by the heat map tool provided by Cistrome (http://cistrome.org/ap). Visualized peaks figures were obtained from MochiView software, version 1.46, using WIG files generated by MACS or IGV (Integrative Genomics Viewer) software, version 2.3.51, using BIGWIG files. The heat map of ATAC-seq data were drawn by the plotHeatmap tool (https://deeptools.readthedocs.io/en/develop/content/tools/plotHeatmap.html). ToppGene (https://toppgene.cchmc.org/) were used to perform Gene Ontology (GO) analysis and pathway enrichment.
RNA sequencing and data analysis
RNAs from FACS-sorted GFP+ cells of embryonic cortices were extracted by TRIzol (Life Technologies) following the standard protocol, then purified using RNeasy Mini Kit (Qiagen). The RNA quality was determined by Bioanalyzer (Agilent). NEBNext Poly(A) mRNA Magnetic Isolation Module (New England BioLabs) was used with a total of 1 µg of good-quality total RNA to isolate the poly A RNA as input. Libraries were prepared with the NEBNext Ultra II Directional RNA Library Prep Kit (New England BioLabs), which were dUTP-based stranded libraries. The library was indexed and amplified under a PCR cycle number of 8. After library Bioanalyzer QC analysis and quantification, individually indexed and compatible libraries were proportionally pooled and sequenced using NovaSeq 6000 Sequencing System (Illumina). Under the sequencing setting of single read 1 × 50 bp, ∼25 million pass filter reads per sample were generated.
All RNA-Seq data were aligned to mm10 using TopHat with default settings (http://tophat.cbcb.umd.edu/). We used the cuffdiff command to estimate fragments per kilobase of transcript per million mapped reads values for known transcripts and to analyze differentially expressed transcripts. In all differential expression tests, a difference was considered significant if the FDR value was <0.1 (p < 0.01). Both raw and adjusted p values were used for the RNA-seq analysis. For the volcano plot, the upregulated and downregulated genes were represented by red or blue dots, respectively (fold change > 1.5, FDR < 0.1, p < 0.01 between control and Chd8 cKO). R language (http://www.r-project.org) was used for heat map drawing of gene expression. GO analysis and pathway enrichment were generated by ToppGene (https://toppgene.cchmc.org/).
For quantitative Real Time-PCR (qRT-PCR), cDNA was synthesized from 0.5 µg RNA by iScript Reverse Transcription Supermix (Bio-Rad) according to the instructions from the manufacturer. qRT-PCR was performed using the StepOnePlus Real-time PCR System (Applied Biosystems) with quantitative SYBR green PCR mix (catalog #170-8880, Bio-Rad).
Statistical analysis
Statistical data analyses were done using GraphPad Prism 7.0a software (https://www.graphpad.com). All data are shown as mean ± SEM. Data distribution was assumed to be normal, but this was not formally tested. Count data were assumed to be nonparametric, and appropriate statistical tests were used. Statistical significance was determined using two-tailed unpaired Student's t test and one-way or two-way ANOVA as indicated. Significance was set as *p < 0.05, **p < 0.01, and ***p < 0.001, unless otherwise indicated. Quantifications were performed from at least three independent experiments or animals as indicated. No randomization was used to collect all the data, but they were quantified blindly.
Data availability
All the high-throughput sequencing data from RNA-seq and ATAC-seq generated in this study are deposited in the National Center for Biotechnology Information Gene Expression Omnibus under accession number GSE148608.
Results
Chd8 ablation in cortical NPCs leads to cortical malformations
To examine the expression pattern of CHD8 in the developing cortex, we performed immunostaining and showed that CHD8 was expressed in PAX6+ RGCs/NPCs localized in the VZ and TBR2+ IPC populations in the SVZ of the developing cortex at E14.5, as well as DCX+ neural progenitors and NeuN+ neurons in the postnatal cortex (Fig. 1A,B).
To determine whether CHD8 functions in cortical neurogenesis, we generated Chd8 conditional knock-out (cKO) mice by crossing our Chd8fl/fl-Exon-4 strain carrying exon-4 floxed alleles (Zhao et al., 2018) with an Emx1-cre line (Fig. 1C), which drives deletion mainly in cortical progenitors and their progeny starting from E9.5 (Gorski et al., 2002). In Chd8fl/fl;Emx1-cre+/− mice (referred to here as Chd8 cKO), CHD8 expression was depleted in cortical neurons compared with Chd8fl/fl mice (referred to as control mice; Fig. 1D). These Chd8 cKO animals were born in expected Mendelian ratios and were viable throughout adulthood. Although the gross appearance and weight of mice appeared comparable to those of control mice, anatomic examination revealed severe forebrain malformations (Fig. 1E–H). Reduced cortical thickness was detected at both perinatal and adult stages in Chd8 cKO mice (Fig. 1G,H). The phenotype of neocortical development in Chd8 cKO mice is similar to that of another Chd8 mutant mice (Emx1-Cre;Chd8fl/fl-Exon-3) carrying Chd8 exon-3 floxed alleles (Kweon et al., 2021). As Emx1-Cre predominantly targets dorsal cortical progenitors (Gorski et al., 2002), cerebellar and ventral brain regions are not grossly altered in the Chd8 cKO relative to controls (Fig. 1I).
CHD8-deficient mice develop abnormal behaviors
Behavioral abnormalities, such as social deficits and other neuropsychiatric conditions like anxiety, are hallmarks of ASD (Simonoff et al., 2008; Lord et al., 2018; Takumi et al., 2020). To examine the potential abnormalities in social interaction and anxiety-related behaviors directly or indirectly caused by CHD8 deficiency, we performed two behavioral tests, an open field test) and a three-chamber test in control, heterozygous Chd8 cHet (Chd8fl/+;Emx1-cre+/−), and Chd8 cKO mice. Previous studies examining ASD-associated phenotypes in Chd8 mutant mice had used both of these tests to assess anxiety-related behaviors and sociability deficits (Katayama et al., 2016; Jung et al., 2018; Suetterlin et al., 2018; Hurley et al., 2021; Kweon et al., 2021). We first subjected the mice to an open field test, which is widely used to measure anxiety-like behaviors in rodents (Gould et al., 2009; Seibenhener and Wooten, 2015; Kraeuter et al., 2019). Chd8 cKO mice spent significantly less time in the center zone of the open field than control and Chd8 cHet mice while reciprocally spending a longer time in the outer zones (Fig. 2A–C). The movement velocity was also higher in Chd8 cKO mice than control and Chd8 cHet mice (Fig. 2D), indicating hyperactivity in Chd8 cKO mice. Notably, compared with control mice, Chd8 cHet mice exhibited a phenotype similar to Chd8 cKO mice, although to a lesser extent (Fig. 2A–C). These observations suggest that mice with CHD8 loss or dosage reduction in cortical NSCs develop anxiety-like behaviors and hyperactivity.
To further examine autism-related behavioral abnormalities in Chd8 cKO animals, we conducted the three-chamber test, the widely used test for general sociability and social interactions in mice (Kaidanovich-Beilin et al., 2011). In the initial three-chamber sociability analysis, control mice spent more time in the chamber with an unfamiliar mouse than a novel object, whereas Chd8 cHet or Chd8 cKO mice spent less time (Fig. 2E,F), suggesting that Chd8-deficiency in Chd8 cHet and cKO mice impairs social preference. In the subsequent social novelty test, control and Chd8 cHet mice spent more time in the chamber with a novel mouse than a familiar mouse, but Chd8 cKO mice did not display preference for a novel mouse over a familiar mouse (Fig. 2G), suggesting that the Chd8 cKO mice display attenuated preference for social novelty. Decreased exploratory behavior is consonant with increased anxiety, which is in line with the observations from the open field tests. This observation is in contrast to the previous report of Emx1-Cre;Chd8fl/fl-Exon-3 mice showing decreased anxiety and increased preference for social novelty (Kweon et al., 2021), a discrepancy that is discussed later. Nonetheless, our observations that mice with Chd8 ablation in the cortical progenitors exhibit increased anxiety-like behavior and reduced social preference are consistent with characteristic features observed in ASD (Frye, 2018; Hollocks et al., 2019).
Chd8 ablation in cortical NSCs leads to defects in cortical development and lamination
To determine the impact of Chd8 ablation on cortical development, we analyzed the cortices of control and Chd8 cKO mice at different developmental stages. Immunostaining with cortical layer-specific markers, such as CUX1 for the neurons in upper layers II–IV and CTIP2 for the neurons in deep layers V–VI (Molyneaux et al., 2007), showed a decrease in cortical thickness in Chd8 cKO mice compared with controls (Fig. 3A). The upper cortical layers were substantially thinner, but the deep cortical neuronal layers were comparable to those of control mice at postnatal stages (Fig. 3A–C). Lamination in Chd8 cKO mice was also disrupted as expression of the deep layer marker CTIP2 was observed in the upper cortical layers (Fig. 3A,B). No substantial defects were observed in layer-specific marker expression, cortical size, or the hippocampus in Chd8 cHet mice (Fig. 3D–G). The loss of the upper-layer neurons, the population most responsible for callosal projections, also led to agenesis of the corpus callosum in Chd8 cKO mice (Fig. 3H). Thus, Chd8 deletion in the embryonic cortical NSCs leads to cortical malformations by disrupting cortical lamination and upper-layer neurons.
The decreased size of the cortex in Chd8 cKO mice could be because of the defects in neural stem/progenitors. To interrogate this, we analyzed PAX6+ stem-like RGC/NSC populations in the VZ/SVZ. In Chd8 cKO mice, the number of PAX6+ RGCs was lower at E12.5 than in control or Chd8 cHet mice at the same stage (Fig. 4A). Similarly, at the neonatal stage P0, the PAX6+ RGC population in the dorsal SVZ of Chd8 cKO mice was substantially reduced (Fig. 4B), suggesting that CHD8 is required for the expansion of the RGC/NSC population in the developing cortex. Consistently, there was a substantial reduction in the number of M (mitosis) phase cells [marked by phosphorylated H3 (PH3)] in the VZ at E14.5 in Chd8 cKO mice (Fig. 4C), whereas the number of PH3+ cells was comparable between Chd8 cHet and control mice (Fig. 4D), suggesting that proliferation of NSCs is compromised in the VZ of Chd8 cKO animals. Similar results were observed using BrdU incorporation assays with the markers for SOX2+ NSCs (Fig. 4E). To further examine the role of CHD8 in NSC proliferation, GFP+ cells were isolated from Chd8 cKO carrying a Cre-dependent GFP reporter using FACS at E14.5 and subjected to a neurosphere formation assay (Lu et al., 2016). The GFP+ cells from Chd8 cKO animals were defective in neurosphere growth (Fig. 4F,G). After passaging, Chd8-deficient cells formed neurospheres that were significantly smaller than those formed by VZ cells isolated from control mice (Fig. 4G). Together, these observations suggest that CHD8 is critical for NSC proliferation.
Chd8 loss results in precocious cell-cycle exit and defects in intermediate progenitor expansion
Given the defects in NSC proliferation in the VZ of Chd8 cKO animals, we then investigated the effects of Chd8-deficiency on cell-cycle progression by performing sequential injections of IdU and BrdU at E14.5 (Fig. 5A). At the beginning of the experiment, pregnant mice were intraperitoneally injected with IdU to label a fraction of proliferative cells, followed by BrdU injection 1.5 h later to label synthesis (S) phase cells. After an additional 0.5 h, embryos were processed for immunostaining with antibodies specific for BrdU or IdU. In this system, cells labeled with IdU but not BrdU had left the S phase (Liu et al., 2015). We found that the percentage of cells that remained in the S phase (number of BrdU+ cells divided by number of IdU+ cells) was lower in the Chd8 cKO cortices than in controls (Fig. 5B,C); however, the proportion of cells that had exited the S phase (number of IdU+/BrdU-negative cells divided by number of IdU+ cells) was higher in Chd8 cKO embryos (Fig. 5D). In addition, more IdU+/BrdU-negative cells had migrated into the cortical mantle zone in Chd8 cKO mice than control (Fig. 5B), suggesting a precocious cell-cycle exit of Chd8 cKO NPCs.
Consistently, the number of TBR2+ IPCs, which are basal progenitors in the SVZ essential for mammalian neocortical expansion (Lui et al., 2011; Florio and Huttner, 2014), were substantially lower in the Chd8 cKO at E12.5 (Fig. 5E) and at later stages (e.g., E14.5 and E15.5) than control or Chd8 cHet animals (Fig. 5F,G). Moreover, BrdU pulse-labeling assays indicated that the proliferation of IPC cells (BrdU+TBR2+) was substantially reduced in the Chd8 cKO cortex (Fig. 5H,I). Together, these data suggest that precocious cell-cycle exit caused by CHD8 loss contributes to a defect in IPC proliferation and expansion during cortical development.
Temporally induced ablation reveals a stage-specific requirement of CHD8 for NSC and IPC survival
The decreased number of NSCs and IPCs in the VZ and SVZ, respectively, in the Chd8 cKO animals prompted us to analyze apoptotic events in the developing cortex. We immunostained cortices from different developmental stages with a marker for apoptotic cells, cleaved caspase 3 (C-Casp3). We detected massive C-Casp3+ cells in the cortices of Chd8 cKO mice beginning at E11.5, peaking at E12.5 (Fig. 6A). However, we found that the cell death was nearly absent in the Chd8 cKO cortex after E14.5 (Fig. 6A). Next, we sought to determine the progenitor cell types undergoing apoptosis and found that C-Casp3 expression was predominantly in PAX6+ NPCs and TBR2+ IPCs in the VZ and SVZ of the Chd8 cKO cortex at E12.5 (Fig. 6B,C), as well as in TUJ1+ postmitotic neurons to a lesser extent (Fig. 6D). This observation is in contrast to the minimal cell death previously reported in PAX6+ RGCs/NPCs in Emx1-Cre;Chd8fl/fl-Exon-3 embryos (Kweon et al., 2021). However, C-Casp3 was barely detectable in the cortices of control or Chd8 cHet animals at these stages (Fig. 6A). These data suggest that RGC/NPC and IPC cell death occurs predominantly in the Chd8 cKO cortices.
Consistent with a proposed role of CHD8 for p53-mediated apoptosis (Nishiyama et al., 2009), we found that p53 expression was upregulated in the developing cortex of Chd8 cKO mice at E11.5 and increased at E12.5 but was essentially undetectable at E14.5 (Fig. 6E). Similarly, expression of the downstream targets of p53 signaling, such as p21 and Sesn2, was increased at E12.5 in the Chd8 cKO cortices compared with control cortices (Fig. 6F). Activation of the p53 signaling pathway is often induced by DNA damage with double-strand breaks (Williams and Schumacher, 2016). We therefore immunostained the cortices with a DNA damage response marker, γH2AX, and observed higher levels of γH2AX in the cortices of Chd8 cKO mice at E12.5 compared with controls (Fig. 6G), suggesting that CHD8 loss leads to DNA damage, which may in turn cause the progenitor death.
To further determine the specific time window during which CHD8 is required for cortical progenitor population survival, we then used a tamoxifen-inducible Nestin-CreERT2 line (Imayoshi et al., 2006) to generate Chd8fl/fl;Nestin-CreERT2+/− mice (referred to here as Chd8 iKO mice). Chd8 ablation was induced by tamoxifen administration at different embryonic stages (Fig. 6H). Ablation of Chd8 at E10.5 and E11.5 in Chd8 iKO mice, which were harvested at E13.5 and E14.5, respectively, led to extensive apoptosis in the cortex as indicated by C-Casp3 immunostaining (Fig. 6I,J). C-Casp3 and p53 were also upregulated in the same cells in the cortex of Chd8-iKO mice (Fig. 6K). In contrast, Chd8 ablation after E13.5 did not result in detectable C-Casp3 expression in the cortical region (Fig. 6I,J), suggesting a stage-specific impact of CHD8 loss on neural progenitor cell survival.
Given that TBR2+ IPCs are generated and expanded predominantly between E11.5 and E13.5 during cortical development (Sessa et al., 2008; Vasistha et al., 2015), we then examined whether Chd8 ablation results in IPC cell death at different embryonic stages. Strikingly, we observed a substantial increase of apoptosis in a population of TBR2+ IPCs at E14.5 when CHD8 is ablated at E11.5 but not at other stages (Fig. 6L,M), suggesting that CHD8 is critical for the survival of the intermediate progenitor IPCs. Because the generation of neurons in the upper cortical layers stems predominantly through IPCs (Sessa et al., 2008; Kriegstein and Alvarez-Buylla, 2009), the cell death of IPCs caused by CHD8 loss during embryonic neurogenesis likely contributes to the reduction of CUX1+ upper-layer cortical neurons in Chd8 cKO animals. Together, our data reveal a stage-specific role of CHD8 for the survival of NSCs and IPCs during cortical neurogenesis.
p53-dependent and p53-independent roles of CHD8 in cortical progenitor survival and development
Given that upregulation of p53 in developing Chd8 cKO cortices could trigger cell death, we next asked whether inhibition of p53 could block apoptosis and restore neurogenesis during cortical development. We crossed Chd8fl/fl mice with Trp53fl/fl and Emx1-Cre mice to ablate both Chd8 and p53 in Emx1-expressing cortical NSCs [referred to as Chd8; p53 double KO (dKO) mice]. In cortices of Chd8 cKO animals heterozygous for p53 knockout (referred to as Chd8 cKO; p53 cHet), numbers of C-Casp3+ cells were reduced compared with Chd8 cKO animals (Fig. 7A,B). This is consistent with the observation in Sox1-Cre;Chd8 cKO-p53 heterozygous embryos (Hurley et al., 2021). Moreover, we found that C-Casp3 signals were essentially absent in cortices of Chd8;p53 dKO animals (Fig. 7A,B). This suggests that p53 ablation rescues, if not fully, the apoptotic phenotype in the Chd8-deficient cortex during early development. Next, we sought to address whether the proliferative capacity of NSCs was restored by p53 ablation. We found that the number of PH3+ proliferative cortical NSCs was higher in Chd8;p53 dKO and Chd8 cKO;p53cHet mice compared with Chd8 cKO mice at E14.5 (Fig. 7C,D), but lower than that of controls. This suggests that the rescue of the apoptotic phenotype does not fully restore the NSC proliferation defects in the cortex of Chd8 cKO mice. Similarly, the number of TBR2+ IPCs in Chd8;p53 dKO mice was also lower compared with the control, despite an increase compared with that of Chd8 cKO or Chd8 cKO;p53 cHet mice (Fig. 7E,F). These observations suggest that blocking apoptosis caused by p53 downregulation does not fully rescue the defects in NSC proliferation and IPC production in Chd8-deficient embryos, indicative of a p53-independent role for CHD8 in cortical progenitor cell development.
CHD8 targets neurogenesis-associated genes to control cortical development
To determine the potential mechanisms underlying CHD8 function in neurogenesis, we performed transcriptome profiling of GFP reporter+ cells that FACS isolated from control and Chd8 cKO cortices at E14.5. There were ∼480 downregulated and ∼204 upregulated genes in Chd8 cKO compared with control cells (FDR < 0.1, p < 0.01; Fig. 8A). Gene ontology analysis of the differentially expressed genes revealed that downregulated genes were involved in processes related to neurogenesis, neuron differentiation, neural precursor cells proliferation, and axon development (Fig. 8A–C), which may contribute to the cortical dysgenesis phenotype observed in Chd8 cKO mice. In contrast, the upregulated genes were predominantly associated with apoptotic and p53 pathways (Fig. 8B,C).
To determine whether CHD8 directly regulates expression of these differentially expressed genes, we integrated our transcriptomic data with CHD8 binding sites in the embryonic cortex identified by ChIP-seq (Katayama et al., 2016) along with H3K27ac and H3K4me3 associated with enhancers and promotors, respectively (Platt et al., 2017; Zhao et al., 2018). CHD8 binding sites overlapped considerably with those of the activating histone marks H3K4me3 and H3K27ac, but not the repressive histone mark H3K27me3 (Fig. 8D), suggesting that CHD8 is associated with active promotor or enhancer elements to activate target gene expression. Among the differentially regulated genes in Chd8 cKO cortices, ∼435 gene loci had CHD8 binding sites in their promoter/enhancer regions (Fig. 8E). Gene enrichment analysis indicated that CHD8 directly targets the gene loci related to neurogenesis and NPC proliferation among the downregulated genes in Chd8 cKO cortices (Fig. 8F,G). In particular, CHD8 targeted the regulatory elements of a set of neurogenesis-promoting genes including Tbr2/Eomes, which encodes a key regulator of IPC-mediated indirect neurogenesis, and Emx2, Lhx2, Neurg2, Neurod1, and Nefl, as well as those with NPC proliferation, for example, Nes, Lef1, Ltbp3, and Crb2 (Fig. 8F,G). These data suggest that CHD8 may directly target and activate expression of NSC- and neurogenesis-associated genes including IPC-neurogenic regulators for cortical neurogenesis.
CHD8 is critical for maintaining chromatin accessibility landscape for cortical neurogenesis
CHD8 is an ATP-dependent chromatin remodeler, and thus we hypothesized that CHD8 regulates transcription of neurogenesis-related genes by controlling chromatin accessibility. To evaluate chromatin accessibility, we used ATAC-seq (Buenrostro et al., 2015) to compare open chromatin regions in the GFP+ cells isolated from Chd8 cKO and control cortical tissues at E14.5. We detected an extensive alteration of ATAC-seq signals in Chd8-ablated cells. The number of accessible open chromatin sites appeared to be substantially lower in Chd8-ablated cells compared with control cells (Fig. 8H,I), whereas there was a modest decrease in the ATAC-seq peak signals in Chd8 cHet NPCs (Fig. 8I).
Pathway enrichment analysis indicated that the gene loci with decreased accessibility were enriched in neural differentiation, cell cycle/mitosis, DNA repair, WNT signaling, and Hedgehog signaling pathways (Fig. 8J). Consistent with transcriptome profiling, chromatin accessibility was decreased in critical neurogenesis-promoting regulatory genes including Neurod1, Tbr2/Eomes, and Fezf2, as well as neural progenitor proliferation-related genes including Nes, Smarcd3, and Fzd9 (Fig. 8K), the expression of which was also downregulated in Chd8 cKO cortices (Fig. 8C). In addition, qRT-PCR analysis validated the downregulation in the expression levels of these genes in Chd8 cKO cortices compared with controls (Fig. 8L). These results suggest that CHD8 is required for maintaining the proper chromatin landscape of key regulatory genes important for cortical progenitor proliferation and differentiation.
CHD8 loss impedes neurogenesis from NSCs in the adult SVZ
Although ASD is generally thought of as a neurodevelopmental disease, adults can present with ASD symptoms (Bernier et al., 2014; Courchesne et al., 2019). CHD8 remains expressed in the NPCs in adult neurogenic regions in the SVZ and SGZ as indicated by colabeling of CHD8 with the markers for adult neural stem/progenitor cells, Nestin and SOX2 (Fig. 9A,B). Notably, adult Chd8 cKO mice exhibited forebrain and hippocampal dysgenesis (Figs. 1G, 9C), suggesting a potential role for CHD8 in adult neurogenesis. In adult mice, newly formed neurons are generated from transit-amplifying IPCs in the SVZ and then migrate into the olfactory bulb through the rostral migratory stream (RMS; Ming and Song, 2011). To investigate the effect of CHD8 loss on adult neurogenesis, we treated adult Chd8 iKO mice carrying the inducible NSC-expressing Nestin-CreERT2 line TAM for 5 d to ablate Chd8 and labeled dividing NSCs with BrdU 4 weeks post-TAM, in which BrdU will also label newborn neurons from NSCs subsequently (Lagace et al., 2007; Fig. 9D). Chd8 ablation was confirmed in the GFP reporter+ cells by immunohistochemistry (Fig. 9E). Newborn neurons labeled with BrdU were reduced in the olfactory bulbs of Chd8 iKO mice compared with control mice (Nestin-CreERT2; Fig. 9D). Colabeling with the mature neuronal marker NeuN revealed a significant reduction in the newborn neuronal population in Chd8 iKO mice (Fig. 9F,G). Moreover, the number of DCX+ neuroblasts was lower in the olfactory bulbs and the SVZ of Chd8 iKO mice than controls (Fig. 9H–J). Strikingly, TBR2+ IPCs in the RMS region in the forebrain were substantially reduced in Chd8 iKO mice (Fig. 9K,L), suggesting that depletion of CHD8 impedes IPC production during adult neurogenesis.
In contrast to the developing cortex at early embryonic stages, we did not detect any significant alteration in the number of proliferative cells assayed by Ki67 expression or BrdU incorporation, as well as cell death assayed by activated Caspase-3 or p53 expression in the SVZ of adult Chd8 iKO brains (Fig. 10A–D). These observations suggest that CHD8 deletion does not substantially affect adult NSC proliferation and survival in the adult brain but impairs the production of DCX+ IPCs or neuroblasts.
The presence of reduced newborn neurons in the Chd8 iKO olfactory bulb suggests that the neurogenesis deficit is likely not because of migration defects but instead because of a defect of differentiation from NSCs in the SVZ. Adult NSCs from the SVZ can differentiate into different types of neurons in the olfactory bulb including CB+ and TH + cells (Merkle et al., 2007). Consistently, we found that numbers of both newly generated CB+ and TH+ interneurons labeled by BrdU and NeuN decreased in the olfactory bulb of Chd8 iKO mice compared with controls (Fig. 10E–H). These data indicate that Chd8 ablation results in a defect in interneuron differentiation during adult neurogenesis.
CHD8 loss-induced adult hippocampal neurogenesis defects ameliorated by proneurogenic fluoxetine treatment
As adult neurogenesis also occurs in the hippocampal SGZ, to investigate the effect of Chd8 deletion on adult hippocampal neurogenesis, we induced Chd8 deletion by administrating 8-week-old mice with tamoxifen. Four weeks after tamoxifen treatment, we labeled NSCs in the SGZ with BrdU injections for 5 d and harvested the brains of control and Chd8 iKO mice at 2 d postinjection for assessing SGZ progenitor cell proliferation and 6 weeks postinjection for analyzing newly formed neurons, respectively (Battiste et al., 2007; Lagace et al., 2007; Fig. 11A). The number of BrdU+ proliferative cells in the SGZ was comparable between control and Chd8 iKO 2 d post-BrdU injection (Fig. 11B). However, at 6 weeks postinjection, the number of NeuN+ newly formed mature neurons labeled with BrdU was reduced in Chd8 iKO mice compared with controls (Fig. 11C), suggesting that Chd8 ablation impairs neurogenesis in the SGZ region. As in the olfactory bulb region, we also observed a decrease in the number of DCX+ newborn neurons in the SGZ (Fig. 11D,E). Similar to the SVZ region, we did not detect any significant alteration in NSC proliferation (marked by Ki67) in the SGZ region between control and Chd8 iKO mice (Fig. 11F). These results suggest that the observed deficits in adult neurogenesis was not because of a decrease in the proliferation of NSCs in the SGZ as in the SVZ.
Given the defect in adult neurogenesis from NSCs in the absence of CHD8, we then tested whether a neurogenesis-promoting agent, the FDA-approved antidepressant FLX, could enhance NSC differentiation in Chd8 iKO mice. Fluoxetine has been shown to promote adult neurogenesis and signaling pathways including WNT signaling (Santarelli et al., 2003; Ohira et al., 2013), which is a principal regulator of adult hippocampal neurogenesis (Lie et al., 2005) and downregulated in Chd8 cKO cortices (Fig. 8). We then treated Chd8 iKO and control mice with fluoxetine for 21 d and performed BrdU labeling to trace neurogenesis in the hippocampus (Fig. 11G). As expected, fluoxetine treatment increased neurogenesis in control mice (Fig. 11H). Strikingly, fluoxetine treatment increased BrdU-labeled newborn neurons in Chd8 iKO adult mice compared with vehicle treatment (Fig. 11I,J). Moreover, DCX+ neuroblasts also increased following fluoxetine treatment of Chd8 iKO mice (Fig. 11J,K), despite at a lower number compared with that of control fluoxetine-treated group. These results indicate that fluoxetine treatment partially rescues the deficits in neurogenesis caused by loss of CHD8 in NSCs of adult mice, suggesting that enhancing NSC differentiation by fluoxetine, at least in part, restores the adult hippocampal neurogenesis defect in Chd8 iKO mice.
Discussion
In this study, we use developmentally regulated ablation approaches to resolve CHD8 functions at different stages of neurogenesis. Our data demonstrate that CHD8 is not only required for NSC proliferation, survival, and proper cell cycle exit in the VZ but also, strikingly, for IPC production and survival in the SVZ at a critical period when IPC progenitors emerge in the embryonic cortex. The temporally specific analyses indicate that CHD8 is crucial for the transition from direct neurogenesis in the VZ to indirect neurogenesis in the SVZ. Thus, our studies identify a critical role of CHD8 for IPC-mediated indirect neurogenesis through controlling IPC production, survival, and differentiation into upper-layer neurons, in addition to direct neurogenesis at the early embryonic stages (Fig. 8). Given that the upper-layer excitatory cortical neurons are preferentially impaired in autistic brains (Velmeshev et al., 2019), the discovery of an important role of CHD8 in IPC generation and upper-layer neuron production is relevant to human ASD phenotypes.
Consistent with CHD8 inhibition of p53 transactivation (Nishiyama et al., 2009), we detect massive apoptosis for PAX6-positive NPCs in the VZ along with upregulation of DNA damage marker γH2A.X and p53 in Chd8 cKO cortices at early embryonic stages but to a less extent for TUJ1+ neurons. This is in contrast to the predominantly TUJ1+ apoptotic cells reported in Emx1-Cre;Chd8fl/fl-Exon3 mice (Kweon et al., 2021). This phenotypic discrepancy might be because of the nature of the mutations disrupting specific Chd8 isoforms (exon-4 vs exon-3; Kunkel et al., 2018; Lee et al., 2020). Importantly, we find that Chd8 ablation causes NPC death at the early embryonic stages (e.g., E11.5), but not at the late embryonic stages after E13.5. In adults, Chd8-inducible ablation in NSCs also does not lead to apoptosis. Thus, our data indicate that CHD8 regulates p53 expression and NPC survival in a stage-dependent manner, suggesting that CHD8 does not function as a general inhibitor of p53 transactivation and apoptosis as previously proposed (Nishiyama et al., 2009) but rather maintains genomic integrity and protects against apoptosis on DNA damage at the early embryonic stages, which is particularly detrimental to the rapidly dividing neural progenitors (Allard et al., 2004; Sokpor et al., 2018).
In the hippocampus of Chd8 cKO mice, CA3 and CA1 regions appear to be completely disrupted; the dentate gyrus can be detected, however, its size is much smaller compared with controls, implicating an impairment of adult neurogenesis. CHD8 depletion does not impair the proliferation and survival of adult NSCs, unlike those defects observed in the early embryonic cortex of Chd8 mutants, yet it impedes the differentiation of DCX+ newborn neurons from adult NSCs, leading to the decline in adult neurogenesis (Fig. 12). The stark contrast between early embryonic neurogenesis and later embryonic and adult neurogenesis might be because of the intrinsic differences in embryonic and adult neural progenitors, and extrinsic differences of NSC niches, or both. Similar phenotypes have been reported for another CHD family member, CHD7, which is required for embryonic neural differentiation but dispensable for NSC self-renewal during adult neurogenesis (Feng et al., 2013). CHD7 or other family members might compensate for CHD8 loss at the later stages, as CHD7 and CHD8 can work together to regulate NSC survival and differentiation (Feng et al., 2013; Marie et al., 2018). Nonetheless, the defect in IPC and newborn neuron production in the Chd8-ablated adult brain suggests a previously unrecognized critical role of CHD8 for adult neurogenesis and neuronal lineage transition from adult NSC niches.
Although macrocephaly has been associated with CHD8 mutations in patients and systemic haploinsufficiency in Chd8 heterozygous (Chd8+/−) mutant mice (Katayama et al., 2016; Gompers et al., 2017; Suetterlin et al., 2018; Sood et al., 2020; Hurley et al., 2021), we did not observe a macrocephaly phenotype in the NSC-specific Chd8 cHet and cKO mice, consistent with another Emx1-Cre;Chd8 mutant mice (Kweon et al., 2021). These differences may result because of neural-progenitor-specific versus systemic CHD8 haploinsufficiency. In our NSC-specific Chd8 cHet animals, Chd8 dose is reduced predominantly in the neural progenitors of the cortex, whereas in Chd8 heterozygous or hypomorph animals, CHD8 dose decreases systematically in all brain cells, which may lead to changes in neuropil compositions. In addition, the phenotypic variation between the haploinsufficient human CHD8 mutants and NSC-specific Chd8 cHet mice might be because of a differential sensitivity to gene dosage across species, genetic backgrounds, or both. It is worth noting that patient phenotypes with CHD8 mutations are heterogeneous (Bernier et al., 2014; An and Claudianos, 2016; Courchesne et al., 2019). A cohort of patients with disruptive CHD8 mutations exhibited developmental defects or delay (Bernier et al., 2014; Zhao et al., 2018), which might be potentially related to the phenotype observed in the mice with homozygous CHD8 loss in NPCs.
Initial studies used constitutive Chd8 heterozygous mice or knockdown models to examine the functions of CHD8 in ASD-related phenotypes (Durak et al., 2016; Katayama et al., 2016; Gompers et al., 2017; Platt et al., 2017; Jung et al., 2018). However, the findings from these earlier studies were complicated by stage- and dosage-dependent roles for CHD8 in neural development (Sood et al., 2020; Hurley et al., 2021). Our NPC cell-type- and stage-specific in vivo mutagenesis analyses delineate a conserved and distinct role of CHD8 for direct and indirect neurogenesis at early embryonic stages and adult neurogenesis. Although cortical developmental defects in Chd8 cKO animals are consistent with previous reported Emx1-Cre;Chd8fl/fl-Exon-3 and Sox1-Cre;Chd8 fl/f mice (Hurley et al., 2021; Kweon et al., 2021), we detected phenotypic and behavior differences between our and previously reported Chd8-mutant mice. In heterozygous Chd8 cHet littermates, we did not observe any significant alteration of NSC proliferation and survival, which is in contrast with a reported increase in proliferative NSCs and SVZ progenitors in the Chd8+/− embryonic cortex (Gompers et al., 2017). Moreover, in contrast to an increase in TBR2+ IPCs as reported in hypomorph Chd8neo/neo embryos (Hurley et al., 2021), we observed a decrease in TBR2+ IPCs and an increase on IPC cell death in NPC-specific Chd8 cKO animals. In addition, our Chd8 cKO mice exhibit increased anxiety-like behavior and decreased socialization, characteristics of ASD (Frye, 2018; Hollocks et al., 2019). This is in contrast to the previous report for the decreased anxiety and increased preference of social novelty in Emx1-Cre;Chd8fl/fl-Exon-3 mice (Kweon et al., 2021). This might be because of differences in Chd8 mutations (e.g., exon-4 vs exon-3 deletion), as suggested by the alternative splicing forms of Chd8 (Kunkel et al., 2018; Lee et al., 2020), the extent of defects in neural cell-type formation, or even different experimental conditions. Nonetheless, our behavioral results are consistent with another hypomorph model that reduced CHD8 levels below those of heterozygotes (Hurley et al., 2021). Future work will be required to directly compare these different Chd8 mutant models to determine the specific impact of CHD8 loss on cortical cell-type development and animal behaviors.
Transcriptomics analysis revealed that the genes associated with neurogenesis and neural differentiation were downregulated in Chd8 cKO mice, similar to those from systemic heterozygous Chd8+/− mice (Katayama et al., 2016; Gompers et al., 2017; Platt et al., 2017; Suetterlin et al., 2018) and hypomorph Chd8neo/neo mice (Hurley et al., 2021). However, functional requirements of CHD8 in NPC development are in stark contrast to those during neural differentiation from ESCs, wherein CHD8 inhibits neuronal gene expression and maintains pluripotency (Sood et al., 2020). CHD8 might play a distinct role in specification from ESCs to NPCs versus differentiation from NPCs to neurons. WNT signaling was downregulated in Chd8 cKO mice, consistent with findings from in utero knockdown (Durak et al., 2016); however, it was upregulated in the brain of some Chd8-heterozygous mice at E14.5 (Katayama et al., 2016; Platt et al., 2017). The discrepancy in transcriptomics data across different Chd8 mutants might be because of the differences in cell types, brain regions, and dosage with Chd8 deficiency. Nonetheless, our integrated genomic and transcriptomic analyses reveal that CHD8 can directly target and regulate the chromatin landscape of key regulatory genes associated with NSC development and neuronal differentiation and provide a functional link between the chromatin accessibility landscape and cell-survival- and neurogenesis-promoting programs regulated by CHD8 in neural progenitors.
The FDA-approved antidepressant fluoxetine has been shown to stimulate adult neurogenesis (Santarelli et al., 2003; Kong et al., 2009; Ohira et al., 2013) and attenuates autistic behaviors including obsessive-compulsive behaviors in ASD (Reddihough et al., 2019). We showed that fluoxetine treatment can at least partially overcome adult neurogenesis deficits because of CHD8 loss. Fluoxetine has been shown to regulate multiple signaling pathways including WNT, serotonin, and glucocorticoid signaling for neurogenesis (Santarelli et al., 2003; Ohira et al., 2013; Lee et al., 2016; Cunha et al., 2018), although the exact mechanisms underlying adult neurogenesis remain to be further defined.
Footnotes
This work was supported by Cincinnati Children's Research Foundation to Q.-R.L. K.B. was supported by National Institutes of Health–National Cancer Institute Grant 2T32CA117846-11A1. We thank Dr. Ryo Kageyama for the Nestin-CreERT2 mouse line and Drs. Paul Andreassen, Ed Hurlock, and Rachid Drissi for comments and technical help. C.D. performed the experiments at the Cincinnati Children's Hospital and current affiliation is the Children's Hospital of Fudan University.
The authors declare no competing financial interests.
- Correspondence should be addressed to Q. Richard Lu at richard.lu{at}cchmc.org