Autism-associated CHD8 deficiency impairs axon development and migration of cortical neurons

For light microscopic immunohistochemistry (IHC), human brain sections (30 μm) were incubated in 0.01 M phosphate-buffered saline (PBS) supplemented with 3% hydrogen peroxide for 10 min to block endogenous peroxidase and then in a blocking buffer containing 5% BSA/10% normal goat serum/0.25% Triton X-100 for 60 min at room temperature to prevent nonspecific staining. Following this, IHC was conducted on these free-floating sections and staining was visualized with a standard ABC Elite kit (Vector Labs).

For confocal microscopic double-labeling immunofluorescence, brain sections were incubated in the blocking buffer for 60 min at room temperature and then in a solution containing primary antibodies from different species simultaneously for one night at 4 °C. After washing, sections were incubated with appropriate secondary antibodies from Alexa Fluor series (Invitrogen, USA) for 45 min at 37 °C and then counterstained with Hoechst 33342 (1:5000; #C10022, Beyotime) for 15 min at room temperature to identify cellular nuclei. After mounting the sections, we observed them under a fluorescent confocal microscope (A1R Nikon).

Primary antibodies used were as follows: anti-CHD8 (1: 500; catalog #A301-225A; Bethyl Laboratories, Inc.), anti-GFAP (1:1000, catalog #MAB3402, Millipore), anti-MAP2 (1:2000, catalog #M9942, Sigma), anti-parvalbumin (Pvalb) (1:500, catalog #MAB1572, Millipore), and anti-DCX antibodies (1:200, catalog #Sc-8066, Santa Cruz).

For Nissl staining, Neuro Trace 500/525 green fluorescent Nissl stain (N21480) was bought from Thermo Fisher Scientific and used according to the manufacture’s instruction.

Mice were sacrificed by cervical dislocation at postnatal day 3 (P3) or P7. Their brains were removed and fixed in 4% (wt/vol) paraformaldehyde (PFA) overnight. After sequential dehydration in 15% and 30% sucrose at 4 °C, 50-μm coronal brain sections were cut with a cryostat, fixed in 4% PFA for 20 min at 4 °C, washed three times with PBS, blocked with 5% BSA and 0.3% Triton X-100 in PBS for 1 h at room temperature, and then incubated with rabbit anti-GFP (1:1000, catalog #A11122, Invitrogen) primary antibody overnight at 4 °C. The rinsed sections were then incubated with Alexa 488-conjugated goat anti-rabbit secondary antibody (1:3000; catalog #A-11122; Invitrogen) for 2 h at 37 °C. For endogenous CHD8 and NeuN immunostaining, sections were incubated with anti-CHD8 antibody (1: 500; catalog #A301-225A; Bethyl Laboratories, Inc.) and anti-NeuN antibody (1:500; catalog # MAB377; Millipore).

Cultured neurons were fixed with 4%PFA and then immunostained with primary antibodies against GFAP (1:1000, catalog #MAB3402, Millipore), MAP2 (1:2000, catalog #M9942, Sigma), and Tau1 (1:3000, catalog #MAB3420, Millipore) overnight at 4 °C. After washing, sections were then incubated with Alexa 488-conjugated goat anti-rabbit secondary antibody (1:3000, catalog #A-11122, Invitrogen) and Alexa 568-conjugated goat anti-mouse secondary antibody (1:2000, catalog #A-11004, Invitrogen) for 2 h at 37 °C.

RNA in situ hybridization on the mouse brain sections was performed with digoxigenin (DIG)-labeled RNA probes. Full-length cDNA of Chd8 was amplified with specific PCR primers and cloned into a pGEM-T easy vector (Promega) to generate an antisense probe for Chd8. The DIG-labeled antisense probes were synthesized by in vitro transcription using SP6 RNA polymerase. Mice of different developmental stages were perfused with 4% PFA, and fixed brains were sectioned into 20-μm slices using a cryostat. RNA in situ hybridization was performed as described previously [26]. Brain sections were hybridized for 18 h at 60 °C. The hybridization signal was detected with anti-DIG–alkaline phosphatase Fab fragments (Roche) and nitro blue tetrazolium chloride (NBT) plus 5-bromo-4-chlor-indolyl-phosphate (BCIP) as color reaction substrates.

Human CHD8 cDNA (CHD8, variant 1 - HaloTag® human ORF in pFN21A) was purchased by Promega (catalog: FHC12740). PCR products were cloned into the pCAGGS-IRES-EGFP vector [27]. Oligonucleotides targeting different sequences of mouse Chd8 cDNA and a scrambled control oligonucleotide were designed, synthesized, and cloned into pSuper-basic (Oligoengine). The 21-bp target sequences were as follows: shRNA#1, 5′-GCT GGT GGA CTT GGT ATT AAT-3′; shRNA#2, 5′-GCT GCT GAT ACC TGT ATT ATC-3′; shRNA#3, 5′-GCT ACA AGA GAG AAC AAA TGA-3′; and shRNA-scramble, 5′-TTC TCC GAA CGT GTC ACG T-3′. All constructs were verified by DNA sequencing.

For the analysis of the expression of CHD8 in the developing mouse cortex, cortices were dissected and homogenized as previously described [28]. For knockdown and rescue experiments, electroporated cortical neurons were cultured for 3–7 days and lysed in 1× protein loading buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, 1% 2-ME, and 0.1% bromophenol blue). All protein samples were separated by 10% SDS-polyacrylamide gel electrophoresis (Bio-Rad) and blotted onto PVDF membrane (Millipore). Membranes were blocked with 5% non-fat milk in 0.05% Tween 20 at room temperature for 1 h and probed with rabbit anti-CHD8 (Bethyl Laboratories, Inc.), HRP-coupled mouse anti-actin (Santa Cruz Biotechnology) antibodies. The secondary antibody used for CHD8 was goat anti-rabbit IgG coupled to HRP (Santa Cruz biotechnology). Bands were visualized by enhanced chemiluminescence (TIANGEN). Band intensities were measured with ImageJ Software.

Primary cortical neurons were prepared from brains of P0 C57 mice as described previously [28]. Before plating, neurons suspended in transfection medium (200 μl) were mixed with 1 μg pCAG-EGFP and 3 μg shRNA for immunocytochemistry. Then, neurons were transferred into a 2.0-mm electroporation cuvette (Fisher) and transfected by electroporation using the Amaxa Nucleofector apparatus.

In utero electroporation was performed as described previously with a few modifications [27]. Briefly, pregnant mice at E14.5 were anesthetized with 0.7% sodium pentobarbital at a dose of 10 μl/g and subjected to abdominal incision to expose the uterus. For different experiments, a mixture of GFP, RNAi, and/or rescue constructs was prepared. Plasmids (about 1 μl) with 0.05% Fast Green (Sigma) were injected into the lateral ventricle of embryos through a glass micropipette. Electrical pulses were then delivered to embryos by gently clasping their heads with forcep-shaped electrodes connected to an ECM-830 square-pulse generator (BTX). Five 50-ms pulses of 30 V were applied at 1-s intervals. Uterine horns were repositioned in the abdominal cavity, and the abdominal wall and the skin were sutured. Postsurgery animals were maintained in a warm animal room (25 °C) with plenty of water and food. At different developmental stages, mice were perfused transcardially with 0.9% saline followed by 4% PFA in 0.1 M phosphate buffer (PB; pH 7.4), and the brains were removed and fixed in 4% PFA for another 24 h. Fixed brains were cryoprotected with 30% sucrose and then sectioned using a cryostat.

For all statistical analyses, experimenters were blinded for genotypes of mice and the treatments of the animals or cells. Data are presented as mean ± SEM. The paired or unpaired Student’s t test, one-way ANOVA with Dunnett’s post hoc tests or Bonferroni’s tests, or two-way ANOVA with Bonferroni post hoc tests were used for data analyses by SPSS 13.0 software. Statistical significance was defined as P < 0.05.

Except in utero electroporation, for which ICR mice were used, all other experiments were performed in C57BL/6 J mice. Pregnant mice undergoing in utero electroporation were individually housed in plastic cages, and when they delivered, they were housed with their pups.

Frozen postmortem brain tissues of Brodmann 19 region from a 6-year-old child was obtained from the NIH human brain tissue bank at the University of Maryland (http://​www.​medschool.​umaryland.​edu/​btbank/​). The postmortem time was 19 h.

To probe the function of CHD8 and its pathophysiology underlying ASD, we first examined the expression pattern of CHD8 in a human brain sample. We performed immunohistological analysis using postmortem brain tissue (Brodmann 19) of a 6-year-old child. Strong staining of CHD8 was revealed in cerebral cortex (Fig. 1a). CHD8 was highly expressed in Neuro Trace Nissl stain-positive neurons but to a lesser extent in GFAP-positive glia cells (Fig. 1b, c). Consistent with the known function of CHD8 as a chromatin regulator, immunofluorescent staining showed that CHD8 was predominantly localized in the nucleus (Fig. 1b, f). Co-staining of CHD8 and neuronal markers showed that CHD8 was expressed in both MAP2-positive excitatory neurons (Fig. 1d) and parvalbumin (Pvalb)-positive interneurons (Fig. 1e). Moreover, CHD8 was expressed in Doublecortin (DCX)-positive neuronal precursor cells and immature neurons (Fig. 1f), which is consistent with the early onset of clinical manifestations in patients with CHD8 mutations.

Next, we examined the expression pattern of Chd8 in the mouse brain at different developmental stages. Quantitative real-time mRNA expression analysis (qRT-PCR) (Fig. 2a) revealed that Chd8 was detectable in neocortex from E16 to adulthood. The peak of the expression is at E18 and then markedly decreases at P7 and gradually declines to very low levels at adulthood. This expression pattern in neocortex was also confirmed at the protein level by western blot analysis (Fig. 2b, c). To examine the temporal and spatial expression pattern of Chd8 in the developing brain, we performed RNA in situ hybridization on mouse brain sections. Chd8 mRNA was most abundant in embryonic brain and then gradually reduced after birth, consistent with the results from the qRT-PCR and western blot analyses. Specifically, Chd8 mRNA was present in the sub-plate (SP) and proliferative zones of the cortical wall (the ventricular zone [VZ] and the subventricular zone [SVZ]) at E16 and P0 and in the cortical plate (CP) of the cerebral cortex, hippocampus, striatum, cerebellum, and olfactory bulb from P3 to P7, a developmental period with extensive outgrowth and elaboration of dendrites and axons (Fig. 2d, e). The cellular and subcellular localizations of CHD8 in developing mouse brains and cultured neurons were also examined by immunostaining with neuronal cell markers. Consistent with the results in human cerebral cortex, mouse CHD8 was expressed in the neuronal nucleus, with a pattern similar to that of NeuN in neocortex and hippocampus in P3 mouse (Fig. 3a–c). CHD8 expression in neurons and glia cells was also confirmed by western blot analysis of cultured DIV3 neurons and glia cells of mice (Fig. 3d). However, the expression of CHD8 in glia cells was significantly lower than that in neurons. Taken together, the expression analyses of both human and mouse brains imply important roles for CHD8 during the development of cerebral cortex.

Altered short- and long-distance functional connectivity has been reported frequently in brains of human ASD patients by neuroimaging studies [29, 30]. Impaired dendritic development has been reported in numerous mouse models of ASD [31‐33]. However, few studies have examined axonal development in ASD animal models. We therefore examined whether Chd8 deficiency affects the development of both dendritic morphology and callosal axon growth during development using a shRNA approach. We designed three shRNA constructs that were targeting to Chd8 and tested the efficiency in both cultured neurons and in vivo in mouse brains. Compared with the scramble shRNA construct, all shRNAs resulted in a significant decrease in the expression of CHD8 protein in cultured neurons, and among them, shRNA#1 had the highest efficiency (Fig. 4a). The efficacy of the shRNA constructs was confirmed in vivo in mouse brain (Fig. 4b). We also validated the rescued constructs against shRNA-mediated knockdown of mouse Chd8 in transfected neurons in vitro (Additional file 1: Figure S1). To examine axonal growth, we co-transfected dissociated DIV0 mouse cortical neurons with plasmids encoding GFP and shRNA constructs by electroporation and then plated them to allow the neurons to grow in culture medium for 3 days (DIV3). We used Tau1 as an axon-specific marker. We found that depleting CHD8 significantly decreased total axon growth (Fig. 4c, d). However, co-expression of human CHD8 protein (hCHD8) that is shRNA#2-resistant was able to rescue the total axon length defects caused by shRNA#2 construct (Fig. 4c–e; Additional file 1: Figure S1). We next examined the effects of Chd8 knockdown on dendritic outgrowth and branching in vitro. Control neurons were transfected at DIV0 with GFP plus a non-targeting shRNA construct (shRNA-scr). Seven days later, we analyzed the data after staining cells with the somatodendritic marker MAP2 (Fig. 4f). The total and mean dendrite length were significantly reduced in neurons expressing both shRNA#1 and shRNA#2 constructs compared to those expressing shRNA-scr (Fig. 4g). It is worth noting that the intersections were more affected by shRNA#1 construct than by shRNA#2 construct (Fig. 4h). This appears to be consistent with the more significant reduction of Chd8 after shRNA#1 treatment than that of shRNA#2. The effects of shRNA#2 were rescued by co-transfecting shRNA2-resistant hCHD8 (Fig. 4i, j). These results indicate CHD8 is required for both axonal and dendritic development.

To test whether Chd8 deficiency has effects on mouse brains in vivo, we used the shRNA#2 construct for this experiment because of the feasibility of conducting rescue experiments as demonstrated in vitro. At P3, mouse pups electroporated with scrambled shRNA had callosal axons originating from the electroporated neurons that crossed the midline and formed and arborized dense projections to the contralateral side of cerebral cortex, and the axon bundles were restricted to the corpus callosum (CC) (Fig. 5a–c, j, k). In contrast, in mice transfected with the Chd8 shRNA#2 construct, the axon terminals were barely visible in the contralateral cerebral cortex, indicating significantly retarded axon growth and elongation (Fig. 5d–f, j, k). Co-transfection of the shRNA#2-resistant hCHD8 construct successfully rescued these defects (Fig. 5g–k). These findings suggest that reduced expression of Chd8 in callosal projection neurons impairs axon growth and branching in vivo.

To assess the impact of Chd8 knockdown on the development of neuronal dendrites, we performed morphological analysis of the layer II/III pyramidal neurons of somatosensory cortex 1 (S1). The layer II/III pyramidal neurons of S1 constitute the main excitatory neuronal subtypes in the neocortex and typically having an apical dendrite that branches out in an apical tuft that terminates in layer I [31]. Most scrambled shRNA-transfected neurons displayed a typical polarized dendritic arbor (Fig. 6a, b). In contrast, the Chd8 shRNA significantly decreased the secondary branches and the total number and mean length of dendritic branches in neurons (Fig. 6c, d). The complexity of dendrites was also significantly decreased by Chd8 shRNA constructs, revealed by Sholl analysis (Fig. 6e). These results indicated that CHD8 plays an essential role in neuronal morphogenesis in the mouse cerebral cortex.

Because of the expression of CHD8 in DCX-positive neural precursors, we examined whether its deficiency affected the radial migration of cortical neurons. The shRNA and EGFP constructs were co-electroporated into the developing mouse cortex at E14.5 using in utero electroporation (IUE) as previously described [27], and the distribution of EGFP-labeled cells in neocortex was examined at E18.5. While the majority of the EGFP-positive cells (76.8 ± 0.6%) migrated to the upper cortical plate (CP) when scrambled shRNA was co-expressed, only a small percentage of EGFP cells were observed in the upper CP when the Chd8 shRNA constructs were co-expressed (shRNA#1, 10.1 ± 1.2%; shRNA#2, 14.4 ± 0.9%;) and the majority of labeled cells remained in the VZ/SVZ (Fig. 7a, b). Interestingly, the number of EGFP-positive cells that reached the upper CP was comparable between shRNA construct and scrambled controls at P3 and P7 (Fig. 7c–f). Accordingly, there are a comparable number of neurons that remained in the white matter (WM) and layers IV–VI. Together, these results indicate that deficiency of Chd8 could cause a delay in radial migration of cortical neurons during early embryonic development, but this normalizes during postnatal development.