Fig 2: Functional annotation of DEGs and their relationship to CHD8-binding. a Venn diagram of DEGs from CHD8 +/− in iPSC-derived NPCs and neurons. b qRT-PCR validation of seven DEGs, using β2-microglobulin (β2M) as a reference gene to calculate relative expression levels. The qPCR was carried out on neurons derived from two control samples and the two CHD8 +/− clones (the same samples used for the RNA-seq). Each was normalized against another control sample. Samples were analyzed in triplicate. The graph shows the fold change for each sample relative to the neutral control, error bars show the +/− standard deviation. The fold changes were highly significant: SMARCA2, p = 9.7E-7; WNT7A, p = 6.1E-9; HMGA2, p = 4.4E-6; TESC, p = 0.003; TCF4, p = 4.4E-5; TGFB3, p = 5.5E-6; DDR2, p = 6.2E-8 (two-tailed Student’s t test). c Percentage of CHD8-binding genes in all protein-coding, upregulated, and downregulated genes in NPCs, ranked by their expression values (FPKMs) in WT NPCs. Asterisks (“*”) mark the groups of genes showing a significant difference in the proportions of genes with CHD8 binding by comparing DEGs with all protein-coding genes (binomial test, two-tailed, p < 0.05). d Representative-enriched GO terms (top) and canonical pathways (bottom) among DEGs, as reported by DAVID and IPA, respectively. The red color in each cell corresponds to the −log10(p value), corrected by Benjamini-Hotchberg method (color scale on the left and only terms with p < 0.05 were shown)

Fig 3: Generation and characterization of CHD8 +/− lines. a Design of CRISPR guide sgRNA sequences targeting the N-terminal end of CHD8. The red marks Ser62 where premature stop codon mutations were uncovered from whole exome-sequencing analysis of ASD individuals [16]. b DNA-sequencing analysis of CHD8 +/− iPSC clones. Knockout alleles were identified by cloning of PCR products and Sanger sequencing. c Western blot analysis of CHD8 +/− NPCs. CHD8-specific antibodies were used to detect CHD8 protein in NPC lysates. A clone with homozygous CHD8 knockout (G2C2) was also analyzed, but this clone could not differentiate into neurons appropriately and thus was not included in our RNA-seq analysis. d Validation of 2-bp (KO1, left) and 10-bp (KO2, right) deletion by RNA-seq reads. A screen shot of RNA-seq reads mapped to the CRISPR targeting regions, with gap showing deletion

Fig 4: Unexplained specificity of CHD8 binding near chromatin, RNA processing, cell cycle, and metabolism promoters across CHD8 ChIP-seq datasets. (A) Heatmap showing correlation across included CHD8 and control ChIP-seq datasets. Legend indicates the correlation between datasets. (B) Heatmap showing enrichment of gene ontology and pathway terms for the top 2000 significant peaks meeting a MACS2 significance cutoff of p < 0.00001 across ChIP-seq datasets using GSEA. Legend indicates absolute value normalized enrichment scores. Data are hierarchically clustered as indicated by the dendrograms. Control datasets for both panels are indicated in white font outlined by a gray box.

Fig 5: Hypoactivity and enhanced mother-seeking/attachment behavior in male but not female Chd8+/S62X juveniles. (A) Juvenile male and female Chd8+/S62X mice (P24) show genotype-dependent decreases in the locomotor activity in the open-field test compared with WT mice, although the center-zone activity is normal, as shown by the total distance moved and time spent in the center region of the open-field arena and supported by two-way ANOVA. Note, however, that male, but not female, Chd8+/S62X juveniles show hypoactivity, in the Mann–Whitney U test, which was performed in addition to two-way ANOVA for the lack of significant genotype-sex interaction; the Mann–Whitney U test results are indicated in blue labels above the bar graphs. (n = 22 mice [male-WT], 15 [male-HT], 13 [female-WT], and 7 [female-HT], **p < 0.001, two-way ANOVA, #p < 0.05, ns, not significant, Mann–Whitney U test). (B) Juvenile male and female Chd8+/S62X mice (P22) show largely normal direct social interaction in the juvenile play test, compared with WT mice, as shown by total interaction time (nose-to-nose, nose-to-tail, following, mounting, and allo-grooming). (n = 12 mouse pairs [male-WT], 11 [male-HT], 8 [female-WT], and 7 [female-HT], two-way ANOVA). (C) Juvenile male and female Chd8+/S62X mice (P26) spend normal amounts of time in closed and open arms in the elevated plus-maze test compared with WT mice. (n = 9 [male-WT], 9 [male-HT], 9 [female-WT], and 9 [female-HT], two-way ANOVA). (D) Juvenile male and female Chd8+/S62X mice (P28) spend normal amounts of time in the light chamber of the light–dark apparatus compared with WT mice. (n = 11 mice [male-WT], 12 [male-HT], 12 [female-WT], and 12 [female-HT], two-way ANOVA). (E) Male juvenile Chd8+/S62X mice (P19) show enhanced mother-seeking/attachment behavior, whereas female juvenile Chd8+/S62X mice do not. In the first phase of the maternal-homing test, male and female juvenile Chd8+/S62X mice separated from their mothers in the maternal homing test spend comparable amounts of time sniffing familiar home-cage nest, or fresh, material, as compared with WT mice. In the second phase, male, but not female, Chd8+/S62X mice spend more time with the reunited mother and less time with the home-cage nest material, as compared with WT mice. (n = 22 mice [male-WT], 17 [male-HT], 18 [female-WT], and 14 [female-HT] for phase 1, and 20 [male-WT], 17 [male-HT], 15 [female-WT], and 13 [female-HT] for phase 2,**p < 0.01, ***p < 0.001, ns, not significant, two-way ANOVA with Holms-Sidak test [separate ANOVA for males or females], *p < 0.05, two-way ANOVA [preference index], ###p < 0.001, ns, not significant, Mann–Whitney U test [within-sex comparisons of preference index]).