Chd8^+/S62X mice have been recently reported (Lee et al., 2022). Mice were maintained at the Korea Advanced Institute of Science and Technology (KAIST) mouse facility (12-h light–dark cycle).

Electrophysiology was performed using mouse brain slices containing the dorsal hippocampus, a brain region associated with ASD (Schumann et al., 2004), from juvenile mice at P21–28 and adult mice at 9–18 weeks. Isoflurane-anesthetized mice were used to dissect brains, which were placed in a chamber containing dissection buffer (In mM: 212 sucrose, 25 NaHCO₃, 1.25 NaH₂PO₄, 5 KCl, 3.5 MgCl₂, 0.5 CaCl₂, 10 D-glucose, 2 sodium pyruvate, 1.2 sodium ascorbate, 95% O₂, and 5% CO₂) and used to obtain sagittal slices (300 μm; ~1–2 slices/mice) using a vibratome; exact numbers of mice and neurons used for each experiment are described in the figure legends. Slices containing the dorsal hippocampus were recovered in the artificial cerebrospinal fluid (aCSF; in mM: 125 NaCl, 25 NaHCO₃, 1.25 NaH₂PO₄, 2.5 KCl, 1.3 MgCl₂, 2.5 CaCl₂, 10 D-glucose) at 32°C for 30 and 60 min (juvenile and adult, respectively) and at 20°C–25°C for 30 min, followed by a transfer to a recording chamber at 28°C with circulating aCSF. sEPSCs were measured in the presence of picrotoxin (60 μM; Sigma), and mEPSCs were measured in the presence of picrotoxin (60 μM) and tetrodotoxin (0.5 μM; Tocris). sIPSCs were measured in the presence of NBQX (10 μM; Tocris) and D-AP5 (50 μM; Tocris). mIPSCs were measured in the presence of NBQX (10 μM), D-AP5 (50 μM), and tetrodotoxin (0.5 μM). Recording pipettes (2.0~3.5 MΩ resistance) were pulled from borosilicate capillaries (thin-walled, 30-0065, Harvard Apparatus) using a two-step vertical micropipette puller (PC-10, Narishige). The pipette-filling solution for sEPSC and mEPSC recordings contained (in mM) 117 CsMeSO₄, 8 NaCl, 10 TEACl, 10 EGTA, 10 HEPES, 4 Mg-ATP, 0.3 Na-GTP, and 5 QX-314, and the solution for sIPSC and mIPSC recordings contained (in mM) 115 CsCl, 8 NaCl, 10 TEACl, 10 EGTA, 10 HEPES, 4 Mg-ATP, 0.3 Na-GTP, and 5 QX-314. For giga sealing, neurons were approached with the internal solution-filled pipette and gently ruptured through suction. Membrane voltages were held at −70 mV, and recordings were obtained after cells were perfused with aCSF and stabilized (2 min). Signals were filtered (2 kHz) and digitized (10 kHz) using Multiclamp 700B amplifier and Digidata 1,550 digitizer (Molecular Devices). The access resistance, checked before and after recordings, was used to exclude data when it is >20 MΩ. Acquired data were analyzed using Clampfit 10 (Molecular Devices).

Three mice aged P0, P25, and P80 were used for each group (heterozygous, wildtype, male, female). Brains were quickly dissected and deep-freezed in RNAlater solution (Ambion) to stabilize RNAs. RNA extraction, library preparation, cluster generation, and sequencing were conducted by Macrogen. Sequencing was performed with an average read depth of 70 to 90 million reads at paired-ends (2 × 101 bp) using an Illumina HiSeq 4000 (Illumina) via Macrogen Inc. Transcript abundance was estimated in pseudo-mapping-based mode for the Mus musculus genome (GRCm38) using Salmon (v1.1.0; Patro et al., 2017). Differential gene expression analysis was performed using R/Bioconductor DEseq2 (v1.26.0; Love et al., 2014) by importing the estimated abundance data into R (v.3.5.3) using the tximport (Soneson et al., 2015) package. The p-values were adjusted for multiple testing with the Benjamini–Hochberg correction. A threshold for differentially expressed genes was used as a gene with an adjusted p-value of less than 0.05. We did not attempt RT-qPCR validation of RNA-Seq results considering that RNA-Seq results are usually well correlated with RT-qPCR results and that our study extracts most of the biological functions through GSEA (not DEG analysis), which uses a large number of genes with small changes (Subramanian et al., 2005) that are difficult to validate by RT-qPCR.

Fold changes and adjusted p-values from differential gene expression (DEG) analysis were used to perform Gene Set Enrichment Analysis (GSEA; Subramanian et al., 2005). GSEA allows us to capture if the expressions of genes in a specific gene set are changed in a consistent direction, although each might not be significant enough to be counted as differentially expressed genes. The GSEA software (gsea2-2.2.4.jar; http://software.broadinstitute.org/gsea) was used to obtain enrichment scores. All expressed genes were ranked by the sign of fold change multiplied by –log10 (p-value). 1,000 permutations and a classic scoring scheme were used, as recommended by the program. Gene sets used in this study included those from the Molecular Signature Database (MSigDB v7.0; http://software.broadinstitute.org/gsea/msigdb) and ASD-related gene sets from previous studies (gene sets that are either upregulated or downregulated in ASD and ASD-risk gene sets; see the Results section below for details; Voineagu et al., 2011; Iossifov et al., 2014; Werling et al., 2016; Yang et al., 2018). GSEA results were visualized using EnrichmentMap App in Cytoscape 3.8.0. The RNA-Seq results are deposited in the NCBI GEO (National Center for Biotechnology Information, Gene Expression Omnibus) repository as GSE167053.

To determine whether the similarly altered behaviors in adult Chd8^+/S62X males and females (Lee et al., 2022) accompany synaptic dysfunctions, we measured synaptic transmissions in the adult hippocampus, a brain region associated with ASD (Schumann et al., 2004).

Adult (13–18 weeks) male and female Chd8^+/S62X mice showed significant increases in miniature excitatory postsynaptic currents (mEPSCs) in CA1 hippocampal neurons compared with WT mice, as supported by genotype differences in mEPSC frequency and amplitude in two-way ANOVA analysis (Figure 1A; Supplementary Table S1). However, an additional Mann–Whitney U-test performed, based on the insignificant sex x genotype interaction in the two-way ANOVA, indicated a genotype difference in females but not in males in the frequency and no genotype difference in males or females in the amplitude. In contrast to these excitatory synaptic changes, there were no alterations in miniature inhibitory postsynaptic currents (mIPSCs) in Chd8^+/S62X males or females compared with WT mice, although females showed greater mIPSC frequency and amplitude compared with males (Figure 1B).

The excitatory synaptic differences, however, became insignificant when spontaneous excitatory postsynaptic currents (sEPSCs) were recorded by omitting the action-potential blocker tetrodotoxin in recording solutions to allow network activity (Figure 1C). Spontaneous inhibitory postsynaptic currents (sIPSCs) were also not different between genotypes (Figure 1D).

These results collectively suggest that the CHD8-S62X mutation enhances excitatory, but not inhibitory, synaptic transmission in adult male and female mutant mice, and that network activity compensates for the mutation-induced excitatory synaptic changes.

Given the stronger mother-seeking/attachment behavior in juvenile male Chd8^+/S62X mice (Lee et al., 2022), we next tested whether juvenile (postnatal day [P] 22–28) Chd8^+/S62X males and females show any differential changes in hippocampal mEPSCs or mIPSCs.

Intriguingly, juvenile male and female Chd8^+/S62X mice showed opposite changes in mEPSCs, but not mIPSCs, compared with WT mice, with the frequency of mEPSCs decreasing in males but increasing in females; the amplitudes of mEPSCs were unaffected in either sex (Figures 2A,B). In addition, allowing network activity during recordings by omitting tetrodotoxin did not affect the oppositely changed mEPSC frequencies observed in juvenile male and female Chd8^+/S62X mice (Figures 2C,D), in contrast to the abovementioned compensatory effects of network activity on mEPSCs in adult Chd8^+/S62X mice. However, a moderate decrease in sIPSC amplitude (not frequency) was observed in male but not female Chd8^+/S62X mice in the presence of network activity (Figure 2D).

Therefore, the CHD8-S62X mutation results in the opposite, or sexually dimorphic, changes in excitatory, but not inhibitory, synaptic transmissions that are resistant to network correction in the juvenile mice, which contrasts with the sensitivity of the excitatory synaptic change to network correction in adult mutant mice.

The results recently reported (Lee et al., 2022) and mentioned above indicate largely similar synaptic and behavioral phenotypes in adult male and female mutant mice but sexually dimorphic synaptic changes and male-preponderant behavioral deficits in juvenile mutant males and females. To better understand the mechanisms underlying these results, we performed a longitudinal analysis of transcriptomic changes in the whole brain of P0 (newborn), P25 (juvenile), and P80 (adult) Chd8^+/S62X males and females (see Supplementary Table S2 for raw RNA-Seq data).

Analyses of the RNA-Seq results with a focus on differentially expressed genes (DEGs) did not reveal detectable changes in biological functions, likely because of the small numbers of DEGs in P0, P25, and P80 Chd8^+/S62X males and females (Figures 3A–F; see Supplementary Table S3 for the complete list of DEGs). We thus performed a gene set enrichment analysis (GSEA), a method optimized to identify biological functions that depend on a large number of genes with small, but coordinated, changes rather than by a small number of genes with large changes above artificial cutoffs.¹

GSEA results revealed that, at P0, both male and female Chd8^+/S62X mice showed transcriptomic changes similar to those that occur in ASD (hereafter termed ‘ASD-like pattern’). Specifically, P0 male HT/WT transcripts—the complete set of transcripts differentially expressed between heterozygous Chd8^+/S62X mice and WT mice (ranked by fold-change signs x –log10 [p-value])—were positively enriched for ASD-related gene sets (summarized in Supplementary Tables S4, S5) that are upregulated in ASD (DEG_Up_Voineagu and Co_Exp_Up_Voineagu; Voineagu et al., 2011; Werling et al., 2016), although not negatively enriched for ASD-downregulated gene sets (DEG_Down_Voineagu and Co_Exp_Down_Voineagu; Figure 4A).

In addition, P0 male HT/WT transcripts showed negative enrichments for ASD-risk gene sets such as SFARI genes² and FMRP targets (Werling et al., 2016) but not for DeNovoMis (protein-disrupting or missense rare de novo variants; Iossifov et al., 2014; Werling et al., 2016), DeNovoVariants (protein-disrupting rare de novo variants; Iossifov et al., 2014; Werling et al., 2016), or AutismKB (Autism KnowledgeBase; Yang et al., 2018; Figure 4A); these gene sets are thought to be downregulated by ASD-related deletion, non-sense, missense, frame-shift, and/or splice-site mutations. P0 female HT/WT transcripts showed positive and negative enrichments in ASD-related and ASD-risk gene sets. This is similar to the changes observed in P0-male HT/WT transcripts, which indicates comparable transcriptomic changes in P0 mutant males and females.

In contrast, P25 male and female HT/WT transcripts showed distinct enrichment patterns. P25 males showed an “ASD-like” pattern similar to that observed in ASD as well as P0 males and females, whereas P25 females showed a pattern that is largely unrelated to ASD, with minimal enrichments for ASD-related/risk gene sets (Figure 4A; Supplementary Table S6).

Intriguingly, at P80, male HT/WT transcripts showed minimal enrichment for ASD-related/risk gene sets, whereas female HT/WT transcripts displayed a strong ASD-like pattern (Figure 4A; Supplementary Table S7). These results indicate that mutant male transcriptomes show strong ASD-like, monophasic patterns at ~P0 and P25 but not at P80, whereas mutant female transcriptomes show strong ASD-like patterns at ~P0 and P80 (a biphasic pattern), with a weak or no ASD-like pattern at ~P25.

Previous studies have also reported that ASD is associated with cell-type–specific transcriptomic changes, including decreased gene expression in neurons and oligodendrocytes and increased gene expression in astrocytes and microglia (Voineagu et al., 2011; Werling et al., 2016), although alternative splicing of these genes controlled by proteins, including RBFOX1, can provide a distinct layer of regulation (Irimia et al., 2014; Parikshak et al., 2016; Quesnel-Vallieres et al., 2019). Tests of Chd8^+/S62X transcriptomes for the enrichment of cell-type-specific gene sets (Albright and Gonzalez-Scarano, 2004; Cahoy et al., 2008; Kang et al., 2011; Zeisel et al., 2015; Werling et al., 2016; Velmeshev et al., 2019, 2020) showed that both male and female P0 HT/WT transcriptomes were negatively enriched for excitatory neuron-related gene sets but positively enriched for astrocyte-and microglia-related gene sets (Figures 4B,C), consistent with “ASD-like” patterns. However, mixed positive and negative enrichments could also be observed for some gene sets associated with inhibitory neurons and oligodendrocytes, standing in between the ASD-like pattern and an opposite change (hereafter termed “reverse-ASD-like” pattern).

P25 male HT/WT transcripts maintained this tendency toward an ASD-like pattern, with the negative enrichment for inhibitory neurons getting stronger, but P25 female HT/WT transcripts showed a pattern opposite that of the ASD-like pattern (i.e., nearly undetectable neuronal enrichments, upregulated oligodendrocyte genes, and downregulated astrocyte and microglial genes; Figures 4B,C). At P80, male HT/WT transcripts showed enrichment patterns that were partly ASD-like (moderately downregulated neuronal genes and downregulated oligodendrocyte genes) and reverse-ASD (downregulated astrocyte and microglia genes), whereas female HT/WT transcripts showed a strong ASD-like pattern (downregulated neuronal and oligodendrocyte genes and upregulated astrocyte and microglia genes).

These results collectively suggest that male HT/WT transcripts show monophasic ASD-like transcriptomic changes that peak at ~P0 and P25, whereas female HT/WT transcripts show biphasic transcriptomic changes peaking at ~P0 and P80.

We next tested whether Chd8^+/S62X transcriptomes are enriched for specific biological functions by applying GSEA to gene sets in the Gene Ontology (C5) gene-set database (currently containing ~15,700 gene sets; http://www.gsea-msigdb.org/gsea/msigdb).

In the cellular component (CC) domain, P0 male HT/WT transcripts were positively enriched for extracellular matrix-and chromatin-related gene sets, as shown by the top-five gene-set list (Figure 5A; see Supplementary Tables S5–S7 for full GSEA results [GSEA-CC]) and the clusters of enriched gene sets revealed by the EnrichmentMap Cytoscape App (Isserlin et al., 2014; Figure 5B). P0 male HT/WT transcripts were also negatively enriched for synapse-, spliceosome-, ribosome-, mitochondria-, and chromatin remodeling-related gene sets, as shown by top-five and clustered gene sets (Figures 5A,B). P0-female-HT/WT transcripts showed similar positive and negative enrichment patterns.

At P25, male HT/WT transcripts were negatively enriched for synapse-, ribosome-, and mitochondria-related gene sets (Figures 5A,B), a pattern with some similarity to that observed in P0 males. In P25 females, however, HT/WT transcripts were positively enriched for ribosome-and mitochondria-related gene sets (Figures 5A,B), a pattern largely opposite that observed in P25 males.

At P80, male HT/WT transcripts showed minimal enrichment patterns (positive or negative), whereas female HT/WT transcripts showed strong negative enrichments for synapse-related gene sets and positive enrichments for ribosome/mitochondria-related gene sets (Figures 5A,B).

Tests of gene sets in biological process and molecular function domains (C5 database) revealed enrichment patterns that are largely similar to those observed for CC-domain gene sets, although MF-domain gene sets tended to show weaker enrichments (Figure 6A,B; Supplementary Figure S1; Supplementary Tables S5–S7).

These results collectively suggest that synapse genes are downregulated at P0 and P25 in mutant males and at P0 and P80 in mutant females. Notably, downregulations of synapse genes in P25 males and P80 females were strongly correlated with the ASD-like transcriptomic patterns (Voineagu et al., 2011; Werling et al., 2016) observed in these animals.