Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2015 Mar 10:6:6404.
doi: 10.1038/ncomms7404.

The autism-associated chromatin modifier CHD8 regulates other autism risk genes during human neurodevelopment

Justin Cotney  1 , Rebecca A Muhle  2 , Stephan J Sanders  3 , Li Liu  4 , A Jeremy Willsey  3 , Wei Niu  2 , Wenzhong Liu  1 , Lambertus Klei  5 , Jing Lei  4 , Jun Yin  1 , Steven K Reilly  1 , Andrew T Tebbenkamp  6 , Candace Bichsel  6 , Mihovil Pletikos  6 , Nenad Sestan  6 , Kathryn Roeder  7 , Matthew W State  8 , Bernie Devlin  5 , James P Noonan  1
Affiliations

The autism-associated chromatin modifier CHD8 regulates other autism risk genes during human neurodevelopment

Justin Cotney et al. Nat Commun. .

Abstract

Recent studies implicate chromatin modifiers in autism spectrum disorder (ASD) through the identification of recurrent de novo loss of function mutations in affected individuals. ASD risk genes are co-expressed in human midfetal cortex, suggesting that ASD risk genes converge in specific regulatory networks during neurodevelopment. To elucidate such networks, we identify genes targeted by CHD8, a chromodomain helicase strongly associated with ASD, in human midfetal brain, human neural stem cells (hNSCs) and embryonic mouse cortex. CHD8 targets are strongly enriched for other ASD risk genes in both human and mouse neurodevelopment, and converge in ASD-associated co-expression networks in human midfetal cortex. CHD8 knockdown in hNSCs results in dysregulation of ASD risk genes directly targeted by CHD8. Integration of CHD8-binding data into ASD risk models improves detection of risk genes. These results suggest loss of CHD8 contributes to ASD by perturbing an ancient gene regulatory network during human brain development.

PubMed Disclaimer

Figures

Figure 1
Figure 1. CHD8 targets in neurodevelopment are enriched for ASD risk genes.
(a) Reproducible CHD8-binding sites identified in two biological replicates of hNSC and midfetal human brain. The number of reproducible sites in each tissue and the subset identified in both tissues are indicated in each section of the Venn diagram. The number of ASD risk genes from Liu et al. bound by CHD8 in each subset is noted in parentheses. (b) Histogram showing the results of ASD risk gene label permutations (n=10,000, green bars) assessing enrichment of ASD risk genes reported by Liu et al. within targets of CHD8 shared in hNSCs and midfetal human brain. The observed number of ASD risk genes identified is indicated by a vertical red line. (c) List of ASD risk genes identified by Liu et al. with shared CHD8 binding between hNSCs and midfetal human brain (n=47). (d) Representative ChIP-Seq signal tracks for H3K27ac and CHD8 from hNSCs at the high-confidence ASD gene POGZ. CHD8 peak calls from hNSCs and midfetal human brain are indicated by horizontal bars. CHD8 binding is coincident with strong H3K27ac signal surrounding the transcription start site in hNSCs.
Figure 2
Figure 2. CHD8 targets are enriched in an ASD-associated co-expression network.
A gene co-expression network spanning 10–19 post conception weeks (defined as Periods 3–5 in Willsey et al.) was constructed as described, except the set of input genes was further restricted to only include genes exhibiting H3K27ac and/or H3K4me3 promoter marking in hNSCs to match the observed characteristics of CHD8 targets. The resulting network was tested for enrichment of potential ASD genes identified by Willsey et al., and genes with CHD8-binding sites in their promoters. The 20 genes best correlated with each high-confidence ASD gene (‘hcASD gene’) were included in the network provided the correlation value was R≥0.7. The hcASD seed genes are shown as large circles; CHD8 targets are in yellow; and the top 20 genes that are not CHD8 targets are small white circles. The lines (edges) reflect co-expression correlations: positive correlations are in red and negative correlations are in blue.
Figure 3
Figure 3. Conserved CHD8-binding sites are enriched for ASD risk genes.
(a) Reproducible CHD8-binding sites identified in two biological replicates of hNSC, midfetal human brain and embryonic day 17.5 mouse cortex. The number of reproducible sites in each tissue and the subset identified in both tissues are indicated in each section of the Venn diagram. The number of ASD risk genes from Liu et al. bound by CHD8 in each subset is noted in parentheses. (b) Histogram showing the results of ASD risk gene label permutations (n=10,000, green bars) assessing enrichment of ASD risk genes reported by Liu et al. within conserved CHD8 target genes. (c) Selected gene ontology categories reported by DAVID as enriched in the set of genes bound by CHD8 in hNSC, midfetal human brain and mouse cortex. P values were corrected for multiple testing using the Benjamini–Hochberg method.
Figure 4
Figure 4. Depletion of CHD8 in hNSCs significantly affects CHD8 target genes.
(a) Top, Schematic depicting functional domains within CHD8. Sites in CHD8 that are targeted by knockdown shRNA constructs C and G are indicated by vertical grey bars. Bottom, Representative western blot of hNSC protein lysates demonstrating depletion of CHD8 protein levels due to transfection of each shRNA construct compared with a non-targeting transfection control (shCTL). QPCR and western blots were performed for each knockdown experiment. (b) Conserved CHD8 targets are disproportionately affected by CHD8 depletion. For each subset of CHD8 target genes shown, the P value from a Wilcoxon rank test comparing the distribution of differential expression P values in that subset versus active genes not bound by CHD8 in hNSC is plotted on the y axis, and the number of genes in the subset is plotted on the x axis (Supplementary Information). The red curve shows the smoothed (quadratic) spline fit to the data. (c) Residual values for the indicated subsets of CHD8 targets calculated from the fit lines in b. The set of CHD8 targets conserved in mouse holds the greatest fraction of genes showing differential expression by each CHD8 knockdown.
Figure 5
Figure 5. Depletion of CHD8 in hNSCs significantly affects ASD risk genes.
(a) Mean differential expression P values for ASD risk genes from Liu et al. or Willsey et al. bound by CHD8 versus other genes bound by CHD8 but not in the respective ASD risk gene list. The significance of differences between mean differential expression P-values across gene sets was assessed using Wilcoxon rank tests. Note that CHD8 targets in Liu et al. are significantly dysregulated compared with other CHD8 targets in both knockdowns, whereas CHD8 targets in Willsey et al. are significantly dysregulated compared with other targets only in knockdown C. (b) Scatterplot of log2 fold change gene expression values and log2 read counts per million (CPM) for genes strongly dysregulated in hNSCs transfected with shRNA target C, as compared with scrambled control (EdgeR Poisson P value<1.68 × 10−6 and absolute log2 fold change>0.1, and log2(CPM) between 2 and 10). CHD8 is indicated by a red circle. ASD risk genes from Liu et al. are indicated by purple dots.
Figure 6
Figure 6. ASD risk network predicted by DAWN model incorporating CHD8 binding.
The displayed genes meet the false discovery rate (FDR) threshold of 0.05 based on analysis of the DAWN algorithm incorporating scoring of CHD8 binding that is conserved in hNSC and human brain tissues. Red nodes represent genes that are regulated by CHD8. Large nodes depict genes that have at least one de novo loss of function mutation. THSD7A, SPAST and ASH1L are new ASD risk genes discovered only after incorporating CHD8-binding sites information into the DAWN algorithm. Edges connect genes with absolute partial correlation greater than 0 based on gene expression levels in the midfetal PFC-MSC.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Devlin B. & Scherer S. W. Genetic architecture in autism spectrum disorder. Curr. Opin. Genet. Dev. 22, 229–237 (2012) . - PubMed
    1. Jeste S. S. & Geschwind D. H. Disentangling the heterogeneity of autism spectrum disorder through genetic findings. Nat. Rev. Neurol. 10, 74–81 (2014) . - PMC - PubMed
    1. Krystal J. H. & State M. W. Psychiatric disorders: diagnosis to therapy. Cell 157, 201–214 (2014) . - PMC - PubMed
    1. State M. W. & Levitt P. The conundrums of understanding genetic risks for autism spectrum disorders. Nat. Neurosci. 14, 1499–1506 (2011) . - PMC - PubMed
    1. Sebat J. et al.. Strong association of de novo copy number mutations with autism. Science 316, 445–449 (2007) . - PMC - PubMed

Publication types

MeSH terms

Associated data