The chromatin remodeler ADNP regulates neurodevelopmental disorder risk genes and neocortical neurogenesis

Abstract

Although chromatin remodelers are among the most important risk genes associated with neurodevelopmental disorders (NDDs), the roles of these complexes during brain development are in many cases unclear. Here, we focused on the recently discovered ChAHP chromatin remodeling complex. The zinc finger and homeodomain transcription factor ADNP is a core subunit of this complex, and de novo ADNP mutations lead to intellectual disability and autism spectrum disorder. However, germline Adnp knockout mice were previously shown to exhibit early embryonic lethality, obscuring subsequent roles for the ChAHP complex in neurogenesis. To circumvent this early developmental arrest, we generated a conditional Adnp mutant allele. Using single-cell transcriptomics, cut&run-seq, and histological approaches, we show that during neocortical development, Adnp orchestrates the production of late-born, upper-layer neurons through a two-step process. First, Adnp is required to sustain progenitor proliferation specifically during the developmental window for upper-layer cortical neurogenesis. Accordingly, we found that Adnp recruits the ChAHP subunit Chd4 to genes associated with progenitor proliferation. Second, in postmitotic differentiated neurons, we define a network of risk genes linked to NDDs that are regulated by Adnp and Chd4. Taken together, these data demonstrate that ChAHP is critical for driving the expansion of upper-layer cortical neurons and for regulating neuronal gene expression programs, suggesting that these processes may potentially contribute to NDD etiology.

Keywords: Adnp; ChAHP complex; chromatin remodeling; neocortex; neurodevelopmental disorders.

Fig. 1.

Conditional ablation of Adnp and Chd4 during neocortical neurogenesis. (A–E) Adnp immunohistochemistry on wild-type (WT) mouse brains at E11.5 (A), E13.5 (B), E15.5 (C), or adult (D) stages. (E) Conditional genetics strategy. Cre-mediated excision of LoxP-flanked exon 5 leads to the deletion of all but ~60 amino acids from the Adnp coding sequence, including all of the DNA binding domains (green bars). Red color indicates the Emx1-Cre expression domain. (F and G) Adnp immunohistochemistry on Adnp cKO brains at E15.5 (F) or adult (G) stages. (H–J) Adnp, Chd4, Pax6, and DNA (Hoechst) staining on E13.5 Adnp cHet (H), Adnp cKO (I), and Chd4 cKO (J) neocortices. (K and L) Coimmunoprecipitations performed on P3 neocortical or hippocampal protein lysates using the indicated antibodies. (K) Adnp western blot. (L) Chd4 western blot. Ncx: neocortex; Hcx: hippocampus; GE: ganglion eminence; LGE: lateral ganglionic eminence; MGE: medial ganglionic eminence; IFL: inner fiber layer; Se: septum; Str: striatum; VZ: ventricular zone. (Scale bars, 200 μm.)

Fig. 2.

Adnp and Chd4 are required for cortical growth. (A–F) WT (A) Adnp cHet (B) Adnp cKO (C) brains harvested at P28. (D) Quantitation of cortex area at P2, P7, P12, and P28. n-values: P2: 6 cHet, 5 cKO; P7: 9 wt, 7 cHet, 3 cKO; P12: 9 WT, 5 cHet, 4 cKO; P28: 34 wt, 9 cHet, 9 cKO. ** P < 0.01 by Student’s t test. *** P < 0.001, **** P < 0.0001 by one-way ANOVA with Tukey’s post hoc test. (E) Quantitation of cortex area at P28 according to biological sex. **** P < 0.0001 by one-way ANOVA with Tukey’s post hoc test comparing sexes separately. (F–H) Chd4 cHet (F) and Chd4 cKO (G and H) brains harvested at P28. The Chd4 cKO in (H) is hydrocephalic. (I and J) Comparison of Adnp and Chd4 cHet and cKOs by cortical area (I) or brain weight (J) at P28. Red data points indicate hydrocephalic brains. ** P < 0.01, *** P < 0.001, and **** P < 0.0001 by Kruskal–Wallis with Dunn’s post hoc test. (Scale bars, 200 μm.)

Fig. 3.

scRNA-seq atlas of neocortical development in Adnp and Chd4 mutants. (A) UMAP embedding of 28,898 neocortical cells derived from the indicated genotypes and timepoints. (B–D) Individual timepoints as indicated. (E) Cluster identification via the Leiden algorithm. (F) Cell types were annotated according to marker genes identified in a previous study (36). See also SI Appendix, Figs. S3 and S4. (G) Comparison of WT (n = 4) and Adnp cKO (n = 4) replicates at E13.5. (H and I) Adnp transcript expression in E13.5 WT (H) or Adnp cKO (I) replicates. (J) Comparison of Adnp and Chd4 paralog gene expression from all timepoints in APs (Top) or Interneurons (Bottom). Note that interneurons fall outside of the Emx1-Cre lineage, and therefore serve as an internal control. Blue arrows: down-regulated in Adnp cKO; Red arrows: up-regulated in Adnp cKO. AP: apical progenitors; IP: intermediate progenitors; CR: Cajal–Retzius cells.

Fig. 4.

Adnp regulates neurogenic gene expression. (A) Venn diagrams of overlap between differentially expressed genes (adj. P-value <0.05) in Adnp cKO at E13.5, E16.5, and E18.5, or in E16.5 Adnp cHet versus cKO. (B) Volcano plot of DEGs (adj. P-value <0.05; LogFC > 0.4 or < −0.4) from Adnp cKO samples vs. WT control at E13.5. Selected significantly down-regulated genes are depicted in purple, while up-regulated genes are depicted in green. The Xist gene reflects imbalances in biological sex (SI Appendix, Fig. S5). (C and D) Top 10 GO terms (by FDR and LogFC) for Biological Process on genes depicted in (A). (E) Cell-type proportions present in the E13.5 scRNA-seq dataset. * P < 0.05; ** P < 0.01 via two-way Student’s t test.

Fig. 5.

Adnp is required for the production of upper-layer neurons. (A–J) Immunohistochemistry for cell-type markers in the E15.5 embryonic neocortex. (A–D) Immunohistochemistry for the apical progenitor marker Pax6 and the lower-layer marker Bcl11b. (E–J) Immunohistochemistry for Pou3f2. Pou3f2 marks progenitors and differentiating neuronal precursors, as well as upper-layer neurons that have migrated to the cortical plate. This latter population is reduced in the Adnp cKO (G–J; arrows). (K) Measurement of the size of neocortical zones in WT, cHet, and, cKO brains. (L) Quantitation of Bcl11b levels by zone. (M) Quantitation of Pou3f2 levels by zone. ** P < 0.01 via two-way Student’s t test. VZ: ventricular zone; SVZ: subventricular zone; CP: cortical plate. (Scale bars, 200 μm.)

Fig. 6.

Adnp is required for apical progenitor proliferation and self-renewal. EdU was administered 24 h prior to harvest at E13.5 (A–D) or E15.5 (E–N). Upper Left panels in (A, B, E, and F) depict the regions shown at higher magnification in the other panels (dotted squares). Arrows indicate the absence of EdU+/Ki67+ cells within the apical VZ in the Adnp cKO. (C, D, G, and H) Quantitation of the proportion of cells that are EdU+, Ki67+, double-positive, or the fraction of double-positive cells among total EdU+ cells within the ventricular zone (C and G) or subventricular zone/cortical plate (D and H) as indicated. (I–N) At E15.5, EdU was costained with the apical progenitor marker Pax6 (I and J) or the basal progenitor marker Tbr2 (K and L). (M) Quantitation of the proportion of EdU+, Pax6+, or double-positive cells within the VZ and SVZ. (N) Quantitation of the proportion of EdU+, Tbr2+, or double-positive cells within the VZ and SVZ. * P < 0.05; ** P < 0.01; *** P < 0.001 by Student’s t test. VZ: ventricular zone; SVZ: subventricular zone. (Scale bars, 100 μm.)

Fig. 7.

(A) Cut&run-seq comparing the genomic occupancy of Chd4 or Ctcf, with an Adnp dataset previously generated by the John Rubenstein lab (JR) (19)—all from E13.5 forebrain/cortical tissue. Datasets were also compared against ChIP-seq for H3K4me3 (marking active promoters) and H3K27ac (marking promoter/enhancer elements) generated by the Bing Ren lab (BR) from E13.5 forebrain (40). Data are centered on (WT) Chd4 peak summits. (B) Venn diagram of called peaks for Chd4 (blue) or Adnp (red). (C) Comparison of Chd4 occupancy in WT (blue) or Adnp cKO (purple) E13.5 cortices at Adnp peak loci. (D) GO terms analysis for Biological Function on genes co-occupied by Adnp and Chd4. (E and F) UMAP projections of gene expression scores for shared cut&run-seq targets of Adnp and Chd4 in WT and cHet (E) or Adnp cKO (F). (G) Violin plot of gene expression scores for shared cut&run-seq targets of Adnp and Chd4 at E13.5 and E16.5. (H) Venn diagram of overlap between differentially expressed genes in Adnp cKO vs. control (red), or Chd4 cKO vs. control (blue). Shared genes are depicted in purple. Scatterplot of differentially expressed genes comparing Adnp cKO vs. WT (X-axis), or Chd4 cKO vs. cHet control (Y-axis). (I) Regulation of previously identified Chd4 target genes (30) at E16.5. Dot plots depict gene expression in all cells. Blue text/arrows: significantly down-regulated. Red text/arrows: significantly up-regulated. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001 by adjusted P-value (Wilcoxon rank-sum test).

Fig. 8.

Adnp and Chd4 regulate neurodevelopmental disorder risk genes. (A) Venn diagram of mouse orthologs of SFARI ASD risk genes, SysNDD risk genes, E13.5 Adnp cKO DEGs, or E16.5 Chd4 cKO DEGs. (B) Volcano plot of E13.5 Adnp cKO DEGs orthologous to SFARI genes (red), SysNDD genes (blue), or both (purple). (C) Volcano plot of E16.5 Chd4 cKO DEGs orthologous to SFARI genes or SysNDD genes. (D) Violin plot of SFARI gene expression scores in WT cells (E13.5 and E16.5) according to cell type. (E and F) UMAP projections of E16.5 gene expression scores for mouse orthologs of SFARI risk genes in WT and cHet (E) or Adnp cKO (F). Arrows mark upper-layer neurons. (G) Violin plots of gene expression scores for mouse orthologs of SFARI risk genes in all cells, or upper layer neurons as indicated. (H) Violin plot of SysNDD scores in WT cells (E13.5 and E16.5) according to cell type. (I and J) UMAP projections of E16.5 gene expression scores for mouse orthologs of SysNDD risk genes in WT (I) or Adnp cKO (J). (K) Violin plots of gene expression scores for mouse orthologs of SFARI risk genes in all cells, or upper layer neurons as indicated. * P-value < 0.05, ** P-value < 0.01, *** P-value < 0.001 by ANOVA with Tukey’s post hoc test.