A set of regulatory genes co-expressed in embryonic human brain is implicated in disrupted speech development

Abstract

Genetic investigations of people with impaired development of spoken language provide windows into key aspects of human biology. Over 15 years after FOXP2 was identified, most speech and language impairments remain unexplained at the molecular level. We sequenced whole genomes of nineteen unrelated individuals diagnosed with childhood apraxia of speech, a rare disorder enriched for causative mutations of large effect. Where DNA was available from unaffected parents, we discovered de novo mutations, implicating genes, including CHD3, SETD1A and WDR5. In other probands, we identified novel loss-of-function variants affecting KAT6A, SETBP1, ZFHX4, TNRC6B and MKL2, regulatory genes with links to neurodevelopment. Several of the new candidates interact with each other or with known speech-related genes. Moreover, they show significant clustering within a single co-expression module of genes highly expressed during early human brain development. This study highlights gene regulatory pathways in the developing brain that may contribute to acquisition of proficient speech.

Fig. 1

De novo pathogenic variants in CHD3, SETD1A and WDR5. a Sanger validation of de novo pathogenic variants in family 01, 04 and 07. b Conservation of the mutated amino acids across species. Blue boxes mark the mutated amino acid. The blue line indicates the site of the duplication. c Schematic representation of CHD3 (ENST00000380358) organization, with CHD C- and N-terminal domains in black, plant homeodomain (PHD) zinc finger domains in blue, chromatin organization modifier (CHROMO) domains in green, DEAD-like helicases superfamily (DEXDc) domain in yellow, a helicase superfamily c-terminal (HELICc) domain in red and two domains of unknown function (DUF) in gray. Blue line indicates site of the p.R1228W variant. d Schematic representation of SETD1A (ENST00000262519) protein, with an RNA recognition motif (RRM) in red, a complex proteins associated with Set1p (COMPASS) component N (N-SET) domain in yellow, a Su(var)3–9, Enhancer-of-zeste, Trithorax (SET) domain in green and a post-SET domain in blue. Blue line indicates site of the frameshift, and the blue shaded area indicates the deleted part of the protein. e Schematic representation of WDR5 (ENST00000358625), with WD40 repeats in green. Blue line indicates site of the p.T208M variant

Fig. 2

Co-expression network analysis. a A co-expression network was calculated using gene expression data of brain samples collected between 8 weeks post conception up to 1 year of age from the cortex, hippocampus, amygdala, striatum and thalamus. A total of 16 modules were detected. Module 3 (indicated by black arrow) was highly enriched for the genes we implicated in CAS through whole-genome sequencing (CAS-WGS). b Enrichment of developmental disorder gene sets in the 16 modules. Gene sets included are the 10 genes implicated in CAS through WGS (CAS-WGS), and genes with de novo mutations in patients with the following (1) intellectual disability (ID, n = 230), (2) autism spectrum disorder (ASD, n = 2760), and (3) schizophrenia (SCZ, n = 711). Significant enrichments with False discovery rate (FDR)-corrected p value < 0.05 and odds ratio (OR) >1 are shown. Colors indicate FDR-corrected p values for enrichment. Numbers show OR

Fig. 3

Human brain expression pattern and functional enrichment of module M3. a Spatial expression pattern of the CAS-related module at 13–24 weeks post conception, as visualized by the M3 module’s Eigengene. Red shows high expression, blue shows low expression. Multiple samples per region were averaged. No expression data were available for gray regions. b Developmental brain expression pattern of the enriched module during development, as visualized by the module Eigengene. Each dot represents a brain sample, the black line is the loess curve fitted through the data points. The blue vertical line represents time of birth. Pcw: post conception week. c Gene functions enriched in the module identified through gene ontology (GO) term enrichment followed by clustering of GO terms using the functional annotation clustering tool in DAVID. The p values represent the geometric mean of Bonferroni-corrected p values of all GO terms underlying each function. The blue vertical line represents the threshold for significant enrichment (p = 0.05). Cx cortex, PF prefrontal