Missense variants in the chromatin remodeler CHD1 are associated with neurodevelopmental disability

Abstract

Background: The list of Mendelian disorders of the epigenetic machinery has expanded rapidly during the last 5 years. A few missense variants in the chromatin remodeler CHD1 have been found in several large-scale sequencing efforts focused on uncovering the genetic aetiology of autism.

Objectives: To explore whether variants in CHD1 are associated with a human phenotype.

Methods: We used GeneMatcher to identify other physicians caring for patients with variants in CHD1. We also explored the epigenetic consequences of one of these variants in cultured fibroblasts.

Results: Here we describe six CHD1 heterozygous missense variants in a cohort of patients with autism, speech apraxia, developmental delay and facial dysmorphic features. Importantly, three of these variants occurred de novo. We also report on a subject with a de novo deletion covering a large fraction of the CHD1 gene without any obvious neurological phenotype. Finally, we demonstrate increased levels of the closed chromatin modification H3K27me3 in fibroblasts from a subject carrying a de novo variant in CHD1.

Conclusions: Our results suggest that variants in CHD1 can lead to diverse phenotypic outcomes; however, the neurodevelopmental phenotype appears to be limited to patients with missense variants, which is compatible with a dominant negative mechanism of disease.

Keywords: chromatin; epigenetic machinery; human disease; neurological dysfunction; speech apraxia.

Figure 1. Missense variants in the chromatin remodeler CHD1 result in a distinct neurological syndrome with dysmorphic features

(A) We have identified five unrelated subjects with missense variants at highly conserved locations within the coding region of the CHD1 gene. Three of these variants are de novo. Three other de novo variants previously described in large cohorts of subjects with autism,[–27] are noted in gray. Two of these variants occur at heavily conserved sites (1016, 1203) and the other (1517) leads to a frameshift at the C-terminal end of the protein. CHCT, which stands for CHD1 helical C-terminal, is an alpha-helical domain of unknown function,[43]. Domain boundaries are based on Chd1 structures,[–45]. R = Arginine, K = Lysine, D = Aspartic Acid, G = Glycine, Q = Glutamine. (B) Representative images of some of the phenotypic facial features in two of the subjects. (C) Locations of the de novo missense sites on the chromodomain-ATPase motor, based on the yeast Chd1 structure,[44]. Two orthogonal views are shown, with the location of the DNA duplex binding highlighted with dotted gray lines. Note that the Arg618Gln variant occurs at the interface between DNA and ATP, and also is expected to pack against ATPase lobe 2 in the active state. (D) This gene does not tolerate missense variation particularly well as it has a relatively high missense Z score compared to all genes. The data here are based on data from the Exome Aggregation Consortium,[35].

Figure 2. Patient fibroblasts demonstrate increased amounts of a closed chromatin modification (H3K27me3) compared to fibroblasts from control subjects

Since CHD1 is thought to play an active role in the process of opening chromatin we examined the amount of H3K27me3 in fibroblasts from one of our subjects (Subject 1) and fibroblasts from two control individuals. Briefly, samples were cultured in triplicate and stained with antibodies against H3K27me3 (green) and DAPI (blue). Intensity level of H3K27me3 in each nucleus was quantified using ImageJ. (A) Representative images of immunofluorescence staining from Subject 1 and two controls. (B) Quantification of H3K27me3 intensity, normalized to the average control intensity. Gray points represent H3K27me3 intensity in each nucleus and black bars represent the mean intensity of all nuclei measured. Controls were age and sex matched (Karyotypes: 46,XX and 46,XX,del(22)(q11.2q11.2)). * P<0.05, ***P<0.0005, one-way ANOVA with post hoc Tukey’s HSD analysis.