Identification of brain cell types underlying genetic association with word reading and correlated traits

Abstract

Neuroimaging studies implicate multiple cortical regions in reading ability/disability. However, the neural cell types integral to the reading process are unknown. To contribute to this gap in knowledge, we integrated genetic results from genome-wide association studies for word reading (n = 5054) with gene expression datasets from adult/fetal human brain. Linkage disequilibrium score regression (LDSC) suggested that variants associated with word reading were enriched in genes expressed in adult excitatory neurons, specifically layer 5 and 6 FEZF2 expressing neurons and intratelencephalic (IT) neurons, which express the marker genes LINC00507, THEMIS, or RORB. Inhibitory neurons (VIP, SST, and PVALB) were also found. This finding was interesting as neurometabolite studies previously implicated excitatory-inhibitory imbalances in the etiology of reading disabilities (RD). We also tested traits that shared genetic etiology with word reading (previously determined by polygenic risk scores): attention-deficit/hyperactivity disorder (ADHD), educational attainment, and cognitive ability. For ADHD, we identified enrichment in L4 IT adult excitatory neurons. For educational attainment and cognitive ability, we confirmed previous studies identifying multiple subclasses of adult cortical excitatory and inhibitory neurons, as well as astrocytes and oligodendrocytes. For educational attainment and cognitive ability, we also identified enrichment in multiple fetal cortical excitatory and inhibitory neurons, intermediate progenitor cells, and radial glial cells. In summary, this study supports a role of excitatory and inhibitory neurons in reading and excitatory neurons in ADHD and contributes new information on fetal cell types enriched in educational attainment and cognitive ability, thereby improving our understanding of the neurobiological basis of reading/correlated traits.

Fig. 1. Input for LDSC.

For LDSC analysis, GWAS summary statistics for word reading or related traits, educational attainment, cognitive ability, and ADHD, were used. The command munge_sumstats.py converted summary statistic files from.txt to.sumstats.gz files. Single cell and bulk RNA sequencing gene expression data was used to make the annotation files. Mean expression or specific expression was calculated. Only the top 10% most expressed or specific genes were used for LDSC. Baseline annotations were included as controls.

Fig. 2. Results of LDSC for Allen Brain Bank, Kriegstein, and Shen RNA sequencing datasets.

X-axis are the cell types denoted by layer and gene marker. Y-axis is the GWAS summary statistics used. Class is colour coded, purple are excitatory neurons, pink are inhibitory neurons, white are oligodendrocytes, hot pink are astrocytes, grey are newborn inhibitory neurons, forest green are intermediate progenitor cells, light pink are radial glia (RG), and black are non-neural cells (endothelial, pericytes). The blue is the negative log10 of the FDR q-value (mean analysis). Black boxes represent cell types that reached significance (FDR < 0.05). a Allen Brain Bank adult single nuclei RNA sequencing data. b Kriegstein fetal single cell RNA sequencing data. c Shen fetal RNA sequencing data.