Cell-type-specific neuroanatomy of brain-wide expression of autism-related genes

TL;DR

This study uses computational neuroanatomy to analyze the cell-type-specific expression of autism-related gene cliques across the mouse brain, revealing their enrichment in granular layers and specific cell types.

Contribution

It introduces a novel computational approach to rank cell types by similarity to autism-related gene expression profiles in the mouse brain.

Findings

Gene expression is highly co-expressed in the cerebellar cortex.

Expression patterns are enriched in the granular layer.

Granule cells show the highest similarity to autism-related gene cliques.

Abstract

Two cliques of genes identified computationally for their high co-expression in the mouse brain according to the Allen Brain Atlas, and for their enrichment in genes related to autism spectrum disorder, have recently been shown to be highly co-expressed in the cerebellar cortex, compared to what could be expected by chance. Moreover, the expression of these cliques of genes is not homogeneous across the cerebellum, and it has been noted that their gene expression pattern seems to highlight the granular layer. However, this observation was only made by eye, and recent advances in computational neuroanatomy allow to rank cell types in the mouse brain (characterized by their transcriptome profiles) according to the similarity between their density profiles and the expression profiles of the cliques. We establish by Monte Carlo simulation that with probability at least 99 percent, the expression profiles of the two cliques are more similar to the density profile of granule cells than 99 percent of the expression of cliques containing the same number of genes (Purkinje cells also score above 99 percent in one of the cliques). Thresholding the expression profiles shows that the signal is more intense in the granular layer.