Differential Gene Expression in Children with Autism Born to Mothers with Obesity and Diabetes

Background:  Autism spectrum disorder (ASD) is one of the most heritable neurobehavioral conditions, but considerable heterogeneity in ASD-concordant twins supports the need to investigate the role of alternative factors in its etiology. Maternal gestational and environmental factors can modify gene expression independent of the genes inherited in-utero and thus influence subsequent neurogenesis and brain development.  Underlying mechanisms responsible for these associations are poorly understood.  Environmental exposures, maternal physiology and genetic susceptibility may influence offspring ASD risk by altering gene dosage and expression patterns, which may ultimately have a profound influence on the course of neurodevelopment.

Objectives: To determine 1) the extent to which maternal diabetes or pre-pregnancy obesity is associated with differential gene expression in children with ASD and 2) the degree to which expression profiles differ between children with ASD and typical development (TD). 

Methods: This study involved 2-5 year-old children enrolled in the CHildhood Autism Risk from Genetics and the Environment (CHARGE) population-based case-control study.  ASD diagnosis was confirmed with ADI-R and ADOS.  Maternal prepregnancy body mass index and diagnosis of gestational or type-2 diabetes were abstracted from medical records.  RNA sequencing of peripheral blood obtained from child participants at enrollment was performed on the Illumina HiSeq 2000 platform.  Sequences that failed quality control or had fewer than 30 million reads were discarded, leaving 152 ASD and 73 TD samples.  Reads were aligned to hg19 using Bowtie and converted to a counts table using HTSeq.  To address differences in sequencing depth between samples, the counts table was normalized using binomial down-sampling.  Association of gene expression with maternal prepregnancy obesity or diabetes was analyzed using a negative binomial generalized linear model in the Bioconductor package edgeR.  Curated gene interaction data were retrieved from the Comparative Toxicogenomics Database.

Results:  In children with ASD, maternal prepregnancy obesity and diabetes were each associated with significant FDR-adjusted differential expression (30 and 5 genes, respectively), compared to those whose mothers lacked these conditions.   Differential expression was also present among children with TD exposed to maternal obesity or diabetes, but the genes differed from those found in children with ASD. While 34 of the 35 genes with significant differential expression in the children with ASD were up-regulated, none of these genes had significant differential expression in children with TD.  These differentially-expressed gene profiles are affected by hormones such as estradiol, testosterone, and thyroid hormone; micronutrients involved in methylation and immune regulation; medications such as acetaminophen and valproic acid; and exogenous environmental exposures such as endocrine disruptors and volatile organic compounds.

Conclusions:  Fetal exposure to maternal obesity or diabetes in children with ASD was associated significantly with up-regulation of a number of genes responsive to several endogenous and exogenous exposures implicated in ASD risk.  As these findings were limited to children with ASD, it is reasonable to speculate about the involvement of unmeasured environmental and genetic influences in this relationship.