Tissue-Specific Expression Quantitative Trait Loci (eQTL) in GI Symptomatic ASD Children

Background: Previously, we characterized, via whole transcriptome analysis of ileocolonic biopsy tissue, an IBD-like condition that occurs with high frequency in autism spectrum disorder (ASD) children. Using gene expression data from that study, together with a second dataset derived from SNP analysis of DNA from the same individuals, we are exploring expression quantitative trait loci (eQTL).

Objectives: The immediate goal of this study is to search for statistical associations between gene expression data and genetic polymorphisms in a population (ASD) and tissues (colon and terminal ileum) previously uncharacterized by eQTL studies. The longer-term goal is to determine if there are specific eQTLs that track with individual IBD subtypes.   

Methods: Gene expression data (generated from Agilent whole genome microarrays) and SNP data (generated at 23&me on custom Illumina SNP chips) from 64 individuals were used for the eQTL analyses. Standard quality control was completed for association studies (e.g., SNP call rate, Hardy-Weinberg Equilibrium). Analyses were computed separately for colon and terminal illium samples and by ASD^IC (ASD with ileocolitis). For each transcript and tissue type, we computed a genome-wide association analysis using linear regression on single nucleotide polymorphism (SNP) cis (within 500kb) to the probe’s gene. In this eQTL analysis, we regressed the SNP’s genotype and the first principal components onto log₂expression for the transcript. Given the modest sample size, only the dominant genetic model was computed. A fixed effect meta-analysis and the corresponding test for heterogeneity of effects were computed across disease groups. Significance of an eQTL effect was measured as the magnitude of the p-value conditional on an expression fold change of at least 1.5.

Results: The number of children analyzed varies by tissue. There were 18 ASD^IC with colon samples and 22 ASD^IC terminal ileum samples. Within the colon, 27 SNP-transcript combinations met the p<1x10E-6 and fold change >1.5. These 27 SNP-transcript combinations reside in 13 regions. Within the terminal ileum, 24 SNP-transcript combinations, residing in 13 regions, met the same threshold criteria. At this level of significance, one gene was observed in both tissues: Williams Beuren syndrome chromosome region 27 on chr 7q11.23. Deletion of genes in this region is known to cause the neurodevelopmental disorder Williams Beuren Syndrome and was recently implicated with ASD in a linkage study (Nijmeijer 2014). Additional genes at the given eQTL significance threshold, such as RYR2 (colon: rs10802598 p=2.25x10^-8) and EPHB1 (colon: rs10512944 p=1.88x10^-12) have been previously related to ASD and/or ASD-phenotypes through other studies and approaches.  

Conclusions:  This is the first eQTL analysis of colonic and terminal ileum tissues within an ASD population, providing a novel dataset. The benefits of such an eQTL analysis are two-fold: 1) potential elucidation of ASD SNP associations, by relating SNPs to gene expression and, 2) identification of polymorphisms that tag colonic and/or terminal ileum gene expression.  Additional analyses are underway to identify eQTLs in IBD (non-ASD) samples. A subsequent comparison across all groups will potentially identify eQTLs associated with specific tissue and that may have contrasting expression levels between disease and non-disease groups.