Functional Impact of ASD-Associated Genetic Risk Variants in Human Cortical Development

Oral Presentation
Thursday, May 10, 2018: 11:45 AM
Jurriaanse Zaal (de Doelen ICC Rotterdam)
D. M. Werling1, S. Pochareddy2, J. Choi2, J. Y. An3, C. Dastmalchi3, M. Li2, M. State3, K. Roeder4, B. Devlin5, N. Sestan6 and S. Sanders3, (1)Psychiatry, UCSF, San Francisco, CA, (2)Yale University, New Haven, CT, (3)Psychiatry, University of California San Francisco, San Francisco, CA, (4)Carnegie Mellon University, Pittsburgh, PA, (5)Univ of Pittsburgh School of Medicine, Pittburgh, PA, (6)Yale School of Medicine, New Haven, CT
Background: Establishing clear links between genetic variation and gene function is a critical step toward translating genetic association into a mechanistic understanding of the etiology of autism spectrum disorder (ASD). Variants identified by genome-wide association studies (GWAS) and whole-genome sequencing (WGS) are frequently noncoding, but do not clearly implicate a specific gene or the manner in which the gene is impacted. Expression quantitative trait loci (eQTLs) can help clarify these questions by demonstrating functional impact on the expression of specific genes. Though ASD risk genes are strongly expressed during prenatal development, available eQTL discovery data sets are based largely on data from adult brains.

Objectives: To test adult brain eQTLs for intersection with ASD biology and to validate and extend these findings in the human cortex across brain development. Then, to apply these functional data to determine the developmental gene regulatory targets of noncoding ASD risk variants from recent WGS and GWAS data.

Methods: We tested gene targets of the eQTL SNPs identified in adult brain from the Common Mind Consortium (CMC) for enrichment with gene sets implicated in ASD. We also generated WGS and RNA-sequencing (RNA-seq) data from post-mortem dorsolateral prefrontal cortex (DLPFC) samples from 200 individuals aged 6 post-conception weeks (PCW) through 22 years. Gene-level transcript counts have been identified across the dataset. We are in the process of identifying high-quality common and rare single nucleotide variants and indels, allowing eQTLs to be identified using Matrix eQTL. We will identify eQTLs in mid-fetal (13-24 PCW) and adolescent (12-20 years) brains, along with a combined analysis. As with adult eQTLs, we will test for enrichment in ASD-related gene sets. Finally, we will use all of these eQTL sets to assess help identify and interpret variants in GWAS and WGS data from ASD samples.

Results: We find that CMC eQTL gene targets are enriched for CHD8 binding targets (adjusted p=4.1e-31) and an astrocyte-associated, ASD brain-up-regulated co-expression module (adj. p=0.003), but depleted for FMRP targets (adj. p=5.1e-13), genes with higher expression in the female brain (adj. p=0.03), and ASD-associated risk genes (adj. p=0.04). Our DLPFC expression data also confirmed published findings that age is the most significant biological driver of expression differences sample-wide, suggesting that fetal eQTLs are likely to differ substantially from those identified in adult brains.

Conclusions: eQTLs offer an opportunity to identify genomic regions that play important regulatory functions in the human brain and that are also sensitive to genetic variation. The enrichment for CHD8 binding sites and depletion for FMRP targets suggest that eQTLs in adult brain are capturing an important, but complicated, dimension in ASD biology. Replication of the observation that developmental stage is the most important component in determining cortical gene expression suggests that there are likely to be considerable differences between fetal and adult eQTLs. We therefore anticipate fetal eQTLs providing important and novel insights into the role of ASD genes in cortical development and illuminating genomic regions likely to contain noncoding ASD risk factors in GWAS and WGS data.