Integrative functional genomics of age-related macular degeneration

Abstract

Purpose : Mechanistic interpretations of genetic variants associated with complex disease remain an open question. Genome-wide association studies (GWAS) have revealed 52 independent variants at 34 distinct loci associated with age-related macular degeneration (AMD). However, we have little insight into their causal mechanisms as majority of signals reside in the non-coding genome. We hypothesize that a majority of associated variants impact AMD progression and pathology through their effect on context- or tissue-specific gene regulation.

Methods : Donor eyes were evaluated for absence/presence and severity of AMD using the Minnesota Grading System. RNA and DNA was isolated from 128 controls, 197 early, 126 intermediate and 66 advanced AMD retina. RNA-Seq was performed on the Hiseq 2500, reads were aligned using STAR, and RSEM was used for gene-level quantification. We used limma for differential expression, GSEA to examine pathways, and WGCNA to identify co-expression modules. eQTL analysis on ~9 million genotyped (HumanCoreExome array) and imputed variants was performed using QTLtools. Integration of AMD-GWAS with eQTL was executed using SHERLOCK and TWAS.

Results : Analysis of reference transcriptome (from healthy donors) revealed that ~28% of the annotated genes (64% coding and 9% non-coding) were expressed in retina. A comparison with 44 tissues from the GTEx consortium demonstrated 257 retina-enriched/specific genes. Several AMD associated pathways, especially those associated with immune response, were up regulated in early AMD with later alterations in complement and ECM pathways. Co-expression network analysis also showed modules enriched for immune and ECM pathways. We identified a total of 2,002,101 cis-eQTLs representing 60% of expressed transcriptome in retina. Combined evaluation of retinal eQTL with GTEx data allowed the global view of eQTL regulation with 591,740 signals being retina-specific/enriched. Finally, the integration of GWAS data with eQTL analysis fine-mapped 11 known AMD loci and identified four novel candidate genes (p-value ≤ 2.02E-05 , FDR ≤ 0.2).

Conclusions : Our data represent the largest resource of retinal transcriptome and highlights the genes, pathways and regulatory mechanisms associated with AMD progression. Additionally, this resource provides functional genomic foundation for post-GWAS analysis across multiple vision-related complex traits, especially glaucoma and diabetic retinopathy.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.