Natural Variation of Chromatin Accessibility reveals a new epithelial Enhancer model

Enhancers coordinate gene expression levels and the majority of disease related SNPs are located in these non-coding regions. Yet how genomic variation affects enhancer function is poorly understood. To investigate cis-regulatory variation in vivo, in an endogenous enhancer context, we profiled the genome-wide chromatin accessibility of epithelial tissues (imaginal discs), across 30 inbred Drosophila lines from the DGRP project. Statistical analysis identified 4289 chromatin accessibility QTLs (caQTL). We singled out the transcription factor Grainyhead as a key player, for which 70 caQTLs alter its binding site, thereby causing a concordant gain or loss in chromatin accessibility and in vivo enhancer activity. We show a clear uncoupling between enhancer accessibility and activity using an in vivo enhancer-reporter screen combined with cell-sorted ATAC-seq. The epithelial enhancers become accessible in all cells, due to the binding of the master regulatory Grainyhead, but activate gene expression only in a specific subpopulation of cells. Finally, to discriminate between functional and non-functional Grainyhead recognition motifs in the genome, we trained various machine learning algorithms and compared enhancers across different Drosophila species, allowing the identification of key enhancer features required for Grainyhead binding. In conclusion, we find a new epithelial enhancer model in which Grainyhead plays a similar role in epithelial tissues as Zelda in the embryo, uncoupling chromatin accessibility and enhancer activity for thousands of epithelial enhancers.status: publishe