Chromatin accessibility in the Drosophila embryo is determined by transcription factor pioneering and enhancer activation [ChIP-nexus]

Spatiotemporal gene regulation during embryonic development is driven by cis-regulatory DNA sequences called enhancers. Enhancers are activated through a combination of transcription factors (TFs) that bind to short sequence motifs within these sequences, but the order of events by which TFs read out motifs is not clear. Some TFs can only bind chromatin that is already accessible, while other TFs called pioneers can open chromatin themselves. Identifying motifs and the order by which they drive chromatin accessibility is very challenging. The recent implementation of convolutional neural networks, which learn complex cis-regulatory sequence rules that are predictive for genomics data, provides an unprecedented opportunity to dissect this problem. Here, we trained base-resolution deep learning models and applied them to high-resolution TF binding and chromatin accessibility data from the well-studied early Drosophila embryo. We uncover a clear hierarchical relationship between the pioneer Zelda and the TFs involved in the spatiotemporal patterning of the embryo, consistent with Zelda being a pioneer. However, the models predict that patterning TFs can also augment chromatin accessibility in a context-specific manner. Using a series of Drosophila mutant strains, we find that the two types of TFs increase chromatin accessibility by distinct mechanisms. Zelda’s pioneering is proportional to motif affinity, while the patterning TFs specifically increase chromatin accessibility when they mediate enhancer activation. This was conclusive because Dorsal can function both as activator and repressor, and the effect on chromatin accessibility depended on Dorsal’s transactivation effect and not on its binding per se. In conclusion, chromatin accessibility occurs in two phases: one through pioneering, which makes regions first accessible but not necessarily active, and a second when the correct combination of transcription factors lead to enhancer activation. Chromatin immunoprecipitation DNA-sequencing (ChIP-nexus) for Zelda, Dorsal, Twist, Bicoid, Caudal, and GAGA Factor, as well as H3K27ac and H3K4me1 ChIP-seq experiments, in 2-3 hour after egg laying (AEL) Drosophila melanogaster embryos.