The characterization of cis-regulatory modules (CRMs) and of their activity is central to understanding gene regulation and metazoan development. Chromatin immunoprecipitation followed by microarray or deep sequencing (ChIP-seq) against TFs are powerful approaches to map CRMs. To enable in vivo tissue-specific ChIP against ubiquitously expressed factors, we develop a ChIP protocol relying on the sorting of fluorescence activated cells, followed by deep sequencing. Using this protocol, we map histone modifications and RNA Polymerase II (PolII) occupancy in the Drosophila mesoderm, and subsequently study the chromatin state of active CRMs in vivo. We show that active CRMs are enriched for H3K27Ac, H3K79me3 and PolII, and that the presence and shape of these marks dynamically correlate with CRM activity timing and nucleosome positioning. Using Bayesian inference, we predict new CRMs to be active in the mesoderm and validate 89% of them in vivo. Next, we investigate how five TFs essential for cardiac specification operate in cis in the dorsal mesoderm, the developmental precursor of the visceral mesoderm (VM) and the cardiac mesoderm (CM). We demonstrate that they are recruited as a TF collective at cardiac CRMs without strong sequence requirements, thereby suggesting a novel mode for CRM activation. We further observe that cardiac TFs occupy CRMs that are active in the VM sibling lineage, echoing the fact that both cell populations derived from the dorsal mesoderm. We thus conclude that dormant TF binding signatures may reveal a developmental footprint of a cell lineage.
