The spatiotemporal expression of genes is controlled by enhancer sequences that bind transcription factors. Identifying the target genes of enhancers remains difficult because enhancers regulate gene expression over long genomic distances. To address this, we used an evolutionary approach to build two genome-wide maps of predicted enhancer–gene associations in the human and zebrafish genomes. Evolutionary conserved sequences were linked to their predicted target genes using PEGASUS, a bioinformatics method that relies on evolutionary conservation of synteny. The analysis of these maps revealed that the number of predicted enhancers linked to a gene correlate with its expression breadth. Comparison of both maps identified hundreds of putative vertebrate ancestral regulatory relationships from which we could determine that predicted enhancer–gene distances scale with genome size despite strong positional conservation. The two maps represent a resource for further studies, including the prioritization of sequence variants in whole genome sequence of patients affected by genetic diseases.

Developmental genes can harbour multiple transcriptional enhancers that act simultaneously or in succession to achieve robust and precise spatiotemporal expression. However, the mechanisms underlying cooperation between cis-acting elements are poorly documented, notably in vertebrates. The mouse gene Krox20 encodes a transcription factor required for the specification of two segments (rhombomeres) of the developing hindbrain. In rhombomere 3, Krox20 is subject to direct positive feedback governed by an autoregulatory enhancer, element A. In contrast, a second enhancer, element C, distant by 70 kb, is active from the initiation of transcription independent of the presence of the KROX20 protein. Here, using both enhancer knock-outs and investigations of chromatin organisation, we show that element C possesses a dual activity: besides its classical enhancer function, it is also permanently required in cis to potentiate the autoregulatory activity of element A, by increasing its chromatin accessibility. This work uncovers a novel, asymmetrical, long-range mode of cooperation between cis-acting elements that might be essential to avoid promiscuous activation of positive autoregulatory elements.

Cis-regulation plays an essential role in the control of gene expression, and is particularly complex and poorly understood for developmental genes, which are subject to multiple levels of modulation. In this study, we performed a global analysis of the cis-acting elements involved in the control of the zebrafish developmental gene krox20. krox20 encodes a transcription factor required for hindbrain segmentation and patterning, a morphogenetic process highly conserved during vertebrate evolution. Chromatin accessibility analysis reveals a cis-regulatory landscape that includes 6 elements participating in the control of initiation and autoregulatory aspects of krox20 hindbrain expression. Combining transgenic reporter analyses and CRISPR/Cas9-mediated mutagenesis, we assign precise functions to each of these 6 elements and provide a comprehensive view of krox20 cis-regulation. Three important features emerged. First, cooperation between multiple cis-elements plays a major role in the regulation. Cooperation can surprisingly combine synergy and redundancy, and is not restricted to transcriptional enhancer activity (for example, 4 distinct elements cooperate through different modes to maintain autoregulation). Second, several elements are unexpectedly versatile, which allows them to be involved in different aspects of control of gene expression. Third, comparative analysis of the elements and their activities in several vertebrate species reveals that this versatility is underlain by major plasticity across evolution, despite the high conservation of the gene expression pattern. These characteristics are likely to be of broad significance for developmental genes.

Gene regulation by enhancer sequences controls the spatiotemporal expression of target genes, often over long genomic distances. Current methods used to identify these interactions in the human genome require complex experimental setups and are restricted to specific cell types. Here we use PEGASUS, an approach that captures evolutionary signals 5 of synteny conservation to predict such interaction potentially in any biological context. We apply it to the human and zebrafish genomes and exploit the 1.3 million human and 55,000 zebrafish enhancers that we predicted to uncover fundamental principles of gene enhancer function in vertebrates. We show that the number of enhancers linked to a gene positively correlates with expression breadth in space and time and that the enhancer-target distance 10 is evolutionarily neutral. We uncover almost 2000 regulatory interactions ancestral to vertebrates, which are strongly enriched in core developmental processes. Using PEGASUS and its evolutionary view of enhancer-gene interactions, we provide a highly complementary resource to functional assays which uncovers key principles of enhancer biology.