Woolfe, A.; Parker, S. C.; Almouzni, G.

Many human developmental enhancers are characterized by extreme evolutionary constraint in the vertebrate lineage and unique DNA sequence properties, the functional relevance of which is still unknown. Here, we investigate the consequences of their DNA sequence features on three potential aspects important for their function - transcription factor sequence recognition, chromatin accessibility and DNA structure. Using computational predictions in human as well as other vertebrates and invertebrates, we find that conserved non-coding elements (CNEs) are intrinsically nucleosome disfavoring at their core, but favor nucleosome occupancy at their borders driven by distinct nucleotide features conserved over large evolutionary distances. Nevertheless, using genome-wide nucleosome occupancy datasets, we find vertebrate CNEs exhibit higher nucleosome occupancy in comparison to surrounding regions in differentiated cells but a highly accessible conformation in embryonic tissues, suggesting a role for nucleosome positioning in their function. In addition, CNEs are exclusively enriched for homeobox transcription factor motifs, which are found at high density across their sequences. In particular, motifs specifically enriched at the boundary belong to the PBX-HOX, MEIS and POU transcription factor families, known to recognize specific DNA structural features. Consistent with this finding, CNE boundaries are enriched for unusual DNA structural motifs that may constitute a recognition mechanism by transcription factors that bind a narrow minor groove. The finding that extreme nucleotide conservation are likely to be driven by a combination of nucleosome and protein binding constraints provide a potential mechanistic insight into the function of early developmental enhancers.