The role of long-range super-enhancers in craniofacial morphogenesis

One of the most defining characteristics that sets individuals apart from each other is the shape of their face. Craniofacial morphogenesis is a fascinating and highly complex process that involves the coordination of multiple genetic, cellular, and environmental factors. A central component of this process are cranial neural crest cells (CNCCs), a population of cells that temporarily possess stem cell-like characteristics, with the potential to differentiate into various types of cells. Furthermore, growing evidence suggests that distal cis-regulatory elements called enhancers are major contributors to craniofacial development. In particular, large clusters of enhancers known as super-enhancers have been associated with genes that play crucial roles in differentiation and cell identity determination. However, while many putative craniofacial enhancers have been identified, only a few have been functionally validated. Furthermore, the role of super-enhancers (SEs) in craniofacial morphogenesis has not been explored so far. Thus, we decided to focus on the identification of putative craniofacial super-enhancers.
Using a combination of ChIP-seq and Promoter-Capture Hi-C (PCHi-C) data we identified 2,322 putative long-range SEs of which 177 specifically interact with transcription factor genes involved in the establishment of CNCCs’ positional identity in craniofacial development. Out of this list of SEs, five SEs targeting Hoxa2, a master regulator of second pharyngeal arch (PA2) CNCCs fate, specifically in PA2-derived CNCCs stood out due to their extremely long genomic distance to Hoxa2 and their ability to interact with Hoxa2 across a topologically associating domain (TAD) boundary. This configuration struck us since TADs are known to spatially insulate regulatory elements and preferentially show interactions within themselves, thereby creating discrete functional and structural blocks that promote intra-domain enhancer-promoter interactions, while limiting contacts between regulatory elements across TADs. According to their genomic distance to each other and the distribution of H3K27ac, we separated the Hoxa2 SEs into Hoxa Inter-TAD Regulatory Element 1 and 2 (HIRE1 and HIRE2). Both of these large regulatory elements are highly conserved within eutherian mammals and upon deletion in mice, we observed varying phenotypes. Homozygous deletion of HIRE1 resulted in a phenocopy of the full homeotic Hoxa2 knock-out phenotype and abnormalities in PA3- and PA4-derived structures, which correlated a strong downregulation in gene expression of Hoxa2 and Hoxa3 in CNCCs and hindbrain rhombomeres. On the other hand, deletion of HIRE2 only caused mild alterations of CNCC-derived skeletal structures and a small external ear on a Hoxa2 haploinsufficient sensitized background. These phenotypes are particularly striking because both Hoxa2 and Hoxa3 were previously believed to be regulated by proximal enhancers in CNCCs and the hindbrain. A transcription factor motif analysis in combination with the analysis of PCHi-C data of PA2 CNCCs from Hoxa2 knockout mice and published ChIP-seq data for Hoxa2, Pbx, and Meis from PA2 CNCCs suggest that Hoxa2 is at least partially involved in its own regulation together with its co-factors Meis and Pbx.
Taken together, the findings presented in this thesis give new insights into the importance of long-range SEs in controlling key developmental transcription factors in craniofacial morphogenesis, which could have implications for the study of disease and rare congenital disorders. Additionally, they change our understanding of the regulation of Hoxa2 in CNCCs from a model that solely relies on multiple proximal enhancers, to a model that is based on multiple long-range inter-TAD SEs.