Abstract
Enhancers are crucial elements for the control of gene expression during embryonic development. The ascidian Ciona intestinalis offers unique experimental features to study these cis-regulatory sequences: enhancers are generally small and compact and their activity can be tracked at the single cell level thanks to the invariant cell lineage of ascidian embryos.Previous work identified two independent signatures associated with enhancer activity: the presence of specific transcription factors binding sites (TFBS) and a global dinucleotide signature along enhancers (Khoueiry, 2010). Although they correlate with enhancer activity, these signatures are insufficient to identify enhancer sequences from their sole sequence. During my thesis, I used a well-characterized early neural Ciona enhancer, the a-element of the Otx gene, as a model enhancer. This small (55pb) enhancer, is bound by GATA-a and ETS1/2 and is activated by the FGF pathway. To better understand the determinants of early neural enhancer activity, I tested the impact of point mutations affecting the affinity of the a-element TFBS for their binding TF and of the randomization of the spacer sequences that separate the TFBS in four ETS and GATA binding site clusters.Our results suggest at least two levels of cis-regulatory control: spatiotemporal specificity of enhancer activity is encoded in the identity of TF-binding sites, while the level of enhancer activity is set both by the affinity of TFs for their binding sites and by the composition of the spacer sequences. A surprisingly high number of variants of the a-element with randomized spacers are active, always in the same cell lineages as the WT. These variants, however, display a wide range of activity levels. This effect is also observed when the spacers in another active ETS/GATA cluster are randomized. Randomization of the spacers can even confer enhancer activity to a large fraction of inactive cluster variants. Consistent with their early neural activity and with the presence of ETS- and GATA-binding sites, these variants are, like the a-element, responsive to the FGF neural inducer.We could not link the action of the spacers on enhancer activity to any simple nucleotide or dinucleotide sequence features and it currently remains unclear why it is so easy to create a synthetic enhancer while most putative genomic ETS/GATA clusters are inactive. Using in vitro transcription factor binding assays, we showed that randomization of spacer sequences can affect TF binding to the a-element without changing the primary sequence of the binding site, and that extended minimal TFBS do not always recapitulate binding to the whole element. These results suggest that the physical structure of the DNA helix around the binding sites may play an important role in the control of enhancer activity.