Cis-regulatory variation is accountable for serious deleterious effects, and yet it is common [14,20]. Understanding TFBS functional constraints is therefore interesting for at least two reasons. First, it may shed light on the regulatory architecture of the genomes. For example, our finding that CTCF motifs tolerate the lowest load a short distance away from TSSs underlines the importance of chromatin architecture at the distal ends of promoter regions. In addition, TFBS constraints are indicators of how the system deals with noise in cis-regulatory networks, and the variation analyses presented here support such phenomena as homotypic redundancy [48]. Interestingly, it was previously shown that homotypic clustering does not affect Drosophila TFBS turnover rate in the phylogenetic context [36], but the dynamics of selection inside a population need not correspond to that observed between species. For example, retaining multiple instances of neighboring homotypic sites in a given species may in itself bear the selective advantage to provide robust buffering to variation and other perturbations.
