The development from a single cell into a complex organism requires the precise control of gene expression in space and time. To achieve this, the activity of genes is governed by large regulatory chromatin landscapes that when disrupted can cause gene mis-regulation and disease. However, at the same time, the successful modification of these landscapes is thought to be a major driver of phenotypic innovation during evolution. Given the vulnerability of these landscapes in disease settings, it remains largely unknown how their integrity is maintained when novel genes are “safely” incorporated during evolution, which is addressed in this work. Specifically, here, multiple mechanisms are dissected that adapted the Fat1 regulatory landscape to maintain its integrity while simultaneously incorporating a novel gene, Zfp42, during evolution. First, comparative evolutionary genomics was used to reconstruct the history of the locus (section 1). Second, the three-dimensional chromatin configuration of the locus was examined in relationship to the gene activities using genomics-technologies (HiC, DamID) combined with super resolution microscopy and in silico modeling (section 2). Finally, the mechanisms that adapted the landscape in ESCs (section 3) and embryonic limbs (section 4) for the emergence of Zfp42 were investigated using genome engineering and genomics. Two tissue-specific mechanisms were identified that enabled the independent activities of Zfp42 and Fat1 despite sharing the same regulatory chromatin landscape: In ESCs, the landscape physically restructures and isolates the genes together with their regulatory information, from one another, thereby allowing their independent regulation. Surprisingly, this restructuring is not driven by the most recognized chromatin structuring force, loop extrusion, but rather by the underlying epigenetic state of chromatin. A different mechanism operates in embryonic mouse limbs where both genes are exposed to the same regulatory information driving Fat1 activation, but surprisingly not Zfp42. The inactivity of Zfp42 cannot be explained by nuclear envelopment attachment nor by enhancer-promoter specificity. Instead, Zfp42 is kept inactive by a highly context-dependent silencing mechanism driven by DNA methylation. As such, Zfp42 is ectopically active and responsive to the surrounding regulatory information when DNA methylation is removed or when the gene is slightly repositioned within its domain. Combined, we find that 3D-restructuring and context-dependent silencing adapted the Fat1 landscape to integrate Zfp42. More generally, this demonstrates that even single regulatory landscapes harbor an enormous regulatory complexity and, thus can accommodate multiple independently regulated genes. We believe that this has significant consequences for human genetics where similar genomic alterations do not drive disease in patients. This is possible, because additional, yet still unknown, mechanisms control how regulatory information is used in the genome.
