Abstract
The multi-scale organization of eukaryotic genomes defines and regulates cellular identity and tissue-specific functions1–3. At the kilo-megabase scales, genomes are partitioned into self-interacting modules or topologically associated domains (TADs) 4–6. TADs formation seems to require specific looping interactions between TAD borders 7,8, while association of TADs can lead to the formation of active/repressed compartments 9. These structural levels are often seen as highly stable over time, however, recent studies have reported different degrees of heterogeneity 10,11. Access to single-cell absolute probability contact measurements between loci and efficient detection of low-frequency, long-range interactions is thus essential to unveil the stochastic behaviour of chromatin at different scales. Here, we combined super-resolution microscopy with state-of-the-art DNA labeling methods to reveal the variability in the multiscale organization of chromosomes in different cell-types and developmental stages in Drosophila. Remarkably, we found that stochasticity is present at all levels of chromosome architecture, but is locally modulated by sequence and epigenetic state. Contacts between consecutive TAD borders were infrequent, independently of TAD size, epigenetic state, or cell type. Moreover, long-range contact probabilities between non-consecutive borders, the overall folding of chromosomes, and the clustering of epigenetic domains into active/repressed compartments displayed different degrees of stochasticity that globally depended on cell-type. Overall, our results show that stochasticity can be specifically modulated to give rise to different levels of genome organization. We anticipate that our results will guide new statistical models of genome architecture and will be a starting point for more sophisticated studies to understand how a highly variable, multi-scale organization can ensure the maintenance of stable transcriptional programs through cell division and during development.
