The short-sequence design of human chromosomes

Recent investigations have shown that isochores are characterized by a 3-D structure which is primarily responsible for the topology of chromatin domains. More precisely, an analysis of human chromosome 21 demonstrated that GC-poor isochores are low-heterogeneity sequences characterized by the presence of oligo-Adenines that are intrinsically stiff, curved and unfavorable for nucleosome binding. This leads to a structure of the corresponding chromatin domains, the Lamina Associated Domains, or LADs, which is well suited for interaction with lamina. In contrast, the high-heteorogeneity GC-rich isochores are in the form of compositional peaks characterized by gradients of oligo-Guanines that lead to increasing nucleosome depletions in the corresponding chromatin domains, the Topological Associating Domains, or TADs. These results encouraged us to investigate in detail the di- and tri-nucleotide profiles of 100Kb segments of chromosome 21, as well as those of the di- to octa-Adenines and di- to octa-Guanines in several regions of the chromosome. The results obtained show that the 3-D structures of isochores and chromatin domains depend not only upon oligo-Adenines and oligo-Guanines but also, to a lower but definite extent, upon the majority of di- and tri-nucleotides. This conclusion, which applies to all human chromosome, has strong implications for the biological role of non-coding sequences.