Abstract EN:

We analysed the large-scale organisation of the human genome in terms of transcriptional activity, chromatin state and replication timing, focusing on a segmentation of the genome based on strand composition asymmetry. The large-scale variations of compositional skew along the human genome define Mbp scale skew domains that are bordered by large upward jumps. These upward jumps in the skew profile were proposed to be a subset of replication origins active in the germline that have been conserved throughout mammalian evolution. This first described skew domains were named N-Domains because their skew profile is shaped like an ’N’, and they were shown to be strongly linked both to the organisation of transcription and replication. Indeed, the borders of N-Domains colocalize with promoters of genes oriented divergently from the border and often with an early-replicating locus. Here, we described a novel kind of skew domains, Split-N-Domains, similar at their borders to N-Domains but larger (>3Mbp) and presenting a gene desert with a null and constant skew at their heart. Central regions of Split-N-Domains constitute an evidence of the link between absence of genes, heterochromatin and late replication timing, and we proposed that the initiation of replication in these regions could be random. We found that genes close to skew domains borders have in general a CpG island promoter, thus suggesting a link between skew domains borders and DNA hypomethylation and gene expression in the germline. Using both experimental and sequence-derived markers of chromatin state, we showed that the region a few 100 kbp large around skew domains borders is also often characterized by an open chromatin state, particularly when it is early-replicating. We discuss how this open chromatin region can relate to the organisation of genes and to the replication program. In addition, we asked how did the distribution of evolutionary breakpoints relate to genome organisation. We found, contrarily to the expectation, that breakpoint regions are over-represented in small intergenes. In fact, breakpoint density is higher in regions hypomethylated and with a higher sensitivity to DNase I. We proposed that the heterogeneous distribution of breakpoints in mammalian genomes could be due to a mutation bias in relation to chromatin state.