Distribution of segmental duplications in the context of higher order chromatin organisation of human chromosome 7

Abstract

Background: Segmental duplications (SDs) are not evenly distributed along chromosomes. The reasons for this biased susceptibility to SD insertion are poorly understood. Accumulation of SDs is associated with increased genomic instability, which can lead to structural variants and genomic disorders such as the Williams-Beuren syndrome. Despite these adverse effects, SDs have become fixed in the human genome. Focusing on chromosome 7, which is particularly rich in interstitial SDs, we have investigated the distribution of SDs in the context of evolution and the three dimensional organisation of the chromosome in order to gain insights into the mutual relationship of SDs and chromatin topology.

Results: Intrachromosomal SDs preferentially accumulate in those segments of chromosome 7 that are homologous to marmoset chromosome 2. Although this formerly compact segment has been re-distributed to three different sites during primate evolution, we can show by means of public data on long distance chromatin interactions that these three intervals, and consequently the paralogous SDs mapping to them, have retained their spatial proximity in the nucleus. Focusing on SD clusters implicated in the aetiology of the Williams-Beuren syndrome locus we demonstrate by cross-species comparison that these SDs have inserted at the borders of a topological domain and that they flank regions with distinct DNA conformation.

Conclusions: Our study suggests a link of nuclear architecture and the propagation of SDs across chromosome 7, either by promoting regional SD insertion or by contributing to the establishment of higher order chromatin organisation themselves. The latter could compensate for the high risk of structural rearrangements and thus may have contributed to their evolutionary fixation in the human genome.

Figure 1

Distribution of segmental duplications (SDs) and bundled long distance interactions in relation to acetylation of H4K8, transcriptional activity and lamina associated domains on human chromosome 7 (derived from IMR90 unless indicated otherwise). A) H4K8 acetylation profile, dark yellow: hyperacetylation of H4K8; blue: hypoacetylation of H4K8. B) the red and blue curve represent RNA-seq read counts/100 kb bin for coding and non-coding RNA, respectively (IMR91L). C) grey areas underlying the two histograms mark lamina associated domains (LADs, Tig3 cells). D) idiogram of chromosome 7, the Williams-Beuren syndrome region is highlighted in yellow beside the idiogram (at 72-74 Mb, hg18). E) transparent blue shading of the idiogram illustrates the inversion-affected segments of chromosome 7 depicted in Figure 2A-C. Bundled long distance interactions (F) and segmental duplications (G) are depicted in the inner circle; green ribbons: long distance interactions between genomic regions; grey: SDs with sequence similarity <98%; yellow: SDs with sequence similarity 98-99%; orange: SDs with sequence similarity >99%.

Figure 2

Long distance interactions of human chromosome 7 connect sequences syntenic to the most proximal 17.9 Mb of marmoset chromosome 2 and cluster in regions rich in SDs, Alu repeats and G4 motifs. A-C) Circos plots showing the patterns of long distance interactions (green bundles) in relation to SDs (following the colouring scheme of Figure 1) within the three segments of human chromosome 7 affected by the pericentric and paracentric inversions (as highlighted in blue in the idiogram of Figure 1); (A) before and (B) after in silico reversion of the paracentric inversion and (C) after reverting the pericentric inversion. The partial red and blue shading of the idiogram in A and B indicates the genomic interval inverted by the paracentric and pericentric inversion, respectively. D) distribution of SDs, long distance interactions (LDIs), G4 motifs and Alu repeats across human (Hs) chromosome 7 (100 kb bins) and its relation to marmoset (Cj) chromosome 2 syntenic regions (green blocks). Pink blocks highlight sequences syntenic to regions of marmoset chromosome 8. E-G) enrichment of SDs, Alu repeats and G4 motifs within chromosome 7 segments homologous to sequences of marmoset chromosome 2 (highlighted in blue). Chromosome 7 segments (binned in 200 kb windows) are displayed in ranked order according to feature count. The red curve and red dot above each plot indicate the hypergeometric score and its minimum (mHG), respectively.

Figure 3

Higher order chromatin organisation and SD localisation around the Williams-Beuren syndrome region. All data are referring to genome release hg19 and are derived from IMR90 unless indicated otherwise. The proximal, central and distal SD clusters (P, C, D) of the 7q11 segment encompassing 4.8 Mb are highlighted in green within the chromosome banding track. A-C) localisation of SDs; colouring according to sequence similarity; grey: <98%, yellow: 98%-99%; orange: >99%; D) genomic interval commonly deleted in WBS and the distal 7q11.23 deletion syndrome; E) topological domains as defined by Dixon et al. [36]; F) topological domains identified in the corresponding region in mouse [36] after conversion to human hg19. Note that the murine topological domain homologous to sequences deleted in the distal 7q11.23 syndrome is not fully represented due to a break of synteny within this genomic interval. See Figure 4 for details; G-H) heatmap and arc view of CTCF binding sites as detected by ChIA-PET in MCF7; I) number of G4 motifs/100 kb bins; J) average GC-content within 100 kb bins; K) number of Alu repeats/100 kb bin; L) number of structural variants as annotated by Database of Genomic Variance (DGV) [104], *maximum of 1080 CNVs not shown; M) log2 ratio scores of the LaminB1 DamID Map (Tig3 cells) as reported by Guelen et al. [45]; N) log2 ratio scores of DNA regions prone to early apoptotic DNA degradation in 20 kb windows, turquoise: degraded DNA segments; O) log2 ratio scores of H4K8 acetylation profile in 20 kb windows, blue: hyperacetylation, grey: hypoacetylation; P) red curve representing the sum of all intrachromosomal interaction counts/bin divided by the median number of interactions for all bins of chromosome 7; Q) percentage of interactions categorised according to their interaction span size; light grey: <0.5 Mb, grey: 0.5-1 Mb, light blue: 1–5 Mb, light brown: 5–10 Mb, dark grey: 10–25 Mb, black: ≥25 Mb. Gaps in this plot are due to alignment problems of Hi-C data in regions harbouring SDs with high sequence similarity.

Figure 4

Cross-species comparison showing that SDs next to the WBS locus have inserted at topological domain borders. Hi-C interactions and topological domains in the human fetal fibroblast cell line IMR90 are shown in dark green in the upper part as triangle view and bars, respectively. SDs with sequence similarity of 98%-99% and above 99%, respectively, (shown in yellow and orange in the SDs track) coincide with gaps within the Hi-C data. SD distribution and Hi-C data of the corresponding region in mouse are given in the lower part of the image. The position of FKBP6 and WBSCR16, the human orthologues of the two genes next to the murine topological domain borders are highlighted in green and red, respectively. The intervals commonly affected in WBS and the distal 7q11.23 syndrome are indicated by pale red bars. Note that the region distal to SRRM3 including the distal SD block are homologous to a different mouse chromosome.