Gaining biological insights with Pore-C: investigating the spatial organisation of chromatin in the human genome

We generated contact data for human breast cancer cell line HCC1954 and control sample ANA51. Using the tool JaBbA we identified a complex, multi-chromosomal subgraph connecting chromosomes 9, 12 and 20 in HCC1954 (Fig. 1a). We obtained a genome graph by analysis of read-depth and junction calls. Higher-order contacts mapping to the region were used to nominate an amplified multi-junction derivative allele within the graph. The Pore-C alignments and associated contact maps visually confirm this putative derivative allele. For additional confirmation we designed multi-colour FISH probes to target locations on the allele (Fig. 1b). Co- localisation of probes in the HCC1954 metaphase spreads confirm the presence of an amplified multi-junction fusion of chromosomes 9, 12 and 20 (Fig. 1c) which is not seen in ANA51.

We generated a draft assembly with Shasta using a flow cell of nanopore WGS data; the resulting contigs are plotted in length order. Next, using Pore-C virtual pairwise contacts and contigs as inputs, we generated scaffolds with 3D-DNA before a second round of scaffolding using SALSA2, followed by Purge Haplotigs to remove regional duplications. The resulting assembly shows high congruity with the reference genome (Fig. 2a). Fig. 2b shows scaffolding for chromosomes 3, 4 and 5. The top map shows contact density derived from Pore-C reads mapped against the draft assembly (length-ordered), and the bottom contact map is derived from the same Pore-C reads mapped against the final scaffolds. The combination of all three methods gives the best result, with a scaffold NG50 of 125 Mb. (Fig. 2c).

To compensate for the presence of an additional copy of the X chromosome in females (XX) compared to males (XY), a random haplotype of the female X chromosomes is silenced through an epigenetic process known as X-chromosome inactivation (XCI). We made use of the publicly available chromosome-level haplotype information for the human cell line GM12878 to separate NlaIII- derived Pore-C reads according to whether they originated from the inactive (Xi) or active (Xa) X chromosome and to thereby detect differences in the DNA conformation and methylation state between the two haplotypes. Imaging studies have shown that Xi adopts a distinctive structural conformation known as the Barr body. 3C methods have shown that this condensed structure is in fact bipartite, consisting of two superdomains separated by a hinge region centred on the macrosatellite repeat locus DXZ4. Xa on the other hand has a more typical elongated structure with A/B compartments (Fig. 3a). Multiple layers of transcriptional repression ensure that the majority of the approximately 1,000 genes on Xi are not expressed, but ~20% of these genes can ‘escape’ this repression and are characterised by DNA methylation patterns more typical of genes on Xa (Fig. 3b). Allele-specific chromatin contact maps at 500 kb resolution (Fig. 3c) for Xi (above the diagonal) and Xa (below the diagonal) show Xi-specific superdomains as two triangles above the diagonal that touch at the DXZ4 region, while the checkerboard pattern associated with A/B compartmentalisation is visible on Xa. Methylation levels detected using Pore-C reads at CpG islands within 1 kb of transcription start sites confirms the pattern of differential methylation at loci subject to XCI. In contrast, similar levels of methylation are found at loci that escape XCI (Fig. 3d). A positional analysis of the Pore-C-derived methylation signal shows a high level of differential methylation around the promoter region for XCI loci but not for those that escape XCI (Fig. 3e).