Architectures of somatic genomic rearrangement in human cancer amplicons at sequence-level resolution

Abstract

For decades, cytogenetic studies have demonstrated that somatically acquired structural rearrangements of the genome are a common feature of most classes of human cancer. However, the characteristics of these rearrangements at sequence-level resolution have thus far been subject to very limited description. One process that is dependent upon somatic genome rearrangement is gene amplification, a mechanism often exploited by cancer cells to increase copy number and hence expression of dominantly acting cancer genes. The mechanisms underlying gene amplification are complex but must involve chromosome breakage and rejoining. We sequenced 133 different genomic rearrangements identified within four cancer amplicons involving the frequently amplified cancer genes MYC, MYCN, and ERBB2. The observed architectures of rearrangement were diverse and highly distinctive, with evidence for sister chromatid breakage-fusion-bridge cycles, formation and reinsertion of double minutes, and the presence of bizarre clusters of small genomic fragments. There were characteristic features of sequences at the breakage-fusion junctions, indicating roles for nonhomologous end joining and homologous recombination-mediated repair mechanisms together with nontemplated DNA synthesis. Evidence was also found for sequence-dependent variation in susceptibility of the genome to somatic rearrangement. The results therefore provide insights into the DNA breakage and repair processes operative in somatic genome rearrangement and illustrate how the evolutionary histories of individual cancers can be reconstructed from large-scale cancer genome sequencing.

Figure 1.

Somatic rearrangements in BACs. (A) Chromosome 17q21 amplicon in HCC1954 including ERBB2; (B) chimeric amplicon in HCC1954 including MYC; (C) chimeric amplicon in NCI-H2171 including MYC; (D) chromosome 2 amplicon in NCI-H1770 including MYCN. The color of the arrow identifies the chromosome, and the direction of the arrow indicates the orientation of DNA sequence relative to the reference genome (+ or − orientation); IVD, putative inverted duplications; black rectangles, DNA “insertions” that do not align to the reference human genome sequence with either flanking sequence. Figure parts are drawn to emphasize the types and complexity of rearrangements within the clone sequences and are therefore not drawn to scale.

Figure 2.

Sequences at breakage–fusion junctions. (A) Example of putative homologous recombination based repair with extended microhomology; (B) example of nonhomologous end joining showing 5-bp overlapping microhomology; (C) example of nonhomologous end joining including putative nontemplated DNA synthesis. In each example, the BAC sequence is shown in the middle with the genomic sequences contributing to the rearrangement shown above and below. Regions of sequence identity are highlighted.

Figure 3.

Clustered genomic origins of rearranged DNA fragments. DNA segments within BAC 14g18 are shown in their reference genomic locations and orientations (see Supplemental Table 1). Fragments are shown in order 5′ to 3′ in the clone. One fragment from another BAC, 5am21, falls within the chromosome 12 interval (*). The scale shows the 120-kb region separated into 10-kb segments; the location of the expanded region is shown with respect to the amplicon for each chromosomal region. Fragments at the BAC vector insert junction are extended off-scale.