Chromothripsis: chromosomes in crisis

Abstract

During oncogenesis, cells acquire multiple genetic alterations that confer essential tumor-specific traits, including immortalization, escape from antimitogenic signaling, neovascularization, invasiveness, and metastatic potential. In most instances, these alterations are thought to arise incrementally over years, if not decades. However, recent progress in sequencing cancer genomes has begun to challenge this paradigm, because a radically different phenomenon, termed chromothripsis, has been suggested to cause complex intra- and interchromosomal rearrangements on short timescales. In this Review, we review established pathways crucial for genome integrity and discuss how their dysfunction could precipitate widespread chromosome breakage and rearrangement in the course of malignancy.

Figure 1. The spectrum of genetic instability in malignancy

Genomic alterations in tumors can be subdivided into three main groups. At the smallest scale, subtle sequence changes may affect one or a few adjacent nucleotides (left). Examples include deficiencies in mismatch repair (MMR) and nucleotide excision repair (NER) systems, which result in unstable microsatellite repeats (top) and retention of UV-induced photoproducts, such as thymine dimers (bottom). At an intermediate scale (middle), gross chromosomal rearrangements – including deletions, amplifications, inversions, and translocations – are a ubiquitous feature of most cancer genomes. GCRs can arise from multiple mechanisms, including telomere erosion and end-to-end fusion (depicted here), non-allelic homologous recombination, replication stress, and chromothripsis. At the largest scale, whole chromosome instability (right) causes not only aneuploidy but also loss of heterozygosity (LOH), which is crucial for unmasking recessive mutations in tumor suppressor genes. This form of instability can result from errors in virtually any aspect of mitosis, including dysregulation of the spindle assembly checkpoint; biogenesis of supernumerary centrosomes and multipolar spindles; untimely dissolution of sister chromatid cohesion; and formation of merotelic kinetochore-microtubule attachments (arrows).

Figure 2. Manifestations of chromothripsis in human tumors

(A) Sequencing of a chordoma (notochord-derived tumor) reveals 147 linkages between chromosomes 3q, 4q, 7q, 8p, and 9p, as well as various intrachromosomal rearrangements. Copy number profiles and allelic ratios are displayed in outer and inner rings of the circos plot. Adapted from Stephens et al (2011). (B) Spectral karyotyping of pancreatic adenocarcinoma. Enlargement shows a derivative chromosome containing at least six distinct regions from heterologous chromosomes, including 22, 1, 4, 10, and 14. Note that formation of this derivative has resulted in multiple monosomies, indicating widespread allelic and copy-number losses. Adapted from Stephens et al. (2011). (C) Chromothripsis in a Sonic hedgehog (Shh)-type medulloblastoma arising in a patient with Li-Fraumeni syndrome. The circos plot on the left shows a double minute generated through fragmentation of chromosome 2. Two major mediators of Shh signaling, MYCN and GLI2, are contained in this 5-megabase interval. On the right is shown a fluorescence in situ hybridization (FISH) experiment confirming ubiquitous amplification of both MYCN (red) and GLI2 (green) in the primary tumor. Adapted from Rausch et al. (2012).

Figure 3. Replication stress and mitotic errors may synergize to induce high levels of genomic instability and precipitate chromothripsis

(A) Untransformed cells pass through mitosis normally, resulting in correct partitioning of sister genomes in daughter cell nuclei. Once the restriction point is passed, orderly activation of replication origins leads to faithful S-phase progression and genome duplication with minimal loads of incidental DSBs. (B) Under the influence of activated oncogenes, G1/S-phase regulation is corrupted, leading to inappropriate firing of replication origins, depletion of nucleotide precursors, and elevated rates of fork stalling and collapse. Such replication stress may be particularly severe in hard-to-duplicate regions of the genome that harbor low densities of replication origins (i.e., ‘fragile sites’) and thus demand high fork processivity. (C) Replication stress can be induced by antecedent mitotic errors. Whole-chromosome segregation errors frequently result in the formation post-mitotic micronuclei and also trigger a p53-dependent arrest in late G1 phase (Uetake and Sluder, 2010). However, if p53 is mutated or experimentally inactivated, micronucleated cells can proceed into S-phase, resulting in delayed and breakage-prone DNA synthesis within the micronucleus. In addition, persistent replication intermediates within the micronucleus can be further destabilized by transit into mitosis, resulting in haphazard chromatin compaction and ‘pulverization’. A similar fate can also befall lagging anaphase chromatids generated through other mechanisms (for example, telomere erosion and end-to-end fusion, resulting in a dicentric chromosome).

Figure 4. Error-prone modes of DNA repair may underlie chromothripsis

(A) Non-replicative end joining. Intense but localized induction of DSBs may liberate genomic fragments that are recognized, processed, and ligated by one of several end-joining pathways that are intrinsically error-prone (see figure for mechanistic details). Fragments not ligated to the founding chromosome ‘stub’ either self-ligate to form extrachromosomal arrays (double-minutes) or are lost from the genome entirely. (B) Mutagenic fork restart. Under conditions where HR-dependent pathways are compromised or limiting, a collapsed replication fork may be restarted via microhomology-mediated template switching. This error-prone mode of lesion bypass can generate the full spectrum of structural alterations, including deletions, amplifications, inversions, and non-reciprocal translocations (see figure for mechanistic details).

Figure 5. Mechanisms and signatures of mutagenic chromosome reassembly

(A) Non-replicative end-joining pathways. Under NHEJ, DSB ends are first recognized and protected from resection by the Ku70–Ku80 heterodimer. Subsequent interactions between DSB ends occur via short (0–5 bp) patches of microhomology. One to four nucleotides are then added or removed to create a ligatable end, resulting in an imprecise junction or “information scar” at the breakpoint. In contrast, MMEJ uses Ku-independent strand resection to expose longer (5–25 bp) microhomology tracts. After annealing of these tracts, the resulting 3′ flaps are trimmed off prior to fill-in synthesis and ligation. Consequently, the final repair products generated by MMEJ typically contain variably sized deletions as well as nucleotide insertions. (B) MMBIR is initiated when a replication fork collapses, resulting in a single-ended DSB (black lines; for clarity the nicked or gapped duplex representing the opposite side of the collapsed fork has been omitted). Thereafter the 5′ end of the broken arm is resected to reveal a free 3′ ssDNA tail, which then uses short-tract microhomology (2–5 nucleotides) to anneal with other exposed ssDNAs, such as the lagging strand of an upstream replication fork (red lines). Elongation of this primer-template junction occurs with low processivity, resulting in eventual fork disassembly, release of the extension products, and further cycles of template switching. While a switch back to the original sister chromatid can re-establish processive DNA replication (resulting in a segmental duplication or higher-order amplification), interchromosomal switching can also occur, resulting in a complex non-reciprocal translocation.