Chromothripsis and telomere crisis: engines of genome instability

Abstract

In the early stages of carcinogenesis many cells confront two key suppressive checkpoints, senescence and telomere crisis. Crisis is characterized by massive chromosomal instability and cell death. The genetic instability initiated during crisis leaves detectable scars on cancer genomes, the full scope of which is only just beginning to be appreciated. In particular, the dramatic genome reshuffling phenomenon chromothripsis has been mechanistically linked to the resolution of DNA bridges formed by dicentric chromosomes, and by the shattering of DNA inside micronuclei. Furthermore, an intriguing connection to innate immune signaling has begun to position telomere crisis as a crucial stage not only in the evolution of the cancer genome, but also in the interaction between the genome and the immune system.

Mammalian cells have a finite lifespan that is in part dictated by the length of their telomeres, specialized nucleoprotein structures at the end of chromosomes. During normal semi-conservative replication, the repetitive TTAGGG telomeric sequence is shortened with each cell division until a DNA damage response, elicited by the inability of the shelterin complex to adequately protect telomere ends, results in cell cycle arrest and senescence [1, 2]. This checkpoint acts as an important tumor suppressor mechanism. Cells in the germline express telomerase, the reverse transcriptase that can maintain telomere length by addition of TTAGGG repeats to the chromosome ends. In the case of malignant cells which often have disrupted cell cycle checkpoints, this senescence barrier fails to be activated by the de-protection of telomere ends [3, 4]. Further rounds of cell division continue to shorten telomeres until the cells enter a period called telomere crisis. The presence of multiple un-capped chromosome ends results in chromosome fusions, high levels of genetic instability and cell death [5, 6]. In order to escape crisis, there is a strong selective pressure for tumors to engage a telomere maintenance mechanism to re-stabilize their genomes. In the majority of cancers this is achieved through telomerase activation, but can also occur through an alternative recombination-based mechanism known as ALT (alternative lengthening of telomeres) [7, 8].

Genomic scars of telomere crisis

Telomere crisis is thought to occur during the early stages of cancer development, and evidence from diverse cancer types, such as breast adenocarcinoma, colorectal cancer and chronic lymphocytic leukemia supports this hypothesis [9–12].The prevalence of telomere crisis across all cancers is not known, but recent bioinformatic analyses of telomere length across the TCGA data set show a majority of tumors with shorter telomeres than corresponding normal cells, suggesting they have undergone telomere shortening [13].The genetic instability that can arise as a result of critically short de-protected telomere ends is myriad. In mice with defective p53 pathways, telomere fusions result in aneuploidy, loss of heterozygosity (LOH) and translocations in the resulting tumors [14]. In depth analysis of individual telomere fusions during crisis have demonstrated that deprotected telomeres can fuse with a wide range of non-telomeric loci, resulting in complex translocations [15]. Recent work has shown that dicentric chromosome resolution can lead to the genome shattering phenomenon chromothripsis (See Box 1) [16–18].

Box 1: What is chromothripsis?

Chromothripsis refers to a localised chromosome shattering and re-joining event that results in a highly re-arranged area of the genome, usually affecting just one or a few chromosomes [16]. A key characteristic is that chromothripsis is an ‘all-at-once’ event, meaning that simulations of copy number change over multiple generations fail to recreate the observed genotype. Several other criteria have been developed to in order to identify chromothripsis rearrangements [17].

• breakpoints are clustered together, usually surrounded by areas of normal copy number

• only two (or three) copy number states are detected in the chromothriptic region

• a derivative chromosome can be formed unambiguously from the fragmented region

• genome fragments in the derivative chromosome have a random order and orientation

• alternating patterns of retention and loss of heterozygosity are observed

Chromothripsis could contribute to tumorigenesis in a variety of ways. Rearranged fragments could result in oncogenic fusions that drive tumor growth, or deleted regions could contain tumor suppressor genes. Additionally, the formation of double-minute (DM) chromosomes containing oncogenes which are then amplified can occur [19, 20]. Initially chromothripsis was thought to be a rare phenomenon, but high-resolution sequencing has continued to uncover instances of highly complex rearrangements across diverse cancer types, albeit at varying frequencies (for a review see [21]). Interestingly, chromothripsis occurs more frequently on some chromosomes in particular cancer types [22]. Whether this is the result of positive selection for beneficial rearrangements (such as oncogene fusions or amplifications), or negative selection against deleterious rearrangements remains to be seen. However, a large-scale analysis across 39 different cancer types identified focal amplifications of oncogenes (such as CCND1, MDM2 and MYC) coinciding with chromothripsis events, and chromothripsis also underlying significant proportions of tumor suppressor and DNA repair gene loss, strongly suggesting that chromothripsis events can be selectively advantageous [23].

How does chromothripsis happen in telomere crisis?

During crisis chromosomes form end-to-end fusions, as multiple de-protected telomeres are recognized as DNA breaks and fused together by DNA repair mechanisms. These dicentric chromosomes cannot be segregated properly at anaphase and result in the formation of DNA bridges. McClintock hypothesized that these bridges would be broken and re-joined in a cyclical manner across multiple cell divisions, in a process referred to as breakage-fusion-bridge (BFB) cycles [24]. These BFB cycles leave characteristic patterns of inverted repeat sequences at the ends of chromosomes that have been observed in multiple cancer types [20, 25, 26]. However, recent work that directly followed the fate of dicentric chromosomes with time-lapse imaging revealed that dicentric chromosomes formed through telomere fusion rarely break in anaphase, instead persisting as extended DNA bridges for many hours into the next G1 phase [18]. Surprisingly the resolution of these bridges was correlated with the loss of nuclear envelope integrity markers, and the formation of ssDNA on the bridge. The cytoplasmic 3’ exonuclease TREX1, which is usually involved in degradation of DNA species in the cytoplasm (Figure 1), was responsible for timely resolution of the DNA bridges, presumably inappropriately gaining access to the bridge DNA after nuclear envelope rupture [18]. High coverage whole genome sequencing of cells that had been subject to dicentric chromosome formation and bridge resolution revealed common chromothripsis events, suggesting DNA bridges in telomere crisis may be an important source of the chromothripsis observed across many different cancer genomes. In this study chromothripsis was frequently accompanied by kataegis, localized hypermutation involving strand coordinated C>T and C>G transversions [18, 27, 28]. Kataegis is thought to be caused by the activity of the APOBEC cytidine deaminase family of enzymes that act on ssDNA and usually play a role in innate immunity [27–29].

Figure 1: Summary of the cGAS/STING response to cytosolic dsDNA.

Aberrant dsDNA within the cytoplasm can be generated by a variety of extracellular and intracellular sources, where it activates cGAS. This leads to cGAMP production, and subsequent STING activation. STING is located in the ER membrane, but upon activation translocates through the ERGIC and Golgi. STING can then interact with TBK (facilitated by palmitoylation of STING), which results in the phosphorylation of STING. STING phosphorylation results in IRF3 recruitment and phosphorylation by TBK1. Phosphorylated IRF3 can then translocate to the nucleus and induce activation of IFN genes. A major regulator of this signaling pathway is the ER-localized nuclease TREX1, which can degrade cytosolic DNA.

To directly test the hypothesis that TREX1 results in chromothripsis through resolution of DNA bridges, the same authors repeated their experiments in TREX1 null cells [30]. The frequency of complex chromothripsis events after dicentric chromosome formation and resolution was reduced in TREX-1 knock-out clones (from 21% to 8%), and complex events had fewer copy number changes [30]. These data suggest that TREX1 does indeed promote the formation of complex rearrangements, likely through the formation of extensive ssDNA on the bridges [30]. Some key mechanistic questions remain. Firstly, since TREX1 is an exonuclease, it requires nicks or gaps in the DNA to begin resection. These may form spontaneously, or there may be another enzyme involved in their formation at bridges. Secondly, without TREX1 present, bridges are still eventually resolved through an unknown mechanism. This mechanism could involve actomyosin based contractility, which can influence bridge resolution [31]. Presumably this unknown mechanism does not result in the formation of extensive ssDNA fragments, since TREX1 null clones were characterized by simpler rearrangements, mostly BFBs. The formation of the rearrangements defined as ‘local-jumps’, which are prevalent in TREX1 null cells remains unclear [30]. A distinct replicative mechanism has also been proposed for the formation of complex chained rearrangements arising from telomere crisis in systems with defective end-joining pathways [32]. This process was inferred to involve mutagenic repair of deprotected telomeres by replicative mechanisms involving microhomology, and the observed events may be more analogous to chromoplexy, events of linked translocations and deletions first observed in prostate cancer [32, 33]. Since the characteristic inverted repeat rearrangements from BFB cycles are prevalent in cancer genomes, understanding the different mechanisms that lead to their formation remains a priority. Are there particular situations that promote bridge resolution through TREX1 mediated degradation over simple bridge resolution and BFB formation? What factors are involved in the repair of each type of bridge resolution mechanism and are they distinct? How do events involving defective or dysregulated replication arise during crisis? Direct sequencing of cells that have passed through telomere crisis without any experimental intervention may shed light on the frequency of different genomic outcomes.

Chromothripsis in micronuclei

Chromosome missegregation can lead to the formation of micronuclei, a well-established hallmark of malignancy. Due to their spatial separation from the main nucleus, micronuclei are an attractive candidate to explain the localized nature of chromothripsis. Indeed, DNA within micronuclei exhibits extensive DNA damage [34, 35]. This has been proposed to derive from a multitude of factors including defective DNA repair and replication [36], inappropriate chromosome condensation during mitotic entry [34, 37] and defective nuclear envelope formation [38, 39]. Direct evidence demonstrating chromothripsis in micronuclei was provided by the Pellman laboratory who sequenced single cells after chromosome missegregation into a micronucleus [40]. Micronuclei are likely formed during telomere crisis as a result of the extensive genome instability that leads to acentric chromosome fragments that are subject to missegregation [41]. The patterns of rearrangements derived from a chromosome bridge resolution event compared to micronuclei driven chromothripsis may be distinct. Indeed, it has been suggested by the Pellman laboratory that within the first cell cycle after bridge resolution complex events are rare, but after further cell cycles involving defective replication of bridge DNA or subsequently formed micronuclei there is a higher prevalence of complex rearrangements with signatures indicative of a replicative origin [31]. Dicentric chromosome bridge resolution may result in distal rearrangements near telomeres, whereas micronuclei rupture and chromothripsis can affect the entire chromosome if it is contained within one micronucleus. Whether TREX1 plays a role in chromothripsis in micronuclei that undergo a loss of nuclear envelope integrity remains unclear. Since both bridges and micronuclei abound during crisis, they are likely non-mutually exclusive with respect to the generation of chromothripsis during this period.

Tetraploidization during crisis - a link to chromothripsis?

DNA damage signaling during telomere crisis has been shown to lead to the formation of tetraploid cells through a process of endoreduplication [42, 43]. Persistent ATM and/or ATR signaling at deprotected telomeres results in a G2 cell cycle arrest and the eventual bypass of mitosis, leaving cells in a G1-like state but with duplicated DNA content. Re-replication of DNA as cells erroneously enter another S phase results in tetraploid DNA content [42]. Tetraploidy can itself be tumorigenic [44], and tetraploid cells have an increased capacity to tolerate and propagate chromosomal instability [45]. Therefore, the formation of tetraploid cells during telomere crisis could represent a significant risk for further genome instability. Interestingly, in vitro systems have suggested that chromothripsis may be more common in hyperploid cells [46]. Furthermore, in bioinformatics analysis of multiple tumor types, polyploidy in tumors was associated with a 1.5X higher chance of chromothripsis [23]. However, since tetraploid cells are known to provide a permissive background for the development of genome instability [45], this may simply reflect the consequence of reduced negative selection for altered karyotypes in a polyploid background. The frequency of whole genome duplication across multiple cancer types is substantial [47, 48], but in which tumors this is linked to passage through telomere crisis is as-of-yet unknown.

Restricting the formation of complex events - a role for autophagy?

During telomere crisis there is a high rate of cell death, generally presumed to be due to the high levels of genomic instability, although some cells have been shown to undergo mitotic cell death [49]. Karlseder and colleagues showed that during telomere crisis, both epithelial and fibroblast cells displayed the hallmarks of active autophagy [50]. The disruption of autophagy pathways using shRNAs was able to delay the onset of crisis, and these cells showed evidence of increased DNA damage and gross chromosomal aberrations [50]. This was not due to a general effect of DNA damage on autophagy, since autophagy induction required telomere specific DNA damage. This suggests that autophagy plays a role in restricting the accumulation of genetic instability during crisis, and that it could act as a tumor suppressive mechanism [50]. Whether cancer cells can permanently forgo the requirement for telomere maintenance if they can evade autophagic death during crisis remains unclear. Interestingly the authors showed that activation of autophagy partially relied upon the formation of cytoplasmic DNA fragments such as bridges, micronuclei and cytoplasmic chromatin fragments, and their recognition by the cytosolic nucleic acid sensor cGAS (Figure 1, [51]). Indeed, disruption of cGAS or its downstream activator STING was sufficient to attenuate autophagy and delay crisis, despite the presence of cytoplasmic DNA species [50]. This suggests that cytosolic DNA sensing through the cGAS-STING pathway is important for the activation of autophagy in crisis. Recent work has directly shown that cGAS can induce autophagy through a conserved pathway [52], although interestingly this is uncoupled from interferon induction. It therefore remains to be tested whether telomere crisis induces a significant interferon response.

Innate immune signaling and genome instability

Several recent papers have begun to shed light on the connection between innate immune signaling and genetic instability. Mackenzie et al., showed that cGAS plays a key role in the activation of innate immune signaling after DNA damage generated micronuclei [53]. Loss of nuclear envelope integrity in micronuclei permitted cGAS localization and downstream interferon response signaling [53]. cGAS was also observed at DNA bridges [53]. Similarly, Harding et al., showed that double-strand break induced induced STAT1 activation was dependent on the passage through mitosis and the formation of micronuclei [54]. A central tenant of these experiments is the inappropriate exposure of nuclear DNA to the cytosol, which was also shown to occur after DNA bridge formation [18].

Together this work raises the possibility that telomere crisis may act as an evolutionary bottleneck not just for genomic instability, but also for innate immune signaling, Figure 2 [53, 55]. cGAS has been observed at chromatin bridges and micronuclei, which are both abundant during crisis. Presumably this could lead to downstream interferon signaling that has a tumor suppressive effect. It is therefore tempting to speculate that to escape crisis, cells must modulate the downstream response to this signaling. Whether this can be achieved simply by stabilizing the genome through activation of telomere maintenance, thereby reducing the genetic instability that leads to signaling, or whether additional specific immune modulatory mechanisms are at work remains to be seen. Many cancers do have alterations in cGAS/STING signaling pathways, although it is not yet clear how each alteration exactly affects the signaling axis [56]. Interestingly, ALT+ cancers, which can escape crisis but maintain high levels of genetic instability frequently have inactivation of the STING pathway, thereby dampening their interferon response to extrachromosomal telomeric repeat DNA which is abundant in ALT cells [57]. Clearly telomere crisis has the ability to drastically alter the genome of evolving tumors, however we are only just beginning to understand the intimate mechanisms responsible for both causing and restraining this instability.

Figure 2: Schematic overview of mechanisms driving genome instability in telomere crisis.

Recent work has uncovered many types of genomic instability that lead to exposure of DNA to the cytoplasm, for example micronuclei and DNA bridges. Cytosolic DNA exposure can lead to signaling through the cGAS/STING pathway. Autophagy has been shown to be a prominent feature of cells in crisis. The exact details of the mechanisms linking cGAS/STING pathway activation to autophagy induction have not yet been fully elucidated.