Genome Rearrangements Associated with Aberrant Telomere Maintenance

Abstract

There is unequivocal evidence that telomeres are critical for cellular homeostasis and that telomere dysfunction can elicit genome instability and potentially initiate events that culminate in cancer. Mounting evidence points to telomeres having a critical role in driving local and systemic structural rearrangements that drive cancer. These include the classical “breakage-fusion-bridg” (BFB) cycles and more recently identified genome re-shaping events like kataegis and chromothripsis. In this brief review, we outline the established and most recent advances describing the roles that telomere dysfunction has in the origin of these catastrophic genome rearrangements. We discuss how local and systemic structural rearrangements enable telomere length maintenance, by either telomerase or the alternative lengthening of telomeres, that is essential to sustain cancer cell proliferation.

A Brief Re(cap) of Telomeres

Telomeres are nucleoprotein structures found at the ends of linear chromosomes that are critical for the maintenance of chromosomal integrity. Due to the “end replication problem”, telomeric repeats are irrevocably lost with every cell division until such time as their capacity to protect chromosome ends becomes compromised, which triggers replicative senescence [5]. Bypass of senescence by the inactivation of cell cycle checkpoint proteins such as p53 and RB, can potentiate the effects of telomere dysfunction. Most somatic cells succumb to telomere dysfunction-induced genome instability during cell division, i.e., mitotic catastrophe, or by activating programmed cell death mechanisms such as autophagy [6,7] (Figure 1), unless they initiate a telomere maintenance mechanism [8]. In this review, we aim to discuss the repercussions of telomere-induced genome instability, mechanisms of telomerase activation in cancer, and the current status of telomerase-targeting therapies.

Figure 1. Consequences of Telomere Shortening.

Mammalian cells harbor linear chromosomes that possess specialized ribonucleoprotein structures called telomeres (shown in yellow) to protect their ends. Due to the “end replication problem”, telomeres undergo progressive shortening during each cycle of cell division. Once the telomeres become “uncapped”, they begin to elicit a DNA damage response. To protect from telomere dysfunction-induced genome instability, cells can activate the p53/RB-mediated cell cycle checkpoint. The activation of these checkpoints can induce either cellular senescence or cell death via apoptosis. However, inactivation or bypass of the p53/RB-mediated checkpoints can lead to the onset of telomere dysfunction-mediated genome instability, which can lead to the formation of chromosomal rearrangements. Continued cycles of cell division with the burden of chromosomal rearrangements can lead to cell death due to errors during mitosis or autophagy. Alternatively, activation of telomere maintenance mechanisms such as telomerase or the Alternative lengthening of telomeres (ALT) pathway that can restore the “uncapped” telomere, can confer immortality, and lead to the onset of tumorigenesis.

Effect of Telomere Dysfunction on Genome Integrity

Recognition of uncapped telomeres by the ATM kinase activates the end joining pathway that can generate end-to-end fusions between non-homologous chromosomes or sister chromatids giving rise to dicentric chromosomes [9]. Dicentric chromosomes are highly unstable during mitosis since they contain two centromeres that each attach to both spindle poles, which results in the chromosome being dragged in opposing directions during anaphase at both centromeres. The torsional force exerted on these chromosomes during anaphase can lead to random breakage of the chromosome at non-centromeric positions [10]. The resultant fragments of the chromosome can be inherited by the daughter cells in an asymmetric manner. However, in the absence of telomerase reactivation, it is feasible that the newly formed daughter cell continues to harbor a chromosome with an uncapped telomere. Such a chromosome can once again form a dicentric chromosome and prone to breakage during mitosis. This concept of multiple “Breakage-fusion-bridge” (BFB) cycles was originally proposed by Dr. Barbara McClintock [11] and these cycles can give rise to several outcomes such as loss of heterozygosity, non-reciprocal translocations [12], whole chromosome aneuploidy [13], and gene amplification with inverted repeats [14,15]. For example, loss of heterozygosity can occur if the BFB cycle is accompanied with asymmetric segregation of chromosomal fragments. Specifically, if a dicentric chromosome randomly breaks during mitosis, and one of the daughter cells inherits both copies of one allele and the other daughter cell inherits none, then it can lead to loss of heterozygosity in one of the daughter cells. On the other hand, a dicentric chromosome formed by fusion of sister chromatids after DNA replication can create gene duplication events where the two genes are in inverted orientation. If a telomere is acquired at this stage, then the derivative chromosome harbors a gene duplication event with inverted repeats. However, if such a chromosome continues to undergo BFB cycles in the absence of a telomere, then such cycles can result in a gene amplification event with inverted repeats. Finally, it is also possible that the breakage of the dicentric chromosomes may lead to the formation of DNA double strand break ends that are able to invade a homologous sequence on another chromosome. Such a unidirectional recombination event, probably orchestrated by the break-induced replication pathway, can give rise to a non-reciprocal translocation [16] (Figure 2). If the donor chromosome has an intact telomere, then such a recombination event can give rise to a translocation with the acquisition of a telomere, which can stop the BFB cycles. However, the gain of a telomere during the formation of a non-reciprocal translocation can lead to a concomitant loss of the telomere from the donor chromosome, which can subject the donor chromosome to enter BFB cycles by the donor and thereby potentiate genome instability [17]. Such features of BFB cycles are often observed in several types of cancers such as breast cancer, esophageal cancer, osteosarcomas, renal cancers, pancreatic cancer and more. Similar outcomes were also observed in a recent study that used a novel system to reproducibly induce a sister chromatid fusion on the X chromosome and monitor its fate (Kagaya K et al 2019, bioRxiv: doi.org/10.1101/607341). Additionally, BFB cycles have also been determined as the underlying cause of certain recurrent chromosomal aberrations seen in cancer such as the 11q3 amplification [18] and the 17q deletion [19]. However, it is currently unclear why certain chromosomes or genomic loci would be prone to BFB cycles and if this correlates with telomere length of a particular chromosome, proximity of the locus to fragile sites, etc.

Figure 2. Telomere Dysfunction Can Induce Breakage-Fusion-Bridge Cycles and Generate Chromosomal Rearrangements.

Shown here are the different rearrangement outcomes that may arise due to breakage-fusion-bridge cycles post telomere crisis. If the chromosome harboring an uncapped telomere undergoes DNA replication, then the result is the generation of sister chromatids that each have an uncapped telomere end. These uncapped telomere ends are recognized as DNA double strands “breaks” and can be “fused” together creating a dicentric chromosome. During mitosis, the spindles can attach to both centromeres and the dicentric chromosome can be pulled in opposite directions forming an anaphase/chromatin “bridge”. Such force renders the dicentric chromosome prone to breakage at random positions along its length. Such breakage can be accompanied with asymmetric inheritance of the chromosome fragments by the daughter cells. If the daughter cell acquires a portion of one chromosome and activates telomerase, it can lead to telomere acquisition by the derivative chromosome. However, due to the loss of genetic information caused by the breakage-fusion-bridge cycle, the resultant event involves loss of heterozygosity. In contrast, the daughter cell can also resect a telomeric end to reveal homologous sequences to another chromosome. Repair of such ends by the break-induced replication pathway can be a mechanism of telomere acquisition, but also generate a non-reciprocal translocation, where genetic information is gained by only the derivative chromosome and the donor chromosome remains intact. Additionally, if the daughter cells do not activate telomerase, then the cell may continue to harbor a chromosome with an uncapped telomere. Subsequent replication result in the generation of another sister chromatid fusion and a dicentric chromosome. Such a chromosome can once again enter the breakage-fusion-bridge cycle. If a chromosome is broken at similar positions during several cycles of BFB, then the derivative chromosome can harbor gene amplifications, but with inverted repeats.

Dicentric chromosomes can persist through mitosis and give rise to long chromatin bridges between daughter cells in telophase. Resolution of such DNA bridges by cytoplasmic nucleases, such as TREX1, allows for reentry of the DNA into the nucleus [20]. However, the attack on chromatin bridges by nucleases can also shear the DNA into hundreds to pieces, a process known as chromothripsis [21]. Chromothripsis is characterized by the presence of a derivative chromosome with a localized region that contains randomly reassembled fragments of the chromosome along with gain and loss of chromosomal sequence [22]. Junctions from chromosomes that have undergone chromothripsis show signatures of repair by either the canonical non-homologous end joining (NHEJ) or the micro-homology mediated end-joining pathway of DSB repair, underscoring their role in ligating the broken fragments of the chromosome post chromatin bridge fragmentation [20,23–26]. Furthermore, resolution of chromatin bridges by nucleases can also give rise to stretches of single stranded DNA, which can serve as substrates for cytidine deaminases, such as APOBEC3B [20,27]. Evidence of such illicit APOBEC activity is often associated with high density of single point mutations, usually within a few kb of the rearrangement junctions associated with chromothripsis and is termed kataegis [27] (Figure 3).

Figure 3. Telomere Dysfunction Can Lead to Chromothripsis and Kataegis.

If a cell harbors uncapped telomeres on two distinct non-homologous chromosomes, then these two chromosomes can be fused at their telomeres i.e., an end-to-end fusion. This can lead to the formation of a dicentric chromosome that forms a chromatin bridge during mitosis. However, if such a bridge persists beyond mitosis, then it can lead to the formation of daughter cells that are connected to each other by this chromatin bridge. During this process, the nuclear envelope can rupture and the chromatin bridge can become susceptible to attack by cytoplasmic nucleases, such a TREX1, which can lead to the fragmentation of the chromatin bridge. These fragments are randomly inherited by the daughter cells and are prone to ligation in a random manner, both with respect to order and orientation. Furthermore, fragmentation of the chromatin bridge by cytoplasmic nucleases can lead to the generation of stretches of single stranded DNA, which can serve as substrates for cytidine deaminases leading to the presence of clustered point mutations (shown as red and green circles), mostly in vicinity of the rearrangement junctions.

Telomerase dysfunction, specifically telomere-driven crisis, can also give rise to whole genome duplication events or tetraploidy. A study that used cells deficient in p53 and critical members of the Shelterin pathway found that telomere dysfunction can initiate an ATM/ATR kinase-dependent block from entering mitosis, which leads to sustained arrest in G2 after which cells undergo whole-genome reduplication leading to the formation of tetraploid cells (davoli T et al 2010). Furthermore, analysis of telomere crisis-driven tetraploid mouse cells showed that these cells are less stable than their diploid counterparts and demonstrate a higher propensity for chromosome loss during cell division [13,28].Thus, telomere dysfunction can give rise to several types of genomic rearrangements and instability observed in cancer cells. A key aspect of telomere dysfunction induced BFB cycles is that they can persist if the telomeres remain unprotected, but theoretically, restoration of the telomere by telomerase activation should stop this reaction [14]. One of the mechanisms that can help gain telomeric sequence is the activation of telomerase to extend the telomeres, to continue cell division with the faithful passing of the genome.

Telomerase Activation in cancer: micro and macro rearrangements

Telomere lengthening is mediated by an evolutionarily conserved telomerase complex that counteracts the “end-replication problem” by direct addition of TTAGGG repeats to the ends of telomeres. The reverse transcriptase activity of telomerase is encoded by the TERT gene. TERT protein associates with a long non-coding RNA called the human telomerase RNA (hTR), which is encoded by the TERC gene, and dyskerin protein that is involved in the biogenesis and stabilization of the core ribonucleoprotein complex. Telomerase activity appears to be detectable only in germ cells, stem cells, and immortalized human cell lines. Somatic cells lack telomerase either due to repression of TERT expression by epigenetic mechanisms, altered abundance of transcription factors associated with the TERT promoter region or limiting endogenous TERT protein levels necessary to catalyze telomere sequence addition. In accordance dance with this, elevated TERT expression, stimulated by gain-of-function promoter mutations, gene amplifications (i.e. gain in TERT gene copy number) and structural rearrangements proximal to the TERT gene, is a defining feature of the majority of cancer cells [29].

While investigating the presence of recurrent mutations associated with familial and sporadic melanoma patients, two independent studies identified recurring point mutations in the core promoter region of the TERT gene, each of which were posited to create a de novo transcription factor binding sites in the TERT promoter, and thereby increase TERT expression. Introduction of these recurrent mutations into the TERT promoter in a human embryonic stem cells (hESCs) line allowed for delayed repression of TERT and an increase in telomere length in differentiated fibroblasts and neurons. However, long term experiments with these edited cells indicate that TERT promoter mutations are insufficient to stabilize telomeres, and that telomeres can continue to shorten below the “Hayflick Limit” and induce crisis. Thus, these findings indicate that, similar to the onset of tumorigenesis, telomerase needs to be reactivated by mechanisms beyond TERT promoter mutations for such cells to proliferate post crisis [33].

Telomerase Activation: Lessons from Neuroblastoma

Over recent years, new insights of how telomere preservation contributes to cancer evolution have emerged from clinical observations of the molecular pathogenesis of a pediatric cancer, neuroblastoma. This solid tumor originates from immature cells of the sympathetic nervous system and has a remarkably broad spectrum of biological and clinical behaviors, ranging from spontaneous regression to fatal progression. While various mechanisms have been proposed to explain the divergent courses of disease, including activation of telomerase [34], the molecular basis of these distinct phenotypes has remained poorly understood until recently. In 2015, two whole-genome sequencing studies independently reported on the occurrence of genomic rearrangements affecting a chromosomal region at 5p15.33 in neuroblastoma [35,36]. While the types of these structural alterations were diverse, involving a broad range of translocation partners and distinct types of rearrangements, the breakpoints almost all clustered in a 50 kb region upstream of the TERT transcriptional start site without affecting the promoter. All tumors bearing TERT rearrangements harbored high TERT expression levels due to translocation of strong enhancer elements in close proximity to the TERT locus [35]. Interestingly, TERT rearrangements occurred almost exclusively in clinical high-risk neuroblastoma patients and were associated with poor patient outcome [35,36]. In addition, TERT rearrangements were found in mutually exclusive fashion with other mutations occurring in high-risk neuroblastoma, such as amplification of MYCN or inactivating mutations of ATRX, suggesting that these alterations converge on a common molecular mechanism. Indeed, MYCN is a direct transcriptional activator of TERT [37], and MYCN amplification was associated with elevated TERT expression in neuroblastoma [35], whereas inactivation of ATRX is strongly correlated with the Alternative Lengthening of Telomeres (ALT) pathway (see below). By contrast, no TERT promoter mutations were detected in neuroblastoma [35,38]. Furthermore, low-risk neuroblastomas characteristically lacked genomic alterations of TERT, MYCN or ATRX, suggesting that telomere maintenance mechanisms are absent in these favorable tumors [35].

The hypothesis that telomere maintenance is a hallmark of high-risk neuroblastoma, discriminating this subtype from low-risk tumors, was systematically evaluated in larger tumor cohorts recently ([39] & Roderwieser et al., JCO doi.org/10.1200/PO.19.00072). Assessment of telomere maintenance mechanisms as well as mutations in genes of the RAS and the p53 pathway in 208 primary neuroblastomas demonstrated that the divergent clinical courses in this malignancy are indeed primarily driven by the presence of telomere maintenance mechanisms [39]. The absence of such mechanisms was compatible with spontaneous regression, independent of the occurrence of mutations in RAs or p53 pathway genes, and patients whose tumors lacked telomere maintenance had excellent outcome. By contrast, patients whose tumors harbored activated telomerase or ALT had unfavorable clinical courses, and additional mutations in RAS or p53 pathway genes substantially increased the risk of death from disease. Moreover, the clinical course of neuroblastoma also appears to depend on the type of telomere maintenance. In a cohort of 223 neuroblastoma patients, event-free survival was similarly poor in patients whose tumors were ALT- or telomerase-positive, however, overall survival was significantly worse in the latter subgroup. Together, these data strongly suggest that activation of telomere maintenance mechanisms in neuroblastoma is a crucial step in transformation to a fully malignant phenotype, whereas lack of telomere maintenance, and hence cellular immortalization, is probably a prerequisite for spontaneous regression. These findings are well in line with observations in experimental systems, in which telomerase was indispensable for full malignant transformation of human cells bearing oncogenic HRAS [40].

Genomic rearrangements of the TERT locus leading to increased TERT expression have not only been identified in neuroblastoma, but also in other cancer types, such as B-cell neoplasms [41], chromophobe renal cell carcinoma [42] and glioblastoma [43]. A recent pan-cancer survey across 31 cancer types identified structural variants involving the TERT locus in 7 % of the cases [44]. Similar to neuroblastoma, structural TERT variants were associated with strong induction of TERT expression, suggesting that such alterations regularly contribute to telomere maintenance and immortalization in cancer. These findings were corroborated by another pan-cancer study considering 18 cancer types. Importantly, elevated TERT expression was not obviously driven by copy number gains, which occasionally co-occur with TERT rearrangements, but by changes in the chromatin context of the locus following disruption of topologically associating domains (TaDs) [45] (Figure 4). Furthermore, genomic rearrangements can also lead to the placement of super enhancers upstream of the TERT locus, which can also facilitate changes in chromatin and influence gene expression [35,36]. These observations are fully consistent with data from neuroblastoma, in which neither copy numbers nor the distance of the breakpoint from the transcriptional start site were associated with TERT expression in cases bearing TERT rearrangements [35]. Together, genomic TERT rearrangements clearly add to the spectrum of alterations causing up-regulation of TERT in cancer, however, the mechanisms leading to induction of telomerase has remained uncertain in a substantial fraction of TERT-expressing malignancies.

Figure 4. Mechanisms of TERT activation.

Shown here are some of the mechanisms that can contribute to TERT activation. The TERT promoter has been shown to harbor recurrent point mutations within 150bp upstream of the transcription start site. Such point mutations have been shown to generate de novo binding sites for transcription factors that can promote TERT expression. Furthermore, recent studies have shown the presence of structural variates with breakpoints approximately 10–50kb upstream of the TERT locus. Such structural variations can lead to increased TERT expression by either disrupting existing Topologically Associating Domains (TADs) or by juxtaposing new enhancers upstream of the transcription start site (i.e., enhancer hijacking).

Alternative Lengthening of Telomeres (ALT)

A subset of tumors, approximately 10–15%, maintain telomere length independently of telomerase andinstead use a homologous-based mechanism called the Alternative Lengthening of Telomeres (ALT) pathway. In those cancer cells, telomeres rely on multiple, potentially redundant homology directed repair mechanisms including homologous recombination and templated DNA synthesis during G2 and M cell-cycle phases that resembles break-induced replication (BIR) [46]. The activation of ALT is strongly correlated with loss of function missense mutations in the alpha thalassemia/mental retardation syndrome X-linked (ATRX) and Death domain associated X-linked (DAXX) genes [47]. Additional ALT associated mutations have been identified in histones H3.1 (HIST1H3A) and H3.3 (H3F3A) [ 48], as well as in the isocitrate dehydrogenase (IDH1–2) genes [49] whose protein product regulates the metabolism of cofactors utilized by histone and DNA methyltransferase enzymes. Thus, the convergence between perturbed chromatin regulation and ALT appears significant. More recently, long sought-after evidence of deregulated homologous recombination mechanisms in ALT have emerged with the identification of loss of function and expression of key HR factors like SMARCAL1 [50] or SLX4IP [51], or the specific temporal sequestration of HR factors like HOP2, RAD52 or the DNA Polymerase Delta (Polδ) complex to telomeres (Reviewed in [52]).

ALT-positive cells exhibit new copy number gains and losses, with regions of recurrent deletions and amplifications, indicative of an unstable genome [53]. For instance, ~60% of leiomyosarcomas rely on the ALT telomere elongation mechanism [54]. These tumors also exhibit MDM2 amplifications that appear to be independent of p53 mutations suggesting a role for MDM2 in the bypass of senescence. More concrete evidence of telomere driven genome restructuring in ALT cancer cells stems from the detection of non-canonical repeat sequences embedded within tracts of canonical telomeric sequences [55,56]. These sequences include TCAGGG (C-type) variants which were subsequently shown to present de novo sequence specific binding sites for the NR2C/F class (also known as TR4 and COUP-TFII, respectively) of nuclear orphan receptors (NORs) within telomeres [56,57]. How the variant repeats are formed is unknown. It was speculated that they could represent remnants of inaccurate nucleotide incorporation by TERT earlier in the ancestral lineage of the cancer cell-type, prior to ALT activation [55]. Similarly, it was suggested that the discrepant repair of oxidized nucleotides at telomeres could yield nucleotide changes that escape proof-reading by polymerases or subsequent replacement, thereby altering the canonical TTAGGG sequence [55]. The intrinsic inaccuracy of DNA synthesis during BIR [58] could conceivably give rise to these variant repeats, although this possibility remains to be determined.

NORs are typically associated with transcriptional control where they promote inter-chromosomal contacts between transcriptional regulatory domains i.e. promoter and enhancers. In cancer cells, androgen receptors (ARs) have been linked with inducing chromosomal translocations [59]. Accordingly, it seems that the variant TCAGGG repeats enables telomere tethering of NR2C/F to establish new inter/intra chromosomal contacts between telomeric and chromosomal binding sites [56]. These sites also appear to function as hotspots for homologous recombination and can lead to the insertion of several kilobase pairs of telomeric repeat sequences and binding of Shelterin at new regions within the chromosome. This finding presented a new model for how the altered sequence content and predisposition to recombine of ALT telomeres promotes genome instability by causing chromosome breakage during mitosis that may lead to rearrangements. The specific functional association of the NR2C/F class of NORs with ALT telomeres and their amenability to pharmacological inhibition implies that targeting these factors might represent a novel avenue for anti-ALT therapy to pursue.

Concluding Remarks

During their evolution cancer cells are exposed and adapt to a myriad of events that eventually activate diverse telomere length maintenance mechanisms. An important future challenge will be to delineate the series that give rise to the specific telomere elongation pathways discussed in this review. While efforts to model the events subsequent to TPM acquisition are making important progress, the causes and immediate events subsequent to the structural rearrangements that drive neuroblastoma, for instance, remain unclear. For instance, what events, factors and mechanisms determine whether TERT is reactivated through TPMs or gene proximal structural rearrangements? Is a specific type of intrinsic/extrinsic genotoxic insult necessary and are they specific to a particular time-window or cell-of-origin. With respect to ALT, there is clearly much work to be done. It is unknown but possible that ALT cancer cells harbor particular signatures of genomic rearrangements that are specifically derived from the combined effects of ATRX-DAXX mutation and BIR. For instance, given the convergence of deregulated chromatin and potential aberrations in DNA and histone methylation, it might be likely that ALT cells harbor unique epigenomic signatures that again await determination.

With the recent advances in elucidating the cryo-EM structure and the molecular biology of telomerase function, as well as a host of new anti-ALT targets emerging, there is renewed hope that targeting telomere elongation mechanisms could once again become the Achilles heel of cancer. Should this transpire, transferring the knowledge gained to diagnostics, risk stratification and therapy design could be of immense significance in improving treatment of devastating malignancies that for too long have afflicted cancer patients.