Whole-genome doubling in tissues and tumors

Abstract

The overwhelming majority of proliferating somatic human cells are diploid, and this genomic state is typically maintained across successive cell divisions. However, failures in cell division can induce a whole-genome doubling (WGD) event, in which diploid cells transition to a tetraploid state. While some WGDs are developmentally programmed to produce non-proliferative tetraploid cells with specific cellular functions, unscheduled WGDs can be catastrophic: erroneously arising tetraploid cells are ill-equipped to cope with their doubled cellular and chromosomal content and quickly become genomically unstable and tumorigenic. Deciphering the genetics that underlie the genesis, physiology, and evolution of WGD cells may therefore reveal therapeutic avenues to selectively eliminate pathologic WGD cells.

The yin and yang of whole-genome doubling

Proliferating somatic mammalian cells are predominantly diploid, possessing two copies of each chromosome. The complex and highly-regulated cell-cycle evolved to help ensure that the diploid state is maintained across successive generations in most proliferating cells [1]. Nevertheless, somatic diploid cells occasionally undergo what is referred to as a whole-genome doubling (WGD) event, in which the entirety of the genome is doubled to produce tetraploid cells with four copies of each chromosome [2–4]. While successive WGD events give rise to polyploid cells with further increasing ploidies (e.g. octoploid cells), this review will predominantly focus on tetraploid cells as a majority of malignant mammalian cells that undergo a WGD event do so only once [5–7].

Ascension from a diploid to a tetraploid state provides a wealth of biologic functionality to cells. Numerous cell types, from hepatocytes to cardiomyocytes, are developmentally programmed to undergo WGD to generate non-proliferative cells with physiological characteristics that better enable them to perform specialized cellular functions. However, non-programmed WGD, commonly resulting from catastrophic failures in cell division, generates proliferative tetraploid cells that are unprepared to handle the unscheduled doubling of their genomic and cellular content. These erroneously-arising whole-genome doubled cells rapidly become genomically unstable and promote the emergence of rare aneuploid clones imbued with the enhanced capacity to drive tumor growth [8, 9]. Consequently, WGD represents one of the most common genetic alterations in human cancer [5–7, 9, 10]. Cells that have undergone WGD events, either scheduled (i.e., developmentally programmed) or unscheduled (i.e., in error), will hereafter be referred to as WGD⁺ cells.

Herein, we review how WGD⁺ cells originate in native tissues and tumors and dissect the physiological consequences of WGD, with emphasis on how scaling defects in WGD⁺ cells can significantly impair normal cellular functions and induce genome instability. We highlight the unique cellular and genomic adaptations that proliferating WGD⁺ cells must acquire to sustain their proliferation, as well as discuss how these adaptations may give rise to ploidy-specific gene dependencies [7, 11]. Finally, we end with a discussion as to how WGD status may be therapeutically leveraged to selectively eliminate WGD⁺ tumor cells.

Origins and physiologic implications of developmentally programmed whole-genome doubling

Since the discovery of the tetraploid nature of hepatocytes over a century ago [12–14], it is now appreciated that several tissue types are comprised of WGD⁺ cells that arise from intentional developmental programming. WGD has been confirmed in tissues such as the heart [15, 16], liver [14, 17], pancreas [18], lung [19], bone marrow [20, 21], and breast [22] across multiple mammalian species (reviewed in [23–25]). This evolutionary conservation indicates the presence of tissue-specific selective pressures which favor traits inherent to WGD⁺ cells. But how is WGD developmentally programmed? And what benefits does WGD provide?

Four distinct mechanisms have evolved to generate WGD⁺ cells: cell fusion, endoreduplication (also referred to as endoreplication), endomitosis, and cytokinesis failure (Fig. 1, Box 1).

Figure 1: WGD in physiologic and oncogenic development.

Programmed WGD occurs through multiple mechanisms in normal developmental physiology. Some examples include: cell fusion in osteoclasts, cytokinesis failure in hepatocytes, endomitosis in megakaryocytes, and endoreduplication in trophoblasts. Tumor cells undergo WGD via similar mechanisms during their evolution as a result of errors, such as: viral-induced cell fusion, cytokinesis failure, mitotic slippage, and endoreduplication.

Box 1: Mechanisms of WGD.

Cell fusion occurs when two or more cells merge to produce a syncytium. Endoreduplication cell cycles are the process by which the genome is re-replicated in S phase without an intervening mitosis. Endomitosis occurs when a cell enters mitosis but fails to complete chromosome segregation and karyokinesis. Finally, cytokinesis failure arises when mitotic cells complete chromosome segregation but fail to split into two distinct daughter cells.

These mechanisms evolved to permit controlled access to the polyploid state as WGD confers cells with unique advantages not accessible to their diploid counterparts. For example, cell fusion represents a key element of myogenesis [26]. During development, multiple nascent myoblasts fuse to produce multinucleated myotubes and myofibrils [26, 27]. These syncytia represent critical units of skeletal musculature and are ultimately vital for the generation of force and locomotion [26, 28, 29]. Cell fusion is also a crucial process for osteoclastogenesis [30–32], where myeloid precursors differentiate into multinucleated mature osteoclasts that are more efficient than diploids at resorbing bone. Similarly, successive rounds of endomitosis produce WGD⁺ megakaryocytes that are capable of producing more platelets than their diploid equivalents [20, 33]. Lastly, cytokinesis failure is a crucial component of both cardiac [34] and hepatic [35] development. The vast majority of adult mammalian ventricular cardiomyocytes (CM) are tetraploid due to programmed cytokinesis failure or endoreduplication in the late fetal or early postnatal period [36, 37]. Akin to cardiac development, newborn hepatocytes are largely diploid but also rapidly transition to a binucleated tetraploid state. A multitude of hypotheses surrounding the purpose of polyploidization in cardiac and hepatic tissue have been proposed, but the extent to which WGD in these organs represents physiologic functionality or adaptive responses to pathologic stimuli remains debated [38]. A clear example of WGD acting as an adaptation in these tissues is seen with hypertension, where WGD⁺ cell prevalence is noted to rise in both cardiac and vascular musculature [16, 39, 40]. Indeed, the incidence of WGD appears to be coupled to the magnitude of pressure overload as the left ventricle harbors a higher incidence of WGD⁺ cardiomyocytes compared to the right [39]. WGD in hepatocytes has been extensively studied ([17, 35, 41], elegantly reviewed in [42]) and mechanistically centers around programmed failure of cytokinesis through downregulation of genes essential to the localization of active RhoA at the cleavage furrow [43]. Recent research has demonstrated liver polyploidization to be tumor-protective, suggesting genome duplication was evolutionarily selected to better protect hepatocytes against the barrage of toxins the liver metabolizes daily [44]. WGD has also been shown to play a role in the lactating mammary gland, where cytokinesis failures gives rise to WGD⁺ cells that are more efficient at milk production [22].

Additional examples of cells utilizing WGD to adapt to pathogenic stimuli have come to light over the past decade. Research has demonstrated the presence of WGD⁺ cells in tissue repair [45], wound healing [46, 47], and disease [16, 40] (reviewed in [48]); yet, the extent to which WGD shapes the natural aging of organs and/or their response to endogenous and exogenous stimuli remains an active area of exploration [25, 34, 42, 49]. As genome duplication is a frequent route utilized by cells to mediate repair, it would not be surprising to discover that a majority of organ systems, even those not natively polyploid, harbor some basal rate of tetraploidy that increases in response to aging or toxic insults. However, the incidence of WGD in mainly diploid organs and tissues, where WGD is inconspicuous, remains unknown [12, 41, 50, 51]. This is due predominantly to technical limitations that preclude the accurate identification of WGD+ cells in vivo.

Oncogenic conscription of WGD

Programmed WGD is an essential component of natural development and frequently occurs in terminally differentiated cell types; however, inappropriate access to tetraploidy in proliferating tissues via unscheduled WGD can cultivate malignancy. The oncogenic predisposition of WGD cells, first hinted at by Theodor Boveri well over a century ago, has been validated by several experimental studies [8, 9]. The first direct evidence came from the demonstration that WGD⁺ p53^−/− mouse mammary epithelial cells generated by cytokinesis failure are sufficient to promote tumor formation when injected into nude mice, whereas genetically-matched diploid cells are not [8].

Subsequent studies confirmed aberrant WGD in different cell types is also oncogenic, irrespective of their mechanism of origin: WGD⁺ cells generated by viral-induced cell fusion [52–54], by endoreduplication induced through prolonged DNA damage via genotoxic drugs or telomere crisis [54, 55], or by endomitosis through inhibition of mitotic regulators [56] all promote tumorigenesis. WGD⁺ cells have also been observed in neoplasms of other species [57], as well as in many early stage or premalignant conditions that antedate the development of gross aneuploidy, further supporting a causal role of WGD in tumorigenesis [9, 58–60]. Most critically, 30–40% of all solid tumors arise from cells that experience a WGD event at some point during their evolution, with many occurring as early initiating events [5–7, 9].

Of all the mechanisms that can induce WGD in proliferating cells, cell division failure is thought to be the most common (Fig. 1). This is largely owing to the fact that completion of mitosis and cytokinesis requires the complex coordination of hundreds of proteins, thus introducing ample opportunity for error [1]. Oncogene-induced replication stress and telomere crisis, two well-established and common initiating events in tumor development, can also lead to cytokinesis failure by promoting the formation of chromosome bridges that can occlude the ingressing cytokinetic furrow [2, 61]. In addition, numerous cellular and/or genetic defects that prevent normal attachment of mitotic chromosomes to the mitotic spindle activate the spindle assembly checkpoint and induce mitotic arrest (reviewed in [62]). As this arrest cannot be maintained indefinitely, many mitotically-arrested cells return to G₁ phase without executing cell division in a process known as mitotic slippage, which also gives rise to WGD⁺ cells [63–65]. Oncogene-induced genome duplication has been demonstrated in vivo [66, 67], but the extent to which various oncogenic precipitating events (e.g., genetic mutations, toxin exposure) promote WGD across mammalian tissues is unknown. Understanding the permissibility of tissues to WGD, both natively and under oncogenic stress, may reveal why certain tumor subtypes have high or low WGD burden. For example, intraocular melanomas exhibit astonishingly low prevalence of WGD (~6%) compared to those that arise from cutaneous origins (~40%) suggesting WGD may be impacted by both cell-intrinsic and cell-extrinsic cues [5–7].

Mechanisms of genome instability in WGD⁺ cells

Extra centrosomes and micronuclei

Regardless of their mechanistic origins, WGD⁺ cells are well-equipped to catalyze tumorigenesis. Proliferating WGD⁺ cells are chromosomally unstable (CIN), meaning that they exhibit a persistently elevated rate of chromosome missegregation, and thus rapidly accumulate both numerical and structural chromosomal abnormalities [8, 68, 69]. Consequently, WGD⁺ cancers exhibit significantly higher rates of somatic copy number alterations (amplifications and deletions) relative to near-diploid cancers [5]. WGD⁺ cancers also exhibit sub-tetraploid ploidies, indicative of chromosome missegregation and loss [5–7, 70].This chromosomal instability is largely driven by the extra centrosomes gained during WGD [69]. Centrosomes are the microtubule nucleating centers of cells and their cellular quantity is tightly regulated (Box 2) [71].

Box 2: Control of Centrosome Number.

Diploid cells in G₁ have a single centrosome and as cells pass through S phase this centrosome is duplicated along with all chromosomes. The two centrosomes then enable efficient bipolar spindle assembly in mitosis. However, when cells experience a WGD, they arrive in G₁ with two centrosomes. During S phase, these centrosomes are again duplicated, giving rise to WGD⁺ mitotic cells with 4 centrosomes.

Extra centrosomes during mitosis promote multipolar spindle formation which, if uncorrected, precipitates a multipolar cell division producing three or more highly aneuploid, non-viable daughter cells [69] (Fig. 2). Many cells with supernumerary centrosomes adapt to this challenge by clustering their extra centrosomes into two poles during mitosis, thus enabling a return to bipolar division. During the process of centrosome clustering, a transient multipolar state exists which promotes the formation of merotelic attachments, in which a single kinetochore of a chromosome is attached to multiple spindle poles. These merotelic attachments promote chromosome missegregation and drive CIN [69]. Unresolved merotelic attachments also give rise to lagging chromosomes that form micronuclei following mitosis. Chromosomes in micronuclei are prone to massive DNA damage [68, 72] and complex chromosomal rearrangements [73–78] stemming from the rupture of fragile micronuclear envelopes [72]. Sequelae of these ruptures were recently shown to drive metastasis [79, 80] and promote epigenetic dysregulation of micronuclear chromosomes [81, 82] leading to cancer-driving heritable alterations following micronuclear chromosome re-integration. Thus, supernumerary centrosomes from WGD initiate a cascade of genome instability.

Figure 2: WGD requires cellular adaptations.

Cells which experience unscheduled WGD duplicate their cellular contents (e.g., centrosomes) as well as their genome. The doubling of these structures imposes unique stresses which WGD⁺ cells must overcome to proliferate. Clear evidence of this is seen with extra centrosomes. If WGD⁺ cells do not cluster surplus centrosomes during mitosis, the result is a multipolar division with uneven distribution of chromosomes and the production of non-viable daughter cells. To successfully proliferate, WGD⁺ cells must adapt to cluster their extra centrosomes in order to form a pseudo-bipolar spindle and produce viable progeny.

Abnormal DNA Replication

While it is known that developmentally-programmed endoreduplicating cells are prone to genome instability during S phase [83], the genomic instability acquired in erroneously-arising WGD⁺ cells was initially believed to require passage through mitosis. However, a new study suggests such instability can be obtained even in the first S phase following WGD [84]. The first evidence supporting this view was the observation that WGD⁺ cells, whether generated by mitotic slippage, endoreduplication, or cytokinesis failure, accumulated massive DNA damage in the first S phase after becoming WGD⁺. Single cell DNA sequencing of these WGD⁺ cells in S phase cells revealed they had already acquired complex karyotypes with multiple under- and over-replicated genomic regions relative to diploids [84]. Mechanistically, it was demonstrated that these defects accrued in S phase due to inadequate levels of DNA replication factors in WGD⁺ cells, which resulted in both fewer replication sites and increased replication stress (Fig. 3). The authors demonstrated that simply prolonging G₁ duration or overexpressing E2F1 to increase levels of replication factors could absolve WGD⁺ cells of DNA damage in S phase. These data suggest WGD⁺ cancer cells can acquire aberrant karyotypes even before their first division and offer another mechanism by which WGD promotes genomic instability. It should also be noted that unresolved replication stress gives rise to bridging chromosomes during anaphase, which often break into micronuclei (MN) and further contribute to CIN [61, 76, 85].

Figure 3: Replication stress in WGD cells.

Following cell division, DNA replication factors scale up in G1 in preparation for entry into S phase. In diploid cells, this results in sufficient numbers of factors required for successful DNA replication; however, in WGD cells there is insufficient scaling of factors to meet the demand for DNA replication, leading to the firing of fewer replication forks, increased replication stress, and genomic instability.

Disruption of chromatin architecture and organization

Chromosomes are not packaged into nuclei in a random and chaotic manner; rather, they are positioned in a highly spatially-organized way that helps to govern normal gene expression. For example, mapping of structural interactions of chromatin has demonstrated the existence of conserved genomic topologies, such as compartmentalization of chromosomes into segregated regions of transcriptionally active (compartment A) and repressed (compartment B) chromatin [86–89] (reviewed in [90]). It is therefore critical to understand how WGD, which entails doubling the number of chromosomes in the nuclear space, impacts three-dimensional chromatin architecture and gene expression.

A new study has revealed that mononucleate, p53-deficient WGD⁺ cells display loss of chromatin segregation (LCS), regardless of how they form (i.e., from mitotic slippage or cytokinesis failure) [91]. LCS was observed in WGD⁺ cells as an increased association between normally partitioned chromatin regions, as if the boundaries of compartments A/B were blurred. Breakdown of these partitions ultimately led to subcompartment repositioning of chromatin regions associated with oncogenes into compartment A and tumor suppressors into compartment B, contributing to oncogenesis (Fig 4). Mechanistically, LCS in WGD⁺ cells is driven by decreased abundance of CTCF, a critical protein in maintaining boundary insulation [92], and H3K9me3 marks, which are associated with heterochromatin or repressed compartments [93, 94]. Similar to Gemble et al., these deficiencies are likely due to inadequate protein scaling during WGD, as LCS can be rescued after pharmacologic prolongation of G₁ phase. Overall, these findings reveal another mechanism through which WGD⁺ cells acquire oncogenic properties [91].

Figure 4: WGD alters 3D chromosome spatial orientation and gene expression.

The spatial orientation of chromosomes within the nucleus has a direct impact on gene expression. Segments of DNA that fall within “A compartments” are highly expressed while segments of DNA that fall within “B compartments” are not. WGD results in changes in nuclear size and chromosomal numbers which alters the 3D spatial chromosomal orientation within the nucleus driving changes in gene expression by altering which DNA segments are positioned in A and B compartments.

Collectively, the mechanisms described above enable burgeoning WGD⁺ cancer cells to sample a myriad of genomic permutations to find the combination that best supports tumor growth, all while enjoying buffering against genetic alterations that would otherwise be lethal to diploid cells [9, 95]. It is these characteristics that explain why WGD is tumorigenic and often correlated with poorer prognosis [9, 95, 96]. Future studies are needed to identify how the mutational landscape preceding WGD molds tumor evolution and to reveal what three-dimensional chromatin architectural features are most conserved across WGD⁺ human tumors.

Genetic states the favor WGD

The genetic and physiologic properties conferred by WGD are not universally beneficial, nor are they without a fitness cost; it is therefore valuable to understand the genomic factors that influence whether a WGD event will be selected for or not during tumor evolution. Cancer cell genetic evolution can be thought of as a model of asexual reproduction where detrimental mutations continually accumulate in an irreversible procession towards cell death, a phenomenon known as Muller’s ratchet [97]. Overcoming these mutations is therefore paramount to the evolution of tumors. Interestingly, recent work has demonstrated that WGD represents a path for cells to mitigate the deleterious ratchet-like accumulation of deleterious mutations, and thus provide a selection pressure for WGD enrichment in certain tumor contexts [10].

Modeling and sequencing studies have discovered WGD to be preferentially enriched in cancer cells with a high rate of deleterious alterations in previously haploid regions of the genome, as WGD can rescue haploidy and thus buffer against chromosomal losses and/or inactivating mutations in essential genes in these genomic areas [7, 9, 10]. The ability of WGD to help cells compensate for haploidy suggests that cancer cells which lack widespread haploid genomic areas will not select for WGD. Indeed, it has been revealed that hypermutator (MSI-high) tumors, which have relatively fewer regions of genomic haploidy, are rarely WGD [7, 9, 10]. A beautiful recent study by Baslan et al. also supports this view [98]. In this work, the authors used a lineage-traced mouse model to observe pancreatic cells during their evolution from pre-malignancy to adenocarcinoma following complete Trp53 (which encodes murine p53) loss. Their analyses revealed a deterministic pattern in which sporadically-arising p53-null cells first acquire and accumulate genomic deletions before undergoing WGD, and it is only following WGD that chromosomal amplifications become enriched in pancreatic cancer cells. This model also suggests that WGD first mitigates the deleterious effects of gene loss and then subsequently promotes the acquisition and accumulation of cancer-driving chromosome gains [98].

Mechanisms restraining WGD⁺ cell proliferation

The duplicitous nature of WGD offers a gateway to both novel biologic possibility and uncontrolled cellular growth. Multicellular organisms have evolved to balance this dichotomy through the creation of tumor suppressive mechanisms which restrain WGD⁺ cell proliferation. Cells employ multiple layers of tumor suppressor pathways that act in concert to restrain WGD⁺ cell growth, often from non-genetic cues. Upon cytokinesis failure, non-transformed cells have been shown to activate both the p53 and Hippo tumor suppressor pathways triggering a G₁ cell cycle arrest which culminates in senescence, apoptosis, and/or immune clearance [99–102]. Accumulation of p53 is secondary to disruption of the E3 ubiquitin ligase MDM2 either through LATS2 or PIDDosome triggered caspase-2 activation [103–105]. The signal triggering these tumor suppressive responses has been found to be a non-genetic consequence of WGD, the presence of extra centrosomes and other alterations in cytoskeletal elements [103, 104].

To promote tumor formation, nascent WGD⁺ cells must overcome these barriers to proliferation, and several studies demonstrate this can be achieved via impairment of the p53 or Hippo pathways, or via stimulation of the E2F-driven G₁-S transition [99, 104, 106–111]. Consistent with this, WGD⁺ tumors are known to significantly enrich for mutations in TP53 (TP53 mutations are observed in ~50% of all WGD⁺ tumors, though this number varies significantly between different cancer subtypes), and are associated with amplifications in CCNE1 or CCND1 [5, 7, 9]. Even subtle changes in the regulation of these pathways through microRNA can promote WGD⁺ cell proliferation [107, 112] and in some cases even induce their formation [67].

Although many WGD⁺ tumors remain wild-type for TP53, a significant proportion of these tumors inactivate the p53 pathway via other mechanisms such as MDM2 amplification or CDKN2A loss [9]. Furthermore, in the MSK-IMPACT cancer sequencing cohort, Bielski et al. identified known deficiencies in E2F-mediated G₁ arrest in ~32% of WGD⁺ tumors with wild-type TP53 [9]. A recent study demonstrated that elevated cyclin E can promote replication stress and endoreduplication in cells with intact p53 signaling [109]. This path to becoming WGD⁺ presents a possible alternative mechanism for how at least a portion of tumors with wild-type TP53 access a hyperdiploid state. It remains unclear to what extent such a process occurs in human cancer. It will be interesting to examine whether CCNE1 amplifications or other disruptions of E2F-signaling commonly occur prior to WGD against a TP53 wild-type background and thus represent drivers of a tetraploid state. Further studies are required to elucidate all the mechanisms by which WGD⁺ cancer cells evade arrest, senescence, and death.

Unique characteristics and adaptations of WGD⁺ tumors

Simply acquiring the ability to escape G₁ arrest is not sufficient for the persistent proliferation of WGD⁺ cells. These cells must adapt to accommodate the numerous physiological changes which accompany a WGD [113, 114]. Broadly, not all cellular processes scale pro rata following WGD [84, 115, 116]. For instance, surface area to cell volume scales allometrically, where a doubling in volume only increases a cell’s surface area by ~1.6 fold [106, 113]. Similarly, although transcription in yeast has been found to generally increase commensurate with ploidy, new work demonstrates translation fails to proportionally scale in both mammalian and yeast cells [84, 116, 117]. As mentioned above, this is a significant finding given that inappropriate scaling of replication factors in newly formed WGD⁺ cells promotes DNA damage [84]. Genomically, WGD⁺ cells must also cope with managing their increased chromosomal complement with respect to chromatin organization [91]. How WGD⁺ cells adapt to overcome these newly discovered challenges of WGD remains an active area of investigation.

As detailed above, a well-recognized consequence and subsequent adaptation of WGD is the management of extra centrosomes, which can compromise cell viability by promoting catastrophic multipolar divisions. [69]. To combat this defect, WGD⁺ cells adapt by either clustering extra centrosomes into two poles during mitosis or discarding them in order to successfully proliferate [69, 118–121]. WGD⁺ cells can promote clustering genetically, through positive enrichment of PPP2R1A mutations [7, 122], or via utilization of microtubule motor proteins [118, 121]. Mutations in PPP2R1A have been shown to decrease mitotic functionality of protein phosphatase 2A (PP2A) and promote robust centrosome clustering, although the exact phosphorylation sites regulating this phenotype remains undiscovered [122]. Microtubule motors, such as the minus-end directed kinesin-like protein HSET, promote clustering by cross-linking nearby microtubules and then “walking” in a poleward direction towards centrosomes [118, 123]. This prompts the minus-ends of microtubules to focus, thus clustering wayward centrosomes together into a single pole. Alternatively, WGD⁺ cells may undergo asymmetric division, with one pole containing three centrosomes and the other containing one, to rapidly shed their additional centrosomes [69, 120].

Absent these adaptations, elevated centrosome count in addition to an already increased chromosomal load prompts activation of the spindle-assembly checkpoint, causing mitosis to be prolonged in WGD⁺ cells [7, 124, 125]. Mitotic prolongation via an active SAC, which pauses progression to anaphase until chromosome attachment to the mitotic spindle is complete, ensures faithful division despite the increased chromosomal compliment of WGD⁺ cells. This activation is critical to WGD⁺ cell survival as abrogation of this checkpoint reduces cell viability in a ploidy-specific manner [7, 126].

Despite adaptations aimed to preserve genomic integrity, proliferating WGD⁺ cells rapidly become aneuploid due to the chromosomal abnormalities which arise from aberrant mitoses [127, 128]. Initially the aneuploidy landscape of newly formed WGD⁺ cells is stochastic, but as WGD⁺ tumors evolve they appear to enrich for select aneuploidies [95], such as the co-occurrence of 10q and 18p loss in low-grade glioma [70]. The presence of recurrent aneuploidies in WGD⁺ tumors suggests that the enrichment or deletion of genes on these chromosomal regions confer specific survival benefits against a WGD⁺ backdrop. Thus, it is tempting to hypothesize that unique genes exist which act as oncogenes or tumor suppressors specifically in the context of elevated ploidy in cells (i.e., ploidy-specific oncogenes).

WGD in metastasis

The traits conferred upon cancer cells by WGD may impart the genetic and molecular tools necessary to thrive outside the primary tumor microenvironment, as recent sequencing studies have revealed over half of all metastatic tumors are WGD⁺ [129]. Similar to analyses of primary samples [5, 6], metastatic WGD rates are subtype dependent [9, 129, 130], ranging from 80% in esophageal to ~15–20% in CNS and mesothelioma tumors in the Hartwig Medical Foundation (HMF) cohort [129]. Although mutations in TP53 and CDKN2A are frequently observed in subtypes with a high fraction of metastatic WGD (e.g., esophageal, stomach) [129], a comprehensive analysis of mutations that significantly enrich in metastatic WGD⁺ vs WGD negative (WGD⁻) tumors has yet to be reported.

The incidence of WGD in primary and metastatic tumors is likely negatively skewed due to the limited number of samples sequenced per tumor in older cohorts. This is especially true when considering sub-clonal WGD, where WGD is only observed in a portion of samples sequenced in multi-regional studies. In support of this, new multi-biopsy studies have revealed the genomic landscape of advanced melanoma to frequently be WGD⁺ [131–133], higher than the ~40% WGD⁺ noted in metastatic samples in the HMF cohort [129]. With the expansion of multi-regional studies, it will be essential to delineate the distinctive genetic evolution and clinical impact of clonal as compared to sub-clonal WGD in both primary and metastatic tumors. Such analyses are already proving to be valuable as sub-clonal WGD in lung cancer was found to correlate with shorter disease-free survival, whereas clonal WGD did not [96]. It is important to note the degree of selection for WGD in metastasis is likely tumor subtype dependent. For example, a comparison of matched primary and metastatic lung cancer in the TracerX cohort revealed no significant difference in WGD status. This must be interpreted with caution though, as a significant fraction of the tumor pairs were already WGD⁺ as primary malignancies (63%) [134].

Why WGD⁺ cancer cells generally become enriched during metastasis remains enigmatic. The features which allow nascent WGD⁺ tumor cells to thrive early in tumorigenesis may also act to reduce the barriers to successful metastasis. These include the propensity for WGD⁺ cells to promote chromosome instability, which has recently been shown to drive metastasis [79]. Alternatively, the extra complement of centrosomes found in WGD⁺ cells may improve their ability to escape their native environment and invade other tissues [135]. Much work remains to fully describe how WGD not only impacts initial tumor formation but metastasis as well.

Therapeutic leverage of WGD status

The adaptations imposed by WGD opens a phenotypic divide between somatic diploid cells and WGD⁺ cancer cells which generates a window of therapeutic leverage [136]. The concept of WGD⁺ cells becoming conditioned to and reliant upon adaptations not critical for diploid cell growth is known as ploidy-specific lethality (PSL) [7, 106, 137]. Analysis of WGD⁺ compared to WGD⁻ cancer cell lines revealed multiple genes that when lost preferentially impact the survival of WGD⁺ cancer cells [7]. These ploidy-specific lethal genes largely function in pathways consequential to WGD⁺ cell proliferation [7]: DNA replication, given WGD⁺ cells have increased replication stress [85]; SAC signaling and chromosome alignment, as tetraploid cells endure prolonged mitoses [7, 124]; and protein turnover, as WGD⁺ cells rapidly become aneuploid and suffer from proteotoxic stress (reviewed in [138, 139]). One mitotic gene in particular, KIF18A, exemplifies the concept of PSL. KIF18A is a kinesin essential for the suppression of chromosomal oscillations at the metaphase plate and facilitates faithful chromosome alignment and segregation during cell division [140, 141]. Loss of KIF18A is largely dispensable for somatic murine development [142] and diploid cell viability [7, 11], but is essential for non-transformed tetraploid cells and many WGD⁺ cancer cells [7]. KIF18A dependencies have similarly been demonstrated in highly aneuploid [11] and chromosomally unstable cancer cells [143]. Marking the promise of this target, first-in-class inhibitors of KIF18A have already shown potential in vivo and have entered phase I clinical trials [144]. The side-effect profile of KIF18A inhibitors should theoretically be limited given their selectivity for proliferating WGD⁺ and/or chromosomally unstable cells. Since hepatocytes and megakaryocytes are the only known mitotically active WGD⁺ cells in humans, KIF18A inhibition may induce elevated transaminases (i.e., hepatotoxicity) or decreased platelet counts. KIF18A inhibition may also lead to reversible germ cell depletion based upon prior murine studies [142]. Unfortunately, while WGD⁺, high aneuploid score, or CIN may represent biomarkers for KIF18A sensitivity, these are not perfect: WGD⁺ cancer cells display variable sensitivity to KIF18A loss suggesting other factors that modulate KIF18A sensitization remain to be discovered [7, 11, 143].

Although WGD status is associated with treatment resistance and poor prognosis [9, 95, 96, 145–149], the use of WGD in the clinic to prognosticate or guide treatment is lacking. There are two major contributing factors to this deficiency. First, it remains unclear how WGD impacts therapeutic efficacy in a clinically actionable manner. However, recent studies have demonstrated a possible relationship between genomic heterogeneity, WGD positivity, and immunotherapeutic response, as WGD tumors tend to respond better to immune checkpoint inhibitors [7, 150]. Defining the association between these tumor characteristics may allow for the use of WGD to identify tumors with a higher likelihood of displaying anti-PD-1 sensitivity. Second, identifying a tumor as WGD⁺ is currently out of the reach of many clinicians. Academically, WGD⁺ tumors are identified by applying algorithms such as ABSOLUTE [6] or MEDICC2 [151] to whole exome or genome sequencing runs, yet these analyses are complicated and time-consuming. The development of simple tools by which clinicians can identify the WGD status of a patient’s tumor will facilitate the incorporation of WGD into clinical decision making for both prognostic and therapeutic considerations. Recent advances in artificial intelligence now open the possibility that WGD⁺ tumors may be quickly and easily identified based on the analysis of standard hematoxylin and eosin-stained tissue samples [152, 153].

Concluding remarks

Whole genome doubling shapes the development of our tissues and malignancies. Recent major discoveries have illuminated the crucible in which nascent WGD⁺ cancer cells evolve and ultimately how these selective pressures drive adaptations which can be leveraged for novel anti-tumor therapeutics. As we continue to further understand the adaptations WGD⁺ cells employ, both in tumorigenesis and metastatic disease, we will uncover new vulnerabilities to specifically target cancerous WGD⁺ cells. (See Outstanding Questions)

Outstanding Questions Box.

• How common are WGD⁺ cells in different tissue types? Do tissues with higher baseline levels of WGD tend to have a higher propensity for WGD⁺ tumors?

• What factors (genetic or environmental) increase the rate of WGD in vivo?

• What is the predominant mechanism that promotes the erroneous generation of WGD⁺ cells in vivo? Cytokinesis failure, mitotic slippage, endomitosis, endoreduplication, or cell fusion?

• Does the mechanism by which cells become WGD effect their tumorigenic potential?

• Is there evidence for the existence of ploidy-specific oncogenes or ploidy-specific tumor suppressors in human cancers?

• Does WGD impart cells with features that promote metastasis?

• Why are WGD tumors more likely to respond to immune checkpoint blockade therapies?

• How does p53 status prior to WGD shape future karyotypes?

• What are the consequences and side effects of therapeutically targeting all proliferating WGD⁺ cells in the human body?

• Can WGD be easily and reliably identified from tumor samples via methods independent of computational analysis of cancer genome sequencing?

• Does WGD promote resistance to cancer therapeutics? How?

• How does WGD affect the transcriptomes and proteomes of cells in vivo?

Highlights.

• Unscheduled whole-genome doubling events give rise to tetraploid cells that are prone to replication stress-induced DNA damage, chromosome instability, and oncogenic epigenetic alterations.

• Proliferating whole-genome doubled cells are tumorigenic and comprise ~37% of primary and ~56% of metastatic solid tumors.

• Whole-genome doubled cancer cells must acquire specific genetic and physiological adaptations to accommodate the unique stresses imposed by their doubled DNA and cellular content.

• Identifying genes that are essential for the viability of proliferating whole-genome doubled cancer cells, yet dispensable for the viability of diploid cells (i.e., ploidy-specific lethal genes), has the potential to uncover new cancer therapeutics.