Aneuploidy in Human Cancer: New Tools and Perspectives

Abstract

Chromosome copy number imbalances, otherwise known as aneuploidies, are a common but poorly-understood feature of cancer. Here, we describe recent advances in both detecting and manipulating aneuploidies that have greatly advanced our ability to study their role in tumorigenesis. In particular, new CRISPR-based techniques have been developed that allow the creation of isogenic cell lines with specific chromosomal changes, thereby facilitating experiments in genetically-controlled backgrounds to uncover the consequences of aneuploidy. These approaches provide increasing evidence that aneuploidy is a key driver of cancer development and enabling the identification of multiple dosage-sensitive genes encoded on aneuploid chromosomes. Consequently, measuring aneuploidy may inform clinical prognosis, while treatment strategies that target aneuploidy could represent a novel method to counter malignant growth.

Chromosomal alterations in cancer

David von Hansemann first observed mitotic chromosomal mis-segregation in epithelial cancers more than 130 years ago [1], but it was another 24 years before Theodor Boveri and Marcella O’Grady postulated that this chromosomal variability drove tumor development [2,3]. A century later, the functional causes and consequences of aneuploidy in cancer development, maintenance, and metastasis remain elusive.

It is important to distinguish aneuploidy from polyploidy and from chromosomal instability (CIN). The term “aneuploidy” refers to an unbalanced gain or loss of whole chromosomes or chromosomal arms [4], while polyploidy refers to the gain of an equal set of all chromosomes. CIN, meanwhile, is an elevated rate of chromosomal mis-segregation during cell division. While CIN can lead to aneuploidy, not all aneuploid cells exhibit CIN [5]. Aneuploidies are a type of somatic copy number alteration (SCNA) that spans at least the length of an entire chromosome arm, but SCNAs as a class also include sub-chromosomal amplifications and deletions that may affect only one or a few genes.

Several excellent reviews have recently covered various facets of aneuploid biology, including its causes and consequences [5–8], its context-dependent roles in cancer [4,9], and the interplay and distinctions between CIN and aneuploidy [10,11]. The history of aneuploidy research has included diverse model systems, beginning with the study of sea urchin embryos [3] and encompassing yeast, chicken, and mouse cells [4,8,9]. In this review, we describe several new tools and approaches that have been developed in the past few years and that have vastly expanded our ability to manipulate and interrogate cellular karyotypes in human cancers. In total, these recent findings suggest that aneuploidy is not a passenger or simple byproduct of tumorigenesis, and instead, as Boveri and O’Grady hypothesized, aneuploidy can drive cancer development.

Detecting Aneuploidy in Cancer

Historically, chromosomal aberrations were mainly detected through cytogenetics. Optimization of the metaphase spread protocol in 1956 convincingly demonstrated that human cells had 46 chromosomes [12,13], and subsequently chromosomal aberrations were identified in Down [14], Turner [15], Klinefelter [16], Edwards [17], and Patau [18] syndromes. Using these same techniques, a translocation involving chromosome 22, called the “Philadelphia chromosome,” was identified in the leukemic cells of chronic myeloid leukemia patients and was found to activate the ABL1 oncogene [19,20]. With improvements in methods of widespread chromosome analysis, the era of clinical cytogenetics began [21].

Chromosome banding techniques such as Q-banding [22] and G-banding [23] were introduced in the 1970s, markedly improving confidence in identification of human chromosomes. Analysis of patient tissues using these approaches established widespread copy number alterations as a key feature of human cancers. Early examples included translocations between chromosomes 8 and 14 in Burkitt’s lymphoma [24], and deletions on chromosome 13 in B-cell lymphocytic leukemia [25]. Subsequently, multicolor fluorescence in situ hybridization (M-FISH) was developed [26], allowing individual chromosomes to be discerned from their ‘painted’ profiles [27]. To reduce time spent per individual sample and increase the resolution of SCNAs, array-based comparative genome hybridization (CGH) was developed, which enhanced the diagnosis of chromosomal disorders [28]. Related techniques were used in the creation of large-scale databases – such as The Cancer Genome Atlas (TCGA) – that incorporate copy number data from thousands of cancer patients. More recently, read-depth analyses of next generation sequencing has facilitated the ultra-precise high throughput detection of copy number changes from whole genome sequencing, whole exome sequencing, and RNA sequencing [29][30]. Commonlyused approaches include SMASH (short multiply aggregated sequence homologies) [31], a highly multiplexed, low-cost method for identifying copy number variants, and FACETS, an approach for detecting aneuploidy in standard next-generation sequencing workflows [32]. Single-cell whole genome sequencing can also be used to detect sub-clonal aneuploidy in tumors, cell lines, and healthy tissue (reviewed in [33]).

Recent analysis of large-scale patient datasets like TCGA have confirmed von Hansemann’s initial observations and demonstrated that aneuploidy is a ubiquitous feature of human tumors. 88% of all tumors exhibit aneuploidy, and the most frequent single arm alterations were present in over 33% of tumors (chromosome 8p and 17p loss, chromosome 8q gain) (Figure 1) [34]. Even though SCNAs are present in 99% of tumors [35], the most common individual focal copy number alterations (MYC amplifications and CDKN2A/B deletions) are observed in just 14% of cancer specimens [36]. Arm-level gains and losses are observed at 30 times the expected frequency when compared to focal copy number alterations, and affect 25% of the genome in a typical cancer sample, while only 10% of the genome is affected by sub-chromosomal SCNAs [36].

Figure 1: Chromosomal Arm Level Alterations Across Patient Tumor Samples.

This figure was adapted from ref [34]. The study used data from 10,522 genomes from TCGA to assess changes in chromosomal copy number across 33 subtypes of cancer.

Some aneuploidies are commonly observed across tumor types while others are specific for one or a few lineages (Figure 2). For example, gains of chromosomes 1q and 8q are found in a diverse array of cancers, while gains of chromosome 3q are most common in squamous cell carcinomas [34]. Additionally, clinical analyses have revealed that aneuploidy can serve as a prognostic biomarker for patient outcomes and therapeutic responses [11,37–39]. In some cases, aneuploidy and/or SCNA burden may harbor more prognostic power than commonly-assessed cancer mutations. For instance, in a study of genomic biomarkers associated with cancer patient survival, copy-number gains affecting the oncogene PIK3CA were associated with disease progression in five of 16 cancer types, while mutations in PIK3CA were not associated with disease progression in any cancer type studied [38]. Similarly, low grade gliomas marked by co-deletion of chromosomes 1p and 19q have been found to respond better to specific chemotherapy regimens [40], and a retrospective analysis of melanoma patients treated with immune checkpoint blockade showed a negative correlation between SCNA burden and patient response, suggesting that analyzing SCNAs could predict immunotherapy efficacy [41]. Translating these observations into validated and widely-applied clinical biomarkers represents an important future endeavor.

Figure 2: Frequently Observed Aneuploidies Across Human Cancer.

The two most frequently observed chromosomal gains (red) and losses (blue) across 14 different cancer types are highlighted. The data for this figure was obtained from Taylor et al. 2018. The study analyzed 10,522 genomes from The Cancer Genome Atlas (TCGA) to assess changes in chromosomal copy number across 33 subtypes of cancer.

When Does Aneuploidy Arise?

Initially, karyotypic dissection using chromosome banding and FISH approaches at distinct tumor grades revealed early and late arising copy number changes. For example, chromosome staining showed gains of chromosomes 1q, 2, 7 and 10 occurred in early stages of endometrial cancer [42], and FISH confirmed the prevalence of chromosome 12 aneuploidy in B-cell chronic lymphocytic leukemia [43]. Similarly, gain of chromosome 3q was shown to mark the transition from severe dysplasia to invasive carcinoma in cervical cancer [44,45]. In fact, the frequency and length of genomic deletion was found to scale with the severity of histopathological preneoplastic changes [46]. Gains of chromosomes 8q, 13q and 20q and loss of chromosomes 18q and 17p were observed more frequently in colorectal carcinomas than adenomas [47–49], indicating copy number changes tended to accumulate with cancer progression. Using whole genome sequencing, a subtype of breast cancer is now defined by chromosome 1q gain and 16q loss [50,51], which have been implicated in the genesis of both ductal and lobular carcinomas [52]. The progression from Barrett’s esophagus to esophageal carcinoma is also associated with aneuploidy [53], and trisomy of chromosome 5 may promote metastasis by inducing a partial epithelial-to-mesenchymal transition [54]. Thus, aneuploidy can be observed during the earliest stages of cancer development, and karyotypic alterations are associated with increased tumor invasiveness and aggressiveness.

Longitudinal studies of large patient cohorts, such as TRACERx [55], offer the opportunity to leverage multi-sample sequencing techniques to trace the evolutionary history of copy number changes at a much higher resolution than was previously possible. Recently, an integrated multi-sample phasing and SCNA analysis of nearly 400 tumors across 22 cancer types [35] resolved the temporal order of SCNA acquisition. By distinguishing the acquisition of sub-clonal and clonal SCNAs in relation to whole genome duplication (WGD) and metastasis, the authors mapped the influence of SCNAs on a tumor’s fitness landscape. For example, loss of 17p13.1, the locus harboring TP53, was found to occur early in 9 out of 13 tumors, suggesting this alteration was necessary for both tumorigenesis and tolerance of subsequent WGD. Meanwhile, losses of TP53 and STK11 loci in lung adenocarcinoma were found to both occur early and persist into metastasis, suggesting these alterations define the tumors’ metastatic potential [35]. A separate method for assessing chromosomal alterations before or after population expansion found that chromosome 5q was gained early in colorectal adenocarcinoma, while chromosome 1q gain occurred early in breast cancer [56]. These findings support a model whereby specific aneuploid chromosomes can be drivers for tumorigenesis, and chromosomal gains occur in punctuated bursts at similar molecular times, resulting in beneficial karyotypes for clonal population growth [57].

Tumor sequencing studies have also revealed that WGD events can elevate the frequency of chromosome mis-segregation. Over 30% of tumors exhibit WGD [58], and WGD is associated with a higher rate of every other type of SCNA [59]. WGD can also mark a shift in bias from chromosomal gain to chromosomal loss [37]. This suggests chromosomal gains are preferred very early on during tumor development, and chromosomal loss occurs following WGD, when multi-gene deletion can be better tolerated in a polyploid background [37]. Combined, these observations highlight a strong association between aneuploidy and tumorigenesis, with early chromosomal gains driving malignancy, and later chromosomal gains and losses further optimizing the karyotypic landscape for tumor expansion.

Experimental Manipulation of Chromosomes

While cytogenetics and sequencing have established associations between specific aneuploidies and cancer, our ability to directly study the impact of individual chromosome arm changes has been limited by the challenges associated with experimentally manipulating chromosome copy number. Over the past several decades, the standard tools of molecular genetics have been successfully used to evaluate the functional role of individual oncogenes and tumor suppressors. Invariably, a thorough demonstration of gene functionality involves its downregulation to assess necessity for an observed phenotype, and its upregulation to assess sufficiency. Downregulation has been accomplished through genetic, transcriptional and proteomic manipulation, for example using homologous recombination [60], CRISPR/Cas9 [61], RNAi [62], CRISPR interference [63], and PROTACs [64,65,65]. Upregulation has been accomplished through the introduction of overexpression constructs, either virally into DNA [67], through transient expression of cDNA or ORFs [68,69], or more recently using CRISPR activation [70]. However, these single-gene approaches are generally insufficient for manipulating whole chromosomes that span millions of base pairs and encode hundreds of genes. Recently, the experimental repertoire for targeted chromosomal manipulation has undergone a significant expansion, allowing direct interrogation of aneuploidy’s role in tumorigenesis (Figure 3).

Figure 3:

Methods for Targeted Chromosomal Gain or Loss (A) Extra chromosomes can be introduced into cells using microcell mediated chromosome transfer (MMCT). (B) dCas9 fused to a component of the mitotic machinery and targeted to unique chromosomal repeats can induce mis-segregation. (C) TALENs and CRISPR/Cas9 can be used to induce targeted chromosomal arm deletions, which may be facilitated by artificial telomeres or negative selection cassettes. All techniques can be used for the generation of genetically-matched lines differing in an aneuploid chromosome.

The isolation of aneuploid cells from individuals affected by disease or developmental abnormalities provided the first opportunity to assess the consequences of aneuploidy. For example, isolation of trisomy 13, 18, and 21 cells from individuals affected by Patau, Edwards or Down Syndrome, respectively, has provided insights into cellular processes affected by the extra chromosome [71–73]. Similarly, a comparison between aneuploid and non-aneuploid cell lines and tumors from cancer patients has revealed common patterns of genomic, transcriptional and proteomic dysregulation [34,74,75]. However, only the three autosomal trisomies mentioned above are compatible with viability, and a lack of matched genetic controls devoid of the aneuploid chromosome introduces many confounding variables into this analysis.

Inducing Specific Chromosome Gains

In order to generate specific chromosomes gains for any chromosome, microcell-mediated chromosome transfer (MMCT) was developed in the 1970s by combining the technologies of microcell formation and cell fusion to introduce chromosomes from donor to recipient cells [76,77] (Figure 3A). At first, this allowed the introduction of a few chromosomal segments into exogenous cells through fusion of mouse [78] or chicken [79] microcells with Chinese hamster ovary (CHO) or human HeLa cells. This facilitated positional cloning for the identification of novel genetic loci implicated in disease, including tumor suppressor genes [80] in cancer and complementation groups in autosomal recessive disorders such as Fanconi anemia [81].

A major advance in studying aneuploidy came at the turn of the century with the development of a library of mouse A9 cell hybrids containing single copies of human chromosomes [82]. This, in turn, facilitated the generation of isogenic human cell lines that differed in the copy number of a single aneuploid chromosome, which have since allowed examination of the oncogenic and tumor suppressive contributions of aneuploid chromosomes [83–86]. The history of MMCT and its current practice have been further reviewed elsewhere [87,88].

It should be noted that the use of MMCT to study aneuploidy has several limitations. Donor chromosomes and target cell lines are almost always genetically distinct, and the transferred chromosome may harbor foreign SNPs and epigenetic alterations that influence the phenotype of the resulting cell population. Micronuclei are inherently unstable, and chromosomes subject to MMCT may undergo massive chromosomal rearrangements and chromothripsis [86,89], which can lead to activation of immunostimulatory cGAS-STING signaling [90,91]. Finally, exogenous introduction of an extra chromosome into a cellular background that hasn’t adapted to aneuploidy may not necessarily reflect how aneuploidy evolves in a tumor and how cells adapt to tolerate that aneuploidy. Therefore, the interpretation of phenotypic consequences of aneuploidy using MMCT-derived models should be undertaken with careful consideration of these effects.

Inducing Random Chromosome Mis-segregation

Induction of chromosomal mis-segregation is another useful technique to generate aneuploid cells. Various components of the spindle machinery can be manipulated to trigger mis-segregation, including microtubules [92,93], and topoisomerase II [94]. However, the prolonged mitotic arrest caused by spindle disruption causes DNA damage, confounding any assessment of the consequences of aneuploidy [95]. To minimize DNA damage, transient inhibition of the Mps1 checkpoint kinases, sometimes preceded by inhibition of the kinesin CENP-E, has been recently used to induce chromosomal mis-segregation without triggering a mitotic arrest [85,89,96–98]. This approach has yielded valuable insights into the consequences of temporary CIN, such as the generation of aneuploid karyotypes beneficial for chemotherapeutic resistance [97,98]. Single-cell cloning after periods of CIN can be used to isolate populations of cells that all harbor the same chromosomal alteration [54,97,99]. Random aneuploidies can also be produced in vivo via genetically-engineered mouse models (GEMMs) that harbor deletions or hypomorphic mutations in genes encoding components of the spindle checkpoint [9]. However, spindle checkpoint mutations are extremely rare in human cancers [5], and many of these checkpoint proteins have been found to moonlight in other cellular processes [100]. Finally, it is important to note that the aneuploidies that arise following either in vitro or in vivo spindle checkpoint perturbations are largely random, and these techniques are unable to produce “designer” karyotypes.

Precise Manipulation of Cellular Karyotypes

In order to generate cells that harbor specific chromosome changes, targeted approaches are necessary. One of the first strategies for selective chromosome targeting utilized the fact that the centromeric proteins CENP-A and CENP-B harbor partially redundant roles promoting chromosome segregation, but CENP-B does not bind to the Y chromosome. By depleting endogenous CENP-A, researchers were therefore able to selectively target the Y chromosome for mis-segregation. While this method only works for a single chromosome, selective mis-segregation in a temporally controlled manner allowed for in depth analysis of cell fate following chromosome mis-segregation and micronucleus formation [101].

Since 2022, several new methods have been published to facilitate the gain or loss of specific, designated chromosomes. Broadly, they utilize CRISPR technology to selectively interfere with normal mitotic progression and target an individual chromosome for mis-segregation (Figure 3B). One promising approach is called KaryoCreate, which utilizes sgRNAs targeting unique centromeric α-satellite repeats to recruit dCas9 fused to a mutant form of the kinetochore scaffold protein KNL1. This alters the normal error-correction process mediated by Aurora B and PP1, resulting in mis-segregation of the targeted chromosome [102]. Similarly, dCas9 can be fused to either the kinetochore-nucleating domain of CENP-T [103] or the Physcomitrella patens Kinesin14VIb motor protein [104], and targeted to repetitive chromosomal regions. CENP-T localization creates an ectopic kinetochore that interferes with accurate segregation, while Kinesin14VIb is a minus-end-directed motor protein that can cause poleward chromosome movement prior to anaphase onset. Thus, these three techniques can be used to trigger the mis-segregation of a targeted chromosome, facilitating the creation of designer karyotypes.

The effects of chromosome dosage changes can also be studied by deleting specific regions from the genome (Figure 3C). These deletions can be produced using technologies that introduce dsDNA breaks at precise loci, in particular TALENs and CRISPR/Cas9. In turn, this has allowed induction of targeted chromosomal arm loss. For example, chromosome 8p loss of heterozygosity was engineered in mammary epithelial cells using two pairs of TALENs to induce loss of a 33 Mb segment [105]. Provision of an artificial telomere to cap a CRISPR-induced dsDNA break subsequently facilitated telomere truncation of chromosome 21 [106], as well as targeted deletion of chromosome 3p [34]. Recently, it was also reported that chromosomal loss could be a consequence of CRISPR/Cas9 induced DNA breaks in embryonic stem cells [107]. This led to CRISPR use for the explicit purpose of inducing chromosomal arm loss in clonal populations of cells [108].

Inspired by these approaches, we introduced a toolkit for targeted chromosomal deletion called ReDACT (Restoring Disomy in Aneuploid Cells using CRISPR Targeting) [109]. Using three different methods – negative selection of the aneuploid chromosome, capping of a centromeric dsDNA break using an artificial telomere, and simply targeting the chromosome with CRISPR/Cas9 – we showed successful deletion of aneuploid chromosomes 1q, 7p and 8q in distinct cellular backgrounds.

A discussion of the practical utility of these new techniques for targeted chromosomal manipulation techniques is presented in Box 1. These methods offer the opportunity to dissect the contribution of each aneuploid chromosome to any cancer phenotype. While a comprehensive study of all clinically relevant aneuploid chromosomes – either individually or in commonly observed combinations – will undoubtedly require a monumental effort, there is tremendous potential for discovering new cancer biology.

Box 1: Comparison of Targeted Arm Loss Methods.

The available methods for targeted chromosomal arm loss offer their pros and cons. Designer karyotypes are now possible, but the efficiency of each technique varies, and demonstrated cases of targeted manipulation are still limited (Table I). Methods using dCas9-CENPT or dCAS9-Kinesin14VIb can lead to either chromosome gain or loss and require chromosomes with repetitive endogenous arrays. While this range can be expanded by targeting dCas9 to an integrated TetO repeat [104], the dual step method will increase length of time required for aneuploidy manipulation, and likely reduce efficiency. Furthermore, any techniques that create opposing tensions on a chromosome to “pull” it apart risk DNA damage. Indeed, γH2AX foci were observed upon use of KaryoCreate [102]. Additionally, any methods requiring multiple instances of dsDNA break creation increase the likelihood of off-target effects and secondary karyotypic alterations. Drug selection for expression of exogenous cassettes also exerts selection pressure on the cells, which may manifest in phenotypic consequences. Therefore, it is important to take these aspects into consideration for any experimental design.

All methods outlined above demand single cell cloning for the generation of aneuploidy gain or loss populations of cells, and none have thus far been used with patient-derived organoids. This means that genetic heterogeneity – such as it may be in cell lines isolated from patients decades ago and propagated under non-physiologic tissue culture conditions – is nevertheless lost [110]. Efficiencies also vary widely and aren’t standardized by a singular metric. Therefore, it is unclear how many single cell clones need to be isolated for each technique to produce a clone with targeted chromosomal arm gain or loss event. The credibility of a single-cell cloning approach relies on strength in numbers – aneuploidy manipulation in diverse and clinicallyrelevant genetic backgrounds, with large numbers of clones and appropriately matched controls, increases confidence in the results. Chromosomal engineering approaches that can be utilized efficiently for a bulk population of cells need to be developed to mitigate the caveats of single cell cloning, which in turn may allow their use for organoids and mouse models.

Chromosome engineering in mice

In addition to the strategies described above for manipulating chromosomes in cell lines, several new techniques have been developed to target mouse chromosomes. MICER (mutagenic insertion and chromosome engineering resource) – a Cre-loxP based approach for engineering deletions – was developed to engineer large genomic changes including deletions and duplications in addition to biallelic gene inactivation [111]. MACHETE (molecular alteration of chromosomes with engineered tandem elements) – a CRISPR-HDR guided positive-negative selection approach enables megabase deletions [112]. These techniques enable investigation of syntenic SCNAs in vivo and are discussed in more detail below.

Identifying Dosage-Sensitive Genes Affected by Aneuploidy

Aneuploidy is one of the most common features in cancer, however, aneuploidy is very rarely found in normal tissues [113]. The introduction of extra chromosomes into diploid cells is not always beneficial for cell fitness [83,85], as it can disrupt both the transcriptome and proteome [114]. The prevalence of SCNAs and aneuploidy in cancer suggests that the simultaneous gain or loss of multiple genes may not only compensate for the negative consequences of having an aneuploid karyotype, but also provide an additional fitness advantage for cancer cells. Using techniques discussed above, it was found that deletions of chromosome regions containing multiple tumor suppressor genes exhibited synergistic growth enhancement compared to deletions of single genes alone. MICER was used to create a 4 Mb deletion in mouse chromosome 11, a region syntenic to human 17p13.1 (where tumor suppressor gene TP53 is located), and the deletion accelerated lymphoma and leukemia development compared to deletion of TP53 alone [115]. MACHETE was used to delete the CDKN2A cluster in a syngeneic mouse model of pancreatic cancer. While deletion of the CDKN2A cluster enhanced proliferation, co-deletion of the neighboring IFN cluster also facilitated immune evasion, thus synergizing to evade both cell intrinsic and extrinsic tumor suppressor mechanisms [112]. Therefore, these strategies indicate that multiple dosage-sensitive genes may drive selection for the sub-chromosomal deletions found in cancer genomes.

Similarly, multiple fitness-enhancing genes may promote the selection for chromosome-spanning aneuploidies [116]. Deletions of mouse orthologues of human genes present on chromosome 8p in a model of hepatocellular carcinoma showed that multiple genes cooperatively inhibit tumorigenesis, and their deletion can synergistically promote tumor development [117]. Deletion of chromosome 9p in low grade gliomas was also attributed to several genes, which in turn can serve as subtype-specific prognostic biomarkers for tumor aggressiveness and patient survival [118]. The simultaneous copy number gain of RAD21 and MYC in EWS-FLI1 driven Ewing sarcomas was found to be crucial for the oncogenic benefits of trisomy 8, as RAD21 dampened replication stress caused by the fusion oncoprotein. Overexpression of MYC promoted proliferation only when combined with increased RAD21 expression [119]. Similarly, in mice, chromosome 15 (syntenic to human chromosome 8q) is frequently gained in T cell lymphomas. When human MYC was expressed on chromosome 6 and endogenous MYC ablated on chromosome 15, karyotypes now included universal chromosome 6 gains in addition to a gain of chromosome 15, whose gain was attributed to Rad21 [120]. In our own work, we found that deletion of one copy of chromosome 1q had a much greater phenotype on anchorage independent growth than single copy deletion of MDM4, a negative regulator of p53 located on 1q. Overexpression of MDM4 and BCL9 was able to rescue anchorage independent growth better than either gene in 1q loss clones, but dual overexpression did not rescue to 1q trisomy (wildtype) levels [109]. These results suggest that while aneuploidies may be partially driven by single oncogenes or tumor suppressors, changes in expression of multiple dosage-sensitive genes on aneuploid chromosomes are required to fully explain the observed phenotypes.

Future Approaches for Aneuploidy Dissection

Now that we have targeted approaches to generate specific SCNAs and chromosome gains and losses, this toolkit can be used to uncover how specific chromosome copy number changes contributes to tumorigenesis. Together, the techniques described in Box 1 could potentially be used to target any chromosome, but so far, their use has been limited to just a handful of chromosomes and chromosome arms, and in depth analysis of the clones generated using these techniques has been limited to a few chromosomes [34,102–105,109]. More work needs to be done to expand which chromosomes can be targeted, and once chromosomes have been targeted and the desired karyotypes generated, to begin to dissect which dosage-sensitive genes on the chromosome are driving cancer phenotypes.

While development of targeted chromosomal mis-segregation and arm loss methods will go a long way towards creating isogenic cellular models for assessment of aneuploid phenotypes, it remains to be seen whether these techniques can be successfully utilized in vivo (see Outstanding Questions). Currently, in vivo chromosome targeting experiments are limited to small-scale SCNAs, and adapting strategies for chromosome targeting developed for cell culture would be difficult given the low efficiency of chromosome targeting and the fact that these approaches require single-cell cloning (see Box 1). Targeted deletion of an aneuploid chromosome or induction of chromosomal gain at various stages of tumor development within an organoid or mouse model would provide incredible insight into both the immediate and the longer-term consequences of aneuploidy. Additionally, a further challenge of using GEMMs to recapitulate human aneuploidy is limited synteny between mouse and human genomes.

Outstanding Questions.

• Can chromosomal engineering approaches be applied to animal models and patient-derived organoids?

• Are cancers addicted to aneuploidy and somatic copy number alterations in a manner similar to oncogenes?

• What is the core subset of genes that drives any given aneuploidy?

• Can therapeutic vulnerabilities created by the presence of an aneuploid chromosome be clinically exploited?

Another useful approach would be chromosomal tiling. While groups have nominated candidates as dosage-sensitive drivers of aneuploid chromosomes – SMAD2 and SMAD4 for chr18q loss [102], RAD21 and MYC for chr8q gain [119,120], and MDM4 and BCL9 for chr1q gain [109] – these genes only partially explain the observed phenotypes. Therefore, chromosomal tiling approaches, where different segments of aneuploid chromosomes are deleted or overexpressed, could shed light on the importance of focal amplifications and deletions as compared to whole chromosome or chromosomal arm aneuploidy. Both ReDACT [109] and MACHETE [112] are capable of engineering genomic deletions spanning tens of megabases, and could be utilized for this purpose. Meanwhile, techniques for creating human artificial chromosomes (HACs) are rapidly improving and may soon offer the possibility of introducing entire chromosomal arms stably into cells [121,122]. Creating designer HACs may bypass the challenges associated with genetic and epigenetic heterogeneity previously observed using MMCT with a generic donor chromosome, though it remains to be determined whether they would also exhibit structural instability [86]. These approaches could help assess the simultaneous necessity and sufficiency of chromosomal segments spanning hundreds of genes.

Furthermore, the development of massively parallel screening approaches, if conducted in a chromosome-specific manner, could allow discovery of dosagesensitive genes. CRISPR interference (CRISPRi) and activation (CRISPRa) screens have already been successfully utilized to discover new facets of oncogene and tumor suppressor biology [123,124]. Thus, libraries designed to downregulate or overexpress chromosome-specific gene expression in an aneuploid and geneticallymatched non-aneuploid setting respectively could identify dosage-sensitive genes present on aneuploid chromosomes. It is critical, however, that gene expression be modulated to levels comparable to chromosome gain or loss. For example, if gene expression scales with chromosomes copy number, CRISPRi in trisomic cells should downregulate targeted gene expression by 33% to assess consequences of chromosome loss, and CRISPRa in disomic cells should upregulate gene expression by 50% to assess consequences of chromosome gain. Modelling gene expression on aneuploid chromosomes is complicated due to the fact that not all copy number gains or losses of genes scale at the expected ratios of gene gain or loss due to buffering at the mRNA and protein level for some genes [75,125]. Although such precise tunability of gene expression is currently difficult, novel techniques are being developed for this purpose, including the use of chemical epigenetic modifiers to implement dose-dependent gene expression [126]. Furthermore, as mentioned above, it is likely that multiple genes on the aneuploid chromosome contribute to its gain or loss. While efficient use of multiplexed CRISPR arrays [127] and multi-gene overexpression constructs [128] remains challenging, their application in aneuploid settings would help uncover the minimal subset of genes necessary and sufficient for the aneuploid phenotypes.

Conclusions

Our ability to detect chromosome copy number changes in cancer cells has improved greatly over the last few decades, and analysis and comparison of thousands of cancer genomes have identified recurrent chromosomal gains and losses. To study the contributions of individual chromosomes and SCNAs, new techniques have been developed and are being used to generate isogenic cell lines with desired karyotypes, which can be used to discover dosage-sensitive tumor driving genes and may help find new aneuploidy-specific druggable targets.

Table I:

Summary of Practical Use for Targeted Chromosomal Arm Loss Methods

Technique	Karyocreate	Targeted Mis-segregation	Targeted Mis-segregation	ReDACT	Telomere Truncation	TALENs
Tools	dCas9-KNL1RVSF/AAAA + centromere-specific sgRNAs	dCas9-CENP-TΔC + Mps1i pulse	dCas9-Kinesin14VIb	Centromere-specific Cas9 gRNAs (+ negative selection cassette or artificial telomere)	Cas9 + artificial telomere & drug selection cassette	TALENs spanning intended deletion
Consequence	Chromosome gain or loss	Arm gain or loss	Arm gain or loss	Arm loss	Arm loss	Arm loss
Targetable Chromosomes	Theoretical: 19/24; Demonstrated: 10/24	Theoretical: 18/24 Demonstrated: 1p & 9q	Theoretical: 18/24 Demonstrated:1p & 9q	Theoretical: all; Demonstrated: 1q, 7p, 8q	Theoretical: all; Demonstrated: 3p	Theoretical: all; Demonstrated: 8p
Efficiency	8% for chromosomal gain; 12% for chromosomal loss	61–70% mis-segregation in transfected cells	Lagging chromosomes in 90% of cells	5–15% (cell-line dependent)	Unreported	Unreported
Cell Lines Used	hTERT TP53-KO colon epithelial and retinal pigment epithelial cells	Embryonic kidney and colorectal carcinoma cells	Retinal pigment epithelial cells	Ovarian and gastric carcinoma, melanoma, and breast epithelial cells	Immortalized lung epithelial cells	Mammary epithelial cells
Reference(s)	[102]	[103]	[104]	[109]	[34]	[105]

Highlights.

• Recurrently observed aneuploidies across human cancers tend to be acquired early in tumorigenesis

• CRISPR-based tools allow targeted mis-segregation or deletion of individual chromosomes

• Isogenic cellular models that differ in a single aneuploidy can reveal insights into underlying aneuploidy biology and their contributions to tumorigenesis

• Multiple dosage-sensitive genes promote selection for aneuploid chromosomes in cancer