In the 20 years since human embryonic stem cells, and subsequently induced pluripotent stem cells, were first described, it has become apparent that during long-term culture these cells (collectively referred to as ‘pluripotent stem cells’ (PSCs)) can acquire genetic changes, which commonly include gains or losses of particular chromosomal regions, or mutations in certain cancer-associated genes, especially TP53. Such changes raise concerns for the safety of PSC-derived cellular therapies for regenerative medicine. Although acquired genetic changes may not be present in a cell line at the start of a research programme, the low sensitivity of current detection methods means that mutations may be difficult to detect if they arise but are present in only a small proportion of the cells. In this Review, we discuss the types of mutations acquired by human PSCs and the mechanisms that lead to their accumulation. Recent work suggests that the underlying mutation rate in PSCs is low, although they also seem to be particularly susceptible to genomic damage. This apparent contradiction can be reconciled by the observations that, in contrast to somatic cells, PSCs are programmed to die in response to genomic damage, which may reflect the requirements of early embryogenesis. Thus, the common genetic variants that are observed are probably rare events that give the cells with a selective growth advantage.
Although little more than 20 years has passed since the first human embryonic stem cells (ES cells) were reported1,2, and less than 14 years since human induced pluripotent cells (iPS cells) were described3,4, clinical trials for regenerative medicine using derivatives of these cells are already under way or on the horizon5,6,7,8,9. Yet over this period it has become evident that both human ES cells and iPS cells (collectively denoted here as ‘pluripotent stem cells’ (PSCs)), although mostly diploid when first derived, can acquire genetic alterations, ranging from large-scale structural modifications readily recognized as karyotypic variants through to single base pair changes on subsequent cell culture passages (Fig. 1). Although not the focus of this Review, epigenetic changes encompassing aberrations in DNA methylation, imprinting and X chromosome inactivation have also been reported in PSCs10.
The observation of genetic and epigenetic changes in PSCs has triggered worries about the significance of such variants for the safety of PSC-based regenerative medicine11,12. Such variants could also impact on other applications, such as in disease modelling and drug discovery. In particular, some recurrent genetic changes, for example mutations in TP53, or gains of the short arm of chromosome 12 (chromosome arm 12p), the long arm of chromosome 17 (chromosome arm 17q) and chromosome arm 20q, have been associated with various cancers. Notably, a gain of chromosome arm 12p has been associated with embryonal carcinoma cells, the malignant counterpart of PSCs13. Indeed, a planned trial of iPS cell-derived retinal pigment cells to treat age-related macular degeneration was halted when a point mutation was detected, although whether this particular mutation may have caused a problem was unknown8,14. Some of the genetic variants found in PSCs may well have been present in the embryos or somatic cells from which they were derived, or may have been induced during derivation15,16.
The outcomes of the currently pursued in-depth studies into the effect of the choice of a starting cell type and/or a method of reprogramming on the overall mutational burden in iPS cells17 will have an important bearing on the applications of PSCs in cell-replacement therapies and their use in basic research and drug discovery. However, regardless of these ultimate findings, a key feature of such ‘mutations of origin’ is that they will be present in all the cells of a given PSC line and so should be readily detectable and assessed for their potential impact before research with a particular line is initiated. Much more problematic, and the subject of this Review, are the ‘acquired mutations’ that were not present in the cell of origin but arise, often recurrently, during the culturing of the cells. Nevertheless, it is important to view these variants in perspective: many PSC lines do not acquire the commonly observed variants, or acquire them only in late culture passages. In a study by the International Stem Cell Initiative (ISCI), 79 of 122 cell lines retained a normal karyotype18. In another study, recurrent mutants of TP53 were detected in only 5 of 140 human PSC lines19. On the other hand, given that the sensitivity of the methods for detecting mutant cells in a mosaic culture is low (Box 1), such variant cells may lurk in cultures for a considerable time until they become prominent as a result of an acquired selective growth advantage or as a result of the cell line becoming subject to a population bottleneck. Furthermore, some genetic variants, for example gains of a small region of chromosome arm 20q, are particularly difficult to detect by G-banding karyotyping and may go unnoticed, even when present in a substantial proportion of the cells18.
Over the years, many studies have sought to find ways to minimize the appearance of genetic variants in PSCs. However, in such endeavours, it is important to recall that the appearance of the common recurrent variants is dependent on two independent events — mutation and subsequent selection (Fig. 1) — and these can be experimentally difficult to disentangle. The mechanisms of selective growth advantage, owing to the altered expression or activity of one or more genes (driver genes) or to the effects of culture conditions on selection, can be relatively easily analysed by introducing small numbers of variant cells into wild-type cell cultures and monitoring their subsequent growth patterns20. By contrast, mutation occurs at a very low frequency, so the appearance of mutations is difficult to monitor directly without expansion of the variant cells, in which case the estimates of mutation rate may be compromised by selection unless this is avoided by often cumbersome clonogenic assays21.
In this Review we discuss the nature of the acquired genetic variants that commonly arise in human PSC cultures and consider the potential consequences of these acquired genetic variants for research on human PSCs and for their clinical applications. We then discuss the mechanisms of selective growth advantage that lead to the recurrent appearance of particular variants. Finally, we focus on the underlying mechanisms of mutation in PSCs; it seems that PSCs differ substantially from somatic cells in both their susceptibility and their response to DNA damage, which may reflect the exigencies of rapid cell proliferation that sustains early embryo development.
Types of acquired genetic variation
Genetic variations can arise from a range of alterations to the genome (Fig. 1). In the following sections, we provide an overview of our current knowledge of the major types of recurrent culture-acquired abnormalities in PSCs.
Traditionally, routine screening of PSC lines for the identification of genetic changes has been performed mainly by cytogenetic and molecular methods that are capable of detecting numerical and structural aneuploidy rather than DNA sequence changes18,22. Consequently, karyotypic abnormalities are the most comprehensively catalogued genetic changes in PSCs to date. In our taking stock of the reports of karyotypic abnormalities in PSCs over the past two decades, it is clear that the aberrant PSC karyotypes can encompass virtually any type of an abnormality, including numerical aneuploidies, such as a whole chromosome gain (trisomy) or loss (monosomy), as well as structural aneuploidies, including interstitial duplications, deletions, inversions, amplifications and translocations18,22,23. That said, the distribution of chromosomal aberrations appears to be non-random, and certain types of variants are more commonly seen18,22,23,24.
The first apparent bias is towards gains rather than losses of chromosomal material. Indeed, it is estimated that more than 70% of all karyotypic abnormalities reported in ES cells are whole or partial chromosome gains, whereas only around 20% of reported abnormalities are losses of chromosomes or chromosomal material24. Losses of entire chromosomes are particularly rare, representing only around 2% of reported abnormalities in PSC cultures24. The under-representation of monosomies in PSCs concurs with the observation that cells in general tolerate gains of genetic material more readily than losses25. Both unbalanced and balanced translocations, that is, with or without the overt net gain or loss of chromosomal material, respectively, have also been reported, but unlike in certain haematological malignancies, for example, no common recurrent translocations or fusion genes have so far been associated with variant PSCs18,22,24,26. On the other hand, some chromosomes are rarely, if ever, reported as gained or lost in PSCs, including chromosomes 2, 4, 19 and 21 (refs18,24).
Finally, most striking is the observation of a consistent pattern of chromosomes affected by aneuploidy in PSCs, with most detected aberrations in PSC karyotypes representing gains of the whole or fragments of chromosomes 1, 12, 17, 20 and X18,24,26,27,28 (Table 1). Of these, a particularly insidious change is the frequent gain of a small, variable region located near the centromere of chromosome arm 20q (refs29,30). Often this gain is below the resolution of G-band karyotyping but, for example, it was noted by SNP array analysis in more than 20% of the cell lines in the ISCI study18,24. That the same repertoire of aneuploidies is observed across different ES cell and iPS cell lines and across different laboratories worldwide18 suggests an enhanced fitness of such variant cells, probably due to increased expression of one or more of the genes located on the amplified chromosomes31,32.
Point mutation screening has not yet become routine in PSC maintenance and, therefore, culture-acquired nucleotide changes in these cells remain largely unexplored. A couple of recent studies have investigated the presence and potential recurrence of cancer-related point mutations in PSCs. DNA from 140 ES cell lines provided by different laboratories worldwide was subjected to whole-exome sequencing19. After the filtering out of inherited polymorphisms and focusing on only variants that were present in a subset of cells, as suspected culture-acquired mutations, 28 of the 263 mosaic variants detected across the 140 lines were predicted to alter gene function. Of these, the tumour suppressor gene TP53 was the only gene in which mutations were detected in multiple cell lines, with six different missense mutations in five independent ES cell lines. All of the identified TP53 missense mutations affected cytosines of highly mutable CpG dinucleotides within four of the residues encoding the DNA-binding domain of p53, therefore rendering the mutant p53 inactive. Further TP53 mutations in both ES cells and iPS cells were uncovered by the leveraging of RNA-sequencing datasets from public repositories19,33, thereby establishing TP53 as a recurrently mutated gene in PSCs. As the observed TP53 mutations represent some of the most frequent mutations in cancer34 and are also known to cause a familial cancer predisposition disorder, Li–Fraumeni syndrome35,36, these findings have brought into focus the need to monitor PSCs for culture-acquired TP53 mutations.
Despite being the most prevalent point mutations, TP53 mutations are not the only recurrent point mutations arising in cancer-related genes on PSC expansion33. Recently, recurrent point mutations were detected in at least 22 other genes that were previously classified within the COSMIC Cancer Gene Census database as genes with a documented cancer-related activity, including CCND2, PCM1, MYH9, HIF1A, BCL9 and VHL33. Intriguingly, the mutational burden seems to differ between different pluripotent states; human PSCs in the naive state, representing the pluripotent state of the pre-implantation epiblast cells37, were estimated to carry four times more mutations than their primed counterparts, which correspond to the pluripotent state of the postimplantation epiblast33. As the samples of naive cells analysed in this study were obtained by ‘resetting’ existing primed PSCs using chemical inhibitors of several cellular pathways, rather than being derived directly from embryos, the mutational load differences may not be intrinsic to different cell states but might reflect a substantial selection pressure imposed on the cells during resetting to naive pluripotency. Supporting the latter view, the genes found to be mutated in the naive cells were in pathways affected by chemical inhibitors used in resetting primed cells to the naive state33.
The studies described above19,33 offered important insights into the mutational landscape of PSCs, albeit they focused on analysing a relatively small portion of the PSC genome (the exome, which represents only about 1% of the genome). Undoubtedly, uncovering the true extent and pattern of point mutations arising in PSCs will require much larger, ideally longitudinal datasets, and scrutiny of PSC sequence changes at a genome-wide level. Important for this endeavour will be the implementation of next-generation sequencing as a component of routine monitoring of PSC genomes. Currently, the turnaround time and cost of sequencing preclude its use as a routine screening method, but the ongoing technological developments, which are driving down the cost and the data processing time, make this a feasible prospect for the coming decade. Nonetheless, we must remain cognizant of the fact that the reliable detection of mutations is only the first step in managing culture-acquired genetic changes. A far more difficult hurdle is ascribing the functional meaning to the detected mutations and predicting their potential impact for applications of human PSCs in regenerative medicine and drug discovery.
Consequences for applications of PSCs
Experimentally derived PSCs are closely related to embryonal carcinoma cells, the malignant stem cells of teratocarcinomas, which occur predominantly as testicular germ cell tumours in young men13,38,39. Moreover, PSCs have the ability to produce teratomas when grown in immunodeficient mice. Together these observations have always provoked concerns that cancer presents a significant safety hazard for human PSC-based regenerative medicine. However, apart from cancer, the genetic variants of PSCs also have the potential to cause a wide range of effects on cellular physiology that could compromise the efficacy of derivative cells used in clinical applications, or the production of such cells, or indeed the use of PSCs in research, for example into disease mechanisms.
It is important to recognize the distinction between teratomas and teratocarcinomas. Teratomas are tumours containing differentiated cells without any persisting PSCs. By contrast, teratocarcinomas are tumours with the characteristics of teratomas that also contain undifferentiated PSCs40,41 (Fig. 2). Clinically, teratocarcinomas are highly malignant cancers, but they can also be effectively treated because PSCs are exceptionally sensitive to the chemotherapeutic agent cisplatin42,43,44 as part of a standard treatment that also includes bleomycin and etoposide45. Although some PSCs do produce teratocarcinomas in which undifferentiated PSCs can be recognized histologically, or by outgrowths of PSCs from explanted tumours46, many of the xenograft tumours derived from PSCs are better classified as teratomas47.
It might be anticipated that variant PSCs carrying mutations that enhance their proliferative potential and, perhaps, reduce their propensity to differentiate would be more likely to generate teratocarcinomas. Certainly, aneuploid PSCs can produce teratocarcinomas46. Furthermore, the transcriptomes of ES cells carrying an extra copy of chromosome 12 clustered more closely with embryonal carcinoma cells from germ cell tumours, which almost always exhibit a gain of chromosome arm 12p, whereas ES cells with a gain of chromosome 12 were more likely to produce teratocarcinomas than the parent diploid cells from which they were derived32. On the other hand, in a recent ISCI study, albeit limited in scope, teratocarcinomas were produced by PSCs without overt karyotypic abnormalities, whereas PSCs with such variants, including gains of chromosome 12, produced teratomas, indicating no clear correlation between the formation of teratocarcinomas and the presence of overt karyotypic changes47. These discrepancies point to the need for a more systematic study of the relationship between genotype and the ability of PSCs to form teratocarcinomas rather than teratomas.
Regenerative medicine involves transplantation of specific differentiated derivatives, not undifferentiated cells, so it is the possibility that genetic variants of PSCs may cause a neoplastic transformation of their derivative differentiated cells that is the greater concern (Fig. 2). Unfortunately, there is very little direct evidence from which to draw any definite conclusions about the extent of the risks. The somatic cells in teratomas of the laboratory mouse are almost always benign and non-tumorigenic39, and this may be generally true of human teratomas. However, pathologists with expertise in clinical gonadal teratocarcinomas do have concerns as in the human tumours, in contrast to those of the laboratory mouse, many of the differentiated elements, such as neural tubes, exhibit features of immaturity that may be regarded as potentially neoplastic41. Certainly, secondary somatic tumours derived from primary germ cell tumours have been found clinically, although they are very rare38. Experimental, PSC-derived teratomas often also contain elements of primitive endoderm, which is a further concern as yolk sac carcinoma, representing malignant primitive endoderm, is a well-known clinical form of germ cell tumours of the newborn48.
Although a relationship between malignant transformation of teratoma elements and particular genetic variants has not been established, some of the common karyotypic variants occurring in human PSCs are associated with other types of somatic cancer — for example, gains of chromosome arm 17q with neuroblastoma49 (Table 1). Furthermore, two of the genes associated with recurrent variants in human PSCs, TP53 (ref.19) and BCL2L1 (the driver gene of the chromosome arm 20q amplicon, see later)50, which derive their selective advantage for PSCs from their antiapoptotic functions, are associated with many cancers51,52. Although the driver genes of the other common recurrent variants of human PSCs have yet to be identified, they provide a selective advantage because of their specific effects on the undifferentiated PSCs, and it is entirely possible that their effects on specific differentiated derivatives may be quite different. However, as many of the recurrent variants involve gains or losses of large chromosomal regions, it is also possible that other hitchhiker genes linked to the driver gene may also cause effects in the differentiated derivatives, and that these effects are separate from those of the driver genes on the undifferentiated PSCs.
There has been very little systematic consideration of the risks posed by particular genetic variants of PSCs for the safety of clinical applications, or for their consequences for other uses, such as in disease modelling or drug discovery. These issues were discussed by international key opinion leaders at a meeting of the ISCI at the Jackson Laboratory in 2016 (ref.53), and again at a meeting hosted by Nature Research in London in 2018 (ref.54), while the Japanese regulatory authorities have issued some guidelines55.
However, there is no international consensus about potential risk assessment, and the ISCI meeting in 2016 suggested the establishment of an advisory group to collate information on the common genetic variants of PSCs, including any evidence of their effects on cell behaviour, and link that information to other cancer and disease-related genomic databases. Meanwhile, the Nature Research meeting strongly recommended that researchers clearly document any genetic variants that may have been present in cells used for particular research, so providing the data for future retrospective analysis of their potential consequences. Certainly, as a minimum, the documentation should include appropriate characterization of the karyotype of the cells, and also assessment of the chromosome 20 amplicon, given that these represent the most commonly observed genomic changes seen in cultured PSCs.
Selection drives genetic variation
In the following sections, we summarize the key phenotypic features of the common genetic variants of human PSCs and discuss the putative driver genes that underpin their phenotypes and provide a selective growth advantage.
Selective growth advantage
Although genetic variants may be occasionally fixed when PSC cultures are passed through a population bottleneck, such as cloning, the recurrence of specific mutations within PSC populations suggests that such genetic changes endow the variant cells with a selective growth advantage. Consistent with this, the proportions of variant cells in a culture typically increase over time from when they are first detected22,56,57 (Fig. 1). Similarly, in experiments designed to recapitulate the takeover of cultures by variant clones, co-mixing a small proportion of commonly occurring genetic variants with their wild-type counterparts led to a gradually increased representation of variant cells in subsequent culture passages, until they eventually dominated the cultures20,50. Commonly, a variant may be first detected when it constitutes around 5−10% of the cells in a culture, rising rapidly to 100% in as few as five passages. On the basis of these longitudinal evaluations, the takeover of PSC cultures by variant cells has been likened to Darwin’s principle of natural selection, whereby the variant PSCs that are best adapted to particular selective conditions outcompete their neighbours and populate cultures with their own progeny. The specific phenotypic features associated with genetic variants hold clues as to the selective pressures operating in PSC cultures, the reduction of which is key to minimizing the appearance of genetic variants in expanding PSC populations.
In principle, genetically variant PSCs could gain a selective advantage by acquiring one or several of the following features: a proliferative advantage underpinned by faster cell cycle time, a decreased rate of differentiation, or altered pattern of differentiation, or an increased rate of survival (Fig. 3). Indeed, a number of studies have reported that such traits typify variant cells harbouring the commonly acquired aneuploidies. For example, the growth advantage of trisomy 12 PSCs was attributed mainly to their significantly reduced cell cycle time, although the variant cells also displayed an increased resistance to apoptosis and a reduced tendency for differentiation32.
With regard to the reduced propensity for differentiation, no studies have so far reported a total block of variant ES cells or iPS cells to differentiation, although nullipotent embryonal carcinoma cells, which are incapable of differentiation, are well known in the context of testicular germ cell tumours58,59. Rather, either a reduced differentiation capacity60 or a delayed differentiation dynamic61 in comparison with wild-type cells has been observed. In some instances, genetically variant PSCs appeared to yield alternative cell types to wild-type cells exposed to the same set of differentiation conditions. For example, the same differentiation protocol applied to wild-type ES cells and variants with a gain of chromosome arm 17q resulted in mesodiencephalic dopaminergic neurons or dorsal telencephalic neurons, respectively62. Given that this gain entails amplification of a large chromosomal region and, hence, increased expression of most of the genes in that region31, it is easy to envision that such extensively altered gene and protein expression profiles could include changes that skew the differentiation trajectory of cells. In this case, the skewed differentiation was attributed to an increased expression of the WNT3 and WNT9B genes localized in the amplified part of chromosome 17 (ref.62). In another case, it has been reported that BCL-XL overexpression perturbs SMAD and transforming growth factor-β signalling in PSCs with the chromosome band 20q11.21 gain, resulting in impaired neurectoderm differentiation63. Although the altered propensity for differentiation of variant cells may be a mere consequence of hitchhiker genes rather than the driver of their growth advantage, the converse may also be true if the differentiation process itself exerts selection on the differentiating cells. For example, in one study, cardiac differentiation favoured cells with a gain of chromosome arm 20q (ref.64), whereas another study19 reported an enrichment of mutant TP53 cells on PSC differentiation. In both cases, a variant PSC population was already present in the starting cultures before differentiation, but it is also possible that variant cells may arise and be selected during the differentiation process itself.
Although faster cell cycle and altered differentiation have been associated with some of the recurrent variants, resistance to apoptosis seems to be a frequent feature of variants commonly detected in PSC cultures. This is perhaps not surprising given that marked sensitivity to apoptosis is one of the notable features of early-passage diploid PSCs. Excessive cell death is particularly prominent when PSCs are grown at a low cell density65, a condition under which single PSCs are confronted with a series of bottlenecks preventing their clonal growth66. At the molecular level, the propensity for apoptosis has been explained by a low apoptotic threshold of PSCs, associated with low expression levels of antiapoptotic proteins and high expression levels of proapoptotic proteins67. In addition to preferential expression of proapoptotic factors, PSCs store a constitutively active proapoptotic factor, BAX, in the Golgi apparatus68. This effectively primes PSCs for a rapid apoptotic response to appropriate cues. Apart from hampering the efficient scale-up of PSCs, the severe reduction in cell numbers during culture clearly creates conditions for selection of genetically variant cells capable of blunting the apoptotic pathways19,50.
The emergence of variant cells in PSC cultures inevitably entails interactions of variants with their wild-type counterparts, as the two populations share their environment and some of their cell–cell contacts. The nature of these interactions can determine the fate of wild-type cells in a non-cell-autonomous manner, thereby impacting on the dynamics of the variant’s overtaking of cultures (Fig. 3). Some of the commonly occurring PSC variants were shown to suppress the growth of wild-type populations by inducing apoptosis in their neighbouring wild-type cells69 in a manner similar to the phenomenon of cell competition described in other model systems70. In PSCs cultures, a differential sensitivity of wild-type and variant PSCs to mechanical pressures imposed by cell crowding allowed variants to effectively eliminate wild-type cells from mixed cultures, thereby enhancing the ability of variants to rapidly achieve clonal dominance69. Therefore, consideration of cell interactions, in addition to cell-autonomous mechanisms, is needed in developing effective strategies for prevention of growth supremacy of variant cells.
Driver genes that provide a growth advantage
The simplest working hypothesis to account for the recurrent selection of a particular chromosomal variant is that it is the altered expression of a single driver gene located in the variant region that provides a growth advantage by altering a cell’s behaviour in response to proliferation, differentiation or cell death cues (Fig. 3). It is, of course, possible that interaction of multiple linked genes in a particular chromosomal rearrangement, or indeed alterations to the chromatin architecture itself, may be responsible. Nevertheless, most studies have focused on seeking a single driver gene.
Often the size of the genomic region affected is too large to home in on a likely candidate, but in the case of amplifications affecting chromosome 20, a common minimal amplicon of 0.55 Mb was identified in the pericentromeric region of the long arm in all reported examples18. Within this minimal amplicon, containing only 13 annotated genes, BCL2L1 was a likely candidate driver gene as its antiapoptotic splice variant, BCL-XL, is expressed in human PSCs18. Experiments in which cells carrying a gain of chromosome 20 or that had been transfected with a BCL2L1-overexpressing vector were mixed with diploid cells confirmed that BCL2L1 and its BCL-XL product were indeed the driver providing a selective growth advantage by blocking apoptosis28,50.
Like chromosome 20, a common minimal amplicon has also been identified on chromosome arm 1q (ref.24). A likely candidate driver gene located in this region is MDM4, which regulates p53 by suppressing its response to cellular stresses and increasing the threshold to apoptosis71. As recurrent dominant negative mutations of TP53 provide a growth advantage to human PSCs19, it is possible that dysregulation of other genes, such as MDM4, that affect apoptosis through p53 confer a similar growth advantage. On chromosome arm 17q, another antiapoptotic gene, BIRC5 (which encodes survivin), encoded in chromosome band 17q25.3, has been proposed to increase resistance to apoptosis as its inhibition leads to apoptosis of human PSCs and cancer cells72,73,74,75. On the other hand, another possible driver gene encoded on chromosome 17 is WNT3, suggested by its involvement in the enhanced proliferation of cells carrying a gain of chromosome arm 17q (ref.62).
Interest in gains of chromosome 12 has a long history because testicular germ cell tumours almost always have a gain of the short arm, mostly as an isochromosome76 or more rarely as an interstitial amplification77,78. However, there is no definitive evidence that identifies the specific driver gene either for the progression of germ cell tumours or for the appearance of variant human PSCs with a gain of all or part of chromosome 12. An obvious candidate driver gene on chromosome arm 12p is NANOG given its central role in maintaining pluripotency and given that its overexpression inhibits differentiation79. Moreover, overexpression of NANOG does allow human ES cells to efficiently form colonies at low density, which is normally associated with extensive apoptosis, perhaps mediated by downregulating LGALS1, which normally promotes apoptosis, and upregulating HSPA1A, which inhibits apoptosis80. Indeed, NANOG is located in a minimal amplicon that has been identified in germ cell tumours, chromosome band 12p13.31 (that is, 13.31 identifies a small region of the short arm of chromosome 12 characterized by the staining pattern of the metaphase chromosome), but so are two other genes, DPPA3 and GDF3, that also may affect the behaviour of human PSCs78. However, a different minimal amplicon has also been reported in human germ cell tumours, at 12p11.2–p12.1, in which a number of other genes have been highlighted, such as the oncogene KRAS77.
Although less frequent, deletions may promote enhanced survival through copy number loss of proapoptotic genes. The BCL-2 apoptotic pathway is controlled by interactions between proapoptotic and antiapoptotic protein family members81. Human ES cells show elevated expression of the proapoptotic genes NOXA (also known as PMAIP1), BIK, BIM (also known as BCL2L11), BMF and PUMA (also known as BBC3), which may contribute to their low apoptotic threshold67,68,82. The two of these most highly expressed in human PSCs, NOXA and BIK, are located in chromosomal regions, 18q21.32 and 22q13.2, that do undergo recurrent deletion. Deletion of NOXA by genetic manipulation decreases the sensitivity of human PSCs to mitotic errors, thereby increasing the survival of the aneuploid cells83, and increases survival during cell dissociation, similar to the overexpression of the antiapoptotic proteins, BCL-2 and BCL-XL65,84.
Mutation rate in PSCs
Mutations occur stochastically and at low frequency in single cells within much larger cell populations, so by the time they become detectable, the frequency of mutation may have been grossly distorted by the effects of selection. To overcome this problem, we recently adopted a clonogenic strategy in which a single cell was isolated and allowed to expand as a clonal colony for a fixed time, after which the clone was subcloned, with about 20 subclones being isolated and, after expansion, subjected to whole-genome sequencing21. Using this approach, in which most of the mutants that arose were in genes and locations unlikely to result in growth advantage or disadvantage, we estimated the mutation rate of two human, clinical grade ES cell lines, MShef4 and MShef11, as 0.37 × 10−9 and 0.28 × 10−9 single-nucleotide variants (SNVs) per base pair, per day, respectively, equating to approximately 0.30 × 10−9 and 0.23 × 10−9 SNVs per cell division, respectively, given that the cell cycle time of human PSCs, in our experience, is approximately 20 hours (±2 hours)66. This rate was not affected by the use of the Rho-associated coiled-coil-containing protein kinase (ROCK) inhibitor Y-27632 commonly used in human PSC cultures85. The frequency of indels was tenfold lower. These low rates are comparable with the estimated mutation rate in another study of human iPS cells of 0.18 × 10−9 SNVs per base pair, per cell division, which was considerably lower than in the endothelial cells from which the iPS cells were derived16. These mutation rates in human PSCs contrast with an estimated much higher rate of 2.66 × 10−9 mutations per base pair per mitosis in somatic cells86. In another, more limited study of a human ES cell line, a slightly higher mutation rate of 1 × 10−9 SNVs per base pair per cell division was found, but again this was much lower than an estimate for a corresponding somatic cell in the same study87, whereas in a study of a single locus, Aprt, in mouse ES cells, the mutation rate was estimated to be tenfold lower than in corresponding somatic cells88. These low rates are consistent with the infrequency of recurrent point mutations observed in PSC lines: for example, it was observed that only 5 of 140 human PSC lines carried mutations in TP53 (ref.19).
In our study of the MShef4 and MShef11 human ES cell lines21, the mutation rate was similar across all chromosomes, with no obvious hotspots, with the exception of a slightly raised rate on the X chromosome, which might have been a consequence of both lines being male. Nevertheless, the mutation rate was significantly higher in intergenic regions than in exons and introns, suggesting an influence of chromatin structure on mutation. Furthermore, the predominant mutation signatures that we detected were consistent with oxidative damage being the predominant cause of mutation and, indeed, the mutation rate for both SNVs and indels was reduced by about 50% when the cells were maintained under low-oxygen (5%) atmospheres.
Mechanisms of mutation
Whereas the mechanisms by which genetic variants offer cells a selective growth advantage are relatively easy to assess and have been extensively studied, addressing the mechanisms that drive the appearance of these variants in the first instance is more problematic. It now appears that the common recurrence of particular genetic variants in PSCs despite their low mutation rate reflects the mechanisms used to maintain genetic integrity in embryonic cells in contrast to those used by somatic cells.
DNA replication stress and mitotic errors
While many SNVs in PSCs, as in other cultured cells89,90,91, are caused by misincorporation of bases due to oxidative stress, the relatively rapid cell cycle of PSCs might also expose them to high levels of DNA replication stress, characterized by reduced rates of DNA replication together with stalling and collapse of replication forks92,93. Errors in the repair of resulting DNA double-strand breaks (DSBs) could then lead to chromosomal rearrangements94. Self-renewal of human PSCs is characterized by an abbreviated G1 phase that bypasses the RB1–E2F checkpoint due to the high expression of cyclin D2 and its CDK4 partner together with the constitutive expression of cyclin E, which together maintain RB1 in a hyperphosphorylated and inactive state95,96,97. By DNA fibre assays, human PSCs were found to exhibit the features of DNA replication stress in comparison with isogenic somatic cells. The features included slower DNA replication speed, stalled replication forks and replication starting from dormant replication origins98 (Fig. 4). When compared with somatic cells, PSCs also display more extensive replication-associated DNA damage99,100.
A similar situation pertains in many cancers where cyclin E is frequently overexpressed and RB1–E2F is constitutively activated101. One of the consequences of this is replication stress, DSBs and genetic instability102,103,104,105. In mouse ES cells, also, molecular hallmarks of replication stress are almost identical to those observed when oncogenes, such as CCNE encoding cyclin E, are dysregulated in somatic cells106, suggesting that atypical cell cycle control with consequent susceptibility to DNA replication stress and genomic damage in PSCs parallels the oncogene-induced DNA damage model for cancer development and progression107.
Replication stress induced by oncogene expression can lead to nucleotide deficiency and collision of replication forks with transcription complexes102,108. Supplementing cancer cells or primary cell lines that overexpress oncogenes, such as CCNE, with nucleosides has been found to alleviate replication stress and its associated DNA damage and genetic instability in these cases102,103. In a similar manner, it was recently found that exogenous nucleosides increase the rate of replication fork progression and decrease DNA damage in human PSC cultures98 (Fig. 4).
While chromosomal non-dysjunction and numerical instabilities may be the product of merotelic kinetochore attachment, in which the microtubules from both poles bind to the same sister chromatid, leading to lagging and potential missegregation of chromosomes109, the persistence of DNA replication defects from S phase into mitosis can also result in the formation of mitotic errors that are a source of chromosomal instabilities103. Under-replicated regions can interlink sister chromatids during segregation, forming anaphase bridges that are prone to breakage, forming DSBs110. Often, to prevent anaphase bridges, nucleases cleave the DNA, which again generates DSBs111. Furthermore, during condensation, chromosomes that harbour replication intermediates are particularly prone to breakage112. These DSBs that result from replication intermediates in mitosis are the substrates for genetic instability caused by error-induced repair.
By fluorescent labelling of human PSCs with histone H2B–mCherry it was observed that 30% of mitoses were abnormal, including a high proportion with lagging chromosomes and anaphase bridges83, a level substantially higher than that observed in somatic cell lines113. In comparison with somatic cell lines, diploid human PSCs show condensation defects that result in partially condensed and entangled chromosomes113. Supplementation of cultures with exogenous nucleosides alleviated replication stress and decreased the frequency of mitotic errors, providing further evidence that these are linked in human PSCs, as well as providing an approach for reducing their appearance in PSC cultures98. However, the continued occurrence of mitotic errors, even with the addition of nucleosides, suggests that there are other factors driving their occurrence.
Response to genomic damage
Human PSCs deploy a number of mechanisms to minimize the effective mutation rate that otherwise might be anticipated from their high susceptibility to DNA damage (Box 2). Genes involved in various DNA damage repair pathways show increased expression compared with those in somatic cells114,115, and nucleotide excision repair, base excision repair and the resolution of interstrand crosslinks caused by ionizing radiation have all been reported to be faster in human PSC lines than in somatic cell lines114,116,117. PSCs also tend to repair DSBs using homologous recombination, which is prone to fewer errors than non-homologous end joining118, although they do also use a higher-fidelity system of non-homologous end joining that is independent of DNA-dependent protein kinase catalytic subunit (DNA-PKcs) and ataxia telangiectasia mutated (ATM)119. Further, in response to the formation of reactive oxygen species as a by-product of respiration, and consequent oxidative stress, PSCs express higher levels of the antioxidant enzymes SOD2 and GPX2 compared with differentiated cell lines120.
Nevertheless, human PSCs generally activate apoptosis when exposed to lower doses of genotoxic insults than do somatic cells, suggesting that a low apoptotic threshold is the key element in their response to genomic damage. After human PSCs are exposed to ultraviolet C radiation to induce nucleotide base adducts or DNA breaks, they respond with extensive apoptosis even at mild doses that have little effect on somatic cell lines99,116,117. Similarly, the treatment of human PSCs with cisplatin (a DNA crosslinking agent) or thymidine to initiate replication block121,122 or with nocodazole to induce mitotic block83 also elicits an extensive apoptotic response, in contrast to the response by somatic cells, while PSCs also efficiently activate apoptosis in response to oxidative stress120. The particular sensitivity of embryonal carcinoma cells, the malignant PSCs of teratocarcinomas, to drugs such as cisplatin42,43 makes germ cell tumours one of the most treatable forms of solid cancer, most likely reflecting this particular low apoptotic threshold.
An atypical cell cycle checkpoint control mechanism most likely underlies the low apoptotic threshold of human PSCs. In response to DNA damage, human PSCs fail to activate p21, which is normally required to execute the G1/S checkpoint, providing less time for repair before apoptosis is initiated in a p53-dependant manner115,117,123. Furthermore, in response to DNA replication stress caused by high levels of thymidine or the presence of cisplatin, human PSCs, unlike somatic cells, do not activate ATR–CHK1, while foci of RPA, which binds to single-stranded DNA at stalled replication forks, are not formed: instead the cells commit to apoptosis121,122. Human PSCs also undergo extensive apoptosis in response to mitotic stress, which may safeguard the genome from abnormal mitosis by clearing the affected cell from the cell pool83.
Collectively, these studies support a model in which genomic stability, and the particularly low observed mutation rate of PSCs, is primarily maintained by a low apoptotic threshold. Consequently, blocking apoptosis seems to be the most likely mechanism that provides selective growth advantage for the common genetic variants found in human PSCs: the two driver genes so far identified, TP53 and BCL2L1, both act to inhibit apoptosis, while other proposed candidates, MDM4 and BIRC5, are also antiapoptotic.
This low apoptotic threshold of human PSCs may reflect their relationship with the early embryo, in which the need for rapid cell doubling is accomplished by the lack of cell cycle checkpoints, rendering the cells particularly susceptible to errors in DNA synthesis and mitosis, which could be catastrophic for subsequent embryonic development. Indeed, almost half of human embryos fail to survive due to chromosomal instability, which does not seem to be a mere artefact of in vitro fertilization124,125,126. It has been observed that the mosaic embryos that survive to the blastocyst stage undergo ‘genetic normalization’ when cultured under routine in vitro fertilization conditions127. The mechanism of genetic normalization is still widely debated, although, in the mouse, activation of apoptosis during the later pre-implantation stages may allow the removal of aneuploid cells from the developing embryo128,129. This model is supported by observations that the proportion of aneuploidy in the inner cell mass is reduced, whereas in the trophectoderm it is enriched as development proceeds130. However, this is still widely debated as apoptosis is a feature of all embryos and may be a mechanism for maintaining cellular homeostasis regardless of their genomic state131.
Response to the DNA damage induced by genome editing
Another context in which the apoptotic response of human PSCs to DNA damage induction is of central importance is the process of genome editing. The rising prominence of genome editing technologies, in particular CRISPR−Cas9-based methods, has fuelled efforts aimed at, for example, correcting germline mutations in PSCs to allow autologous cell therapy or removing HLA antigens to reduce the need for immunosuppressants in patients who have received a transplant. Crucially, as gene editing relies on the induction of DNA DSBs by nucleases, the edited PSCs undergo high levels of apoptosis in comparison to their unedited counterparts132. The rate of cell death was shown to be similar between different edited PSCs, regardless of whether the targeted gene was expressed in PSCs or whether it was not expressed and was dispensable for PSC maintenance. This observation supports the view that the induction of DSBs during the gene editing process commits PSCs to apoptosis132. Mechanistically, DSB induction by Cas9 was shown to trigger differential gene expression in edited cells, most notably by promoting changes in gene expression associated with activation of the p53 pathway132. In line with the importance of p53 activation during the genome editing process, genome editing in PSCs with genetically inactivated TP53 reduced the levels of cell death and increased the efficiency of PSC genome editing132. While genetic inactivation of p53 is deemed too risky for editing of cells destined for clinical use, transient p53 inactivation has been proposed as a possible alternative133. Further work will need to carefully address this possibility to ensure that the transient p53 inactivation does not inadvertently select for TP53 genetic mutants. Such appearance of TP53-inactivating mutations during the CRISPR−Cas9 genome editing process was recently observed in cancer cell lines134.
Conclusions and perspectives
Efforts to collect and catalogue genetic variation in PSCs over the past two decades have demonstrated that particular genetic variants do arise in cell cultures and are sometimes difficult to detect because of the limitation of sensitivity of detection methods. Nonetheless, it is reassuring that the rate of mutation in PSC cultures is low compared with that in somatic cells. Indeed, on the basis of the data from large-scale retrospective studies, such as the ISCI study18, and direct measurements of the mutation rates in PSCs21, there is no evidence to suggest that PSC genomes are particularly unstable. Rather, the clonal expansion of genetic variants against the backdrop of low mutation rates can be explained by the effect of selective pressures operating in PSC cultures. Optimizing culture conditions and protocols to minimize the growth advantages of the common variants is, therefore, a key route to maintaining the genetic integrity of PSC lines. However, the predominant selective force dominating PSC cultures appears to be a high rate of apoptosis66,68. Apoptosis seems to be a default fate choice of PSCs in many different scenarios, including a response of cells to genome damage or mitotic stress68,83,121,122, most likely reflecting its function of maintaining the genetic integrity of the early embryo. Consequently, optimizing culture conditions should entail removing the apoptotic stimuli, but not blocking apoptosis per se, which would be counterproductive.
Armed with knowledge of recurrent karyotypic and sequence changes, our attention now needs to turn to finding ways of minimizing their occurrence by lowering the probability of genome damage and reducing the selective pressures. In that respect, the observations that mutation rates can be decreased by growing cells under low-oxygen conditions21 and that replication stress-induced genome damage can be alleviated by addition of exogenous nucleosides98 provide foundations for optimized culture conditions of PSCs. Further work should also address the contribution of epigenetic variants to aberrant PSC phenotypes, as the understanding of epigenetic variation in PSC cultures remains limited.
Finally, the field awaits deciphering of the functional consequences of the karyotype and sequence changes for PSC traits and for the behaviour of the differentiated derivatives. Interpreting the role of specific variants is complicated by the fact that their consequences are likely to be context dependent. For example, a mutation in a gene expressed specifically in an endodermal lineage may have little impact on the clinical application of neuronal cells. To aid these analyses, ISCI is proposing an international study group to collate and monitor evidence of genetic variants in PSCs and their potential consequences53. However, the success of this approach will require a concerted effort within the field to perform routine monitoring and report the presence of genetic variants, thereby allowing retrospective analyses of their effects. We envisage that these initiatives, in synergy with cancer genome efforts, will provide a rational strategy to assess the potential risk of different mutations, a necessary requirement for routine, safe clinical implementation of cellular therapies.
Thomson, J. A. et al. Embryonic stem cell lines derived from human blastocysts. Science 282, 1145–1147 (1998).
Reubinoff, B. E., Pera, M. F., Fong, C. Y., Trounson, A. & Bongso, A. Embryonic stem cell lines from human blastocysts: somatic differentiation in vitro. Nat. Biotechnol. 18, 399–404 (2000).
Takahashi, K. et al. Induction of pluripotent stem cells from adult human fibroblasts by defined factors. Cell 131, 861–872 (2007).
Yu, J. et al. Induced pluripotent stem cell lines derived from human somatic cells. Science 318, 1917–1920 (2007).
da Cruz, L. et al. Phase 1 clinical study of an embryonic stem cell-derived retinal pigment epithelium patch in age-related macular degeneration. Nat. Biotechnol. 36, 328–337 (2018).
Schwartz, S. D. et al. Embryonic stem cell trials for macular degeneration: a preliminary report. Lancet 379, 713–720 (2012).
Song, W. K. et al. Treatment of macular degeneration using embryonic stem cell-derived retinal pigment epithelium: preliminary results in Asian patients. Stem Cell Rep. 4, 860–872 (2015).
Mandai, M., Kurimoto, Y. & Takahashi, M. Autologous induced stem-cell-derived retinal cells for macular degeneration. N. Engl. J. Med. 377, 792–793 (2017).
Barker, R. A., Parmar, M., Studer, L. & Takahashi, J. Human trials of stem cell-derived dopamine neurons for Parkinson’s disease: dawn of a new era. Cell Stem Cell 21, 569–573 (2017).
Bar, S. & Benvenisty, N. Epigenetic aberrations in human pluripotent stem cells. EMBO J. https://doi.org/10.15252/embj.2018101033 (2019).
Yasuda, S. et al. Tumorigenicity-associated characteristics of human iPS cell lines. PLoS ONE 13, e0205022 (2018).
Sato, Y. et al. Tumorigenicity assessment of cell therapy products: the need for global consensus and points to consider. Cytotherapy 21, 1095–1111 (2019).
Andrews, P. W. From teratocarcinomas to embryonic stem cells. Philos. Trans. R. Soc. Lond. B Biol. Sci. 357, 405–417 (2002).
Garber, K. RIKEN suspends first clinical trial involving induced pluripotent stem cells. Nat. Biotechnol. 33, 890–891 (2015).
Hussein, S. M. et al. Copy number variation and selection during reprogramming to pluripotency. Nature 471, 58–62 (2011).
Rouhani, F. J. et al. Mutational history of a human cell lineage from somatic to induced pluripotent stem cells. PLoS Genet. 12, e1005932 (2016).
Steichen, C., Hannoun, Z., Luce, E., Hauet, T. & Dubart-Kupperschmitt, A. Genomic integrity of human induced pluripotent stem cells: reprogramming, differentiation and applications. World J. Stem Cell 11, 729–747 (2019).
Amps, K. et al. Screening ethnically diverse human embryonic stem cells identifies a chromosome 20 minimal amplicon conferring growth advantage. Nat. Biotechnol. 29, 1132–1144 (2011). Amps et al. report a study of a large panel of cells lines by an international consortium, using karyotype and SNP array analyses to survey the range of genetic variants commonly recurring in human PSCs.
Merkle, F. T. et al. Human pluripotent stem cells recurrently acquire and expand dominant negative P53 mutations. Nature 545, 229–233 (2017). The work by Merkle et al. provides the first demonstration of point mutations that occur recurrently in a single gene in human PSCs, resulting in variants of a tumour suppressor that provide a growth advantage by reducing sensitivity to apoptosis.
Olariu, V. et al. Modeling the evolution of culture-adapted human embryonic stem cells. Stem Cell Res. 4, 50–56 (2010).
Thompson, O. et al. Low rates of mutation in clinical grade human pluripotent stem cells under different culture conditions. Nat. Commun. 11, 1528 (2020). Using a clonogenic strategy, Thompson et al. report the low mutation rate in human PSCs and show that this rate is reduced by growth under low-oxygen conditions.
Draper, J. S. et al. Recurrent gain of chromosomes 17q and 12 in cultured human embryonic stem cells. Nat. Biotechnol. 22, 53–54 (2004).
Taapken, S. M. et al. Karotypic abnormalities in human induced pluripotent stem cells and embryonic stem cells. Nat. Biotechnol. 29, 313–314 (2011).
Baker, D. et al. Detecting genetic mosaicism in cultures of human pluripotent stem cells. Stem Cell Rep. 7, 998–1012 (2016). Baker et al. provide a comprehensive comparison of the sensitivity of commonly used methods for detection of genetic changes in PSCs, providing the basis for interpreting genetic tests used for assessment of PSC cultures.
Torres, E. M., Williams, B. R. & Amon, A. Aneuploidy: cells losing their balance. Genetics 179, 737–746 (2008).
Assou, S. et al. Recurrent genetic abnormalities in human pluripotent stem cells: definition and routine detection in culture supernatant by targeted droplet digital PCR. Stem Cell Rep. 14, 1–8 (2020).
Baker, D. E. et al. Adaptation to culture of human embryonic stem cells and oncogenesis in vivo. Nat. Biotechnol. 25, 207–215 (2007).
Nguyen, H. T., Geens, M. & Spits, C. Genetic and epigenetic instability in human pluripotent stem cells. Hum. Reprod. Update 19, 187–205 (2013).
Lefort, N. et al. Human embryonic stem cells reveal recurrent genomic instability at 20q11.21. Nat. Biotechnol. 26, 1364–1366 (2008).
Spits, C. et al. Recurrent chromosomal abnormalities in human embryonic stem cells. Nat. Biotechnol. 26, 1361–1363 (2008).
Enver, T. et al. Cellular differentiation hierarchies in normal and culture-adapted human embryonic stem cells. Hum. Mol. Genet. 14, 3129–3140 (2005).
Ben-David, U. et al. Aneuploidy induces profound changes in gene expression, proliferation and tumorigenicity of human pluripotent stem cells. Nat. Commun. 5, 4825 (2014). Ben-David et al. provide a detailed analysis of how gains of chromosome 12, one of the common chromosomal variants seen in human PSCs and in germ cell tumours, affect the transcriptome and the growth patterns of human PSCs.
Avior, Y., Eggan, K. & Benvenisty, N. Cancer-related mutations identified in primed and naive human pluripotent stem cells. Cell Stem Cell 25, 456–461 (2019).
Kandoth, C. et al. Mutational landscape and significance across 12 major cancer types. Nature 502, 333–339 (2013).
Malkin, D. et al. Germ line p53 mutations in a familial syndrome of breast cancer, sarcomas, and other neoplasms. Science 250, 1233–1238 (1990).
Srivastava, S., Zou, Z. Q., Pirollo, K., Blattner, W. & Chang, E. H. Germ-line transmission of a mutated p53 gene in a cancer-prone family with Li-Fraumeni syndrome. Nature 348, 747–749 (1990).
Nichols, J. & Smith, A. Naive and primed pluripotent states. Cell Stem Cell 4, 487–492 (2009).
Mostofi, F. K. & Price, E. B. Tumors of the Male Genital System, Atlas of Tumor Pathology, Second Series. (Armed Forces Institute of Pathology, 1973).
Damjanov, I. & Solter, D. Experimental teratoma. Curr. Top. Pathol. 59, 69–130 (1974).
Damjanov, I. & Andrews, P. W. The terminology of teratocarcinomas and teratomas. Nat. Biotechnol. 25, 1212 (2007).
Damjanov, I. & Andrews, P. W. Teratomas produced from human pluripotent stem cells xenografted into immunodeficient mice - a histopathology atlas. Int. J. Dev. Biol. 60, 337–419 (2016).
Einhorn, L. H. & Donohue, J. Cis-diamminedichloroplatinum, vinblastine, and bleomycin combination chemotherapy in disseminated testicular cancer. Ann. Intern. Med. 87, 293–298 (1977).
Oosterhuis, J. W., Andrews, P. W., Knowles, B. B. & Damjanov, I. Effects of cis-platinum on embryonal carcinoma cell lines in vitro. Int. J. Cancer 34, 133–139 (1984).
Einhorn, L. H., Nagy, C., Furnas, B. & Williams, S. D. Nabilone: an effective antiemetic in patients receiving cancer chemotherapy. J. Clin. Pharmacol. 21, 64S–69S (1981).
Williams, S. D. et al. Treatment of disseminated germ-cell tumors with cisplatin, bleomycin, and either vinblastine or etoposide. N. Engl. J. Med. 316, 1435–1440 (1987).
Andrews, P. W. et al. Embryonic stem (ES) cells and embryonal carcinoma (EC) cells: opposite sides of the same coin. Biochem. Soc. Trans. 33, 1526–1530 (2005).
Allison, T. F. et al. Assessment of established techniques to determine developmental and malignant potential of human pluripotent stem cells. Nat. Commun. 9, 1925 (2018).
Cunningham, J. J., Ulbright, T. M., Pera, M. F. & Looijenga, L. H. Lessons from human teratomas to guide development of safe stem cell therapies. Nat. Biotechnol. 30, 849–857 (2012).
Plantaz, D. et al. Gain of chromosome 17 is the most frequent abnormality detected in neuroblastoma by comparative genomic hybridization. Am. J. Pathol. 150, 81–89 (1997).
Avery, S. et al. BCL-XL mediates the strong selective advantage of a 20q11.21 amplification commonly found in human embryonic stem cell cultures. Stem Cell Rep. 1, 379–386 (2013). Avery et al. report the identification of BCL2L1 as the driver gene on chromosome band 20q11.21, which is commonly amplified in PSCs.
Beroukhim, R. et al. The landscape of somatic copy-number alteration across human cancers. Nature 463, 899–905 (2010).
Hainaut, P. & Hollstein, M. in Advances in Cancer Research Vol. 77 (eds Vande Woude, G. F. & Klein, G.) 81–137 (Academic Press, 1999).
Andrews, P. W. et al. Assessing the safety of human pluripotent stem cells and their derivatives for clinical applications. Stem Cell Rep. 9, 1–4 (2017).
Technologies, S. Challenges in Ensuring hPSC Quality. Stemcell.com, https://www.stemcell.com/nature-research-roundtable-hPSC-quality (2018).
Ministry of Health, Labour and Welfare. English translation of Annex of Notification 0613-3. http://www.nihs.go.jp/cbtp/sispsc/pdf/Eg.ver.Annex_0613-3_2016.pdf (2016).
Catalina, P. et al. Human ESCs predisposition to karyotypic instability: is a matter of culture adaptation or differential vulnerability among hESC lines due to inherent properties? Mol. Cancer 7, 76 (2008).
Imreh, M. P. et al. In vitro culture conditions favoring selection of chromosomal abnormalities in human ES cells. J. Cell Biochem. 99, 508–516 (2006).
Andrews, P. W., Bronson, D. L., Benham, F., Strickland, S. & Knowles, B. B. A comparative study of eight cell lines derived from human testicular teratocarcinoma. Int. J. Cancer 26, 269–280 (1980).
Andrews, P. W., Goodfellow, P. N., Shevinsky, L. H., Bronson, D. L. & Knowles, B. B. Cell-surface antigens of a clonal human embryonal carcinoma cell line: morphological and antigenic differentiation in culture. Int. J. Cancer 29, 523–531 (1982).
Fazeli, A. et al. Altered patterns of differentiation in karyotypically abnormal human embryonic stem cells. Int. J. Dev. Biol. 55, 175–180 (2011).
Werbowetski-Ogilvie, T. E. et al. Characterization of human embryonic stem cells with features of neoplastic progression. Nat. Biotechnol. 27, 91–97 (2009). The study by Werbowetski-Ogilvie et al. demonstrates a profound effect of some of the recurrent genetic changes on PSC growth rates and the ability of PSCs to differentiate to specific lineages.
Lee, C. T. et al. Functional consequences of 17q21.31/WNT3-WNT9B amplification in hPSCs with respect to neural differentiation. Cell Rep. 10, 616–632 (2015).
Markouli, C. et al. Gain of 20q11.21 in human pluripotent stem cells impairs TGF-β-dependent neuroectodermal commitment. Stem Cell Rep. 13, 163–176 (2019). Markouli et al. show that altered transforming growth factor-β signalling in PSCs harbouring a 20q11.21 copy number variant impairs their neuroectodermal differentiation.
Laurent, L. C. et al. Dynamic changes in the copy number of pluripotency and cell proliferation genes in human ESCs and iPSCs during reprogramming and time in culture. Cell Stem Cell 8, 106–118 (2011).
Ohgushi, M. et al. Molecular pathway and cell state responsible for dissociation-induced apoptosis in human pluripotent stem cells. Cell Stem Cell 7, 225–239 (2010).
Barbaric, I. et al. Time-lapse analysis of human embryonic stem cells reveals multiple bottlenecks restricting colony formation and their relief upon culture adaptation. Stem Cell Rep. 3, 142–155 (2014).
Liu, J. C. et al. High mitochondrial priming sensitizes hESCs to DNA-damage-induced apoptosis. Cell Stem Cell 13, 483–491 (2013).
Dumitru, R. et al. Human embryonic stem cells have constitutively active Bax at the Golgi and are primed to undergo rapid apoptosis. Mol. Cell 46, 573–583 (2012).
Price, C. J. et al. Genetically variant human pluripotent stem cells selectively eliminate wild-type counterparts through YAP-mediated cell competition. bioRxiv https://doi.org/10.1101/854430 (2019).
Bowling, S., Lawlor, K. & Rodríguez, T. A. Cell competition: the winners and losers of fitness selection. Development https://doi.org/10.1242/dev.167486 (2019).
Haupt, S., Mejía-Hernández, J. O., Vijayakumaran, R., Keam, S. P. & Haupt, Y. The long and the short of it: the MDM4 tail so far. J. Mol. Cell Biol. 11, 231–244 (2019).
Blum, B., Bar-Nur, O., Golan-Lev, T. & Benvenisty, N. The anti-apoptotic gene survivin contributes to teratoma formation by human embryonic stem cells. Nat. Biotechnol. 27, 281–287 (2009).
Mesri, M., Wall, N. R., Li, J., Kim, R. W. & Altieri, D. C. Cancer gene therapy using a survivin mutant adenovirus. J. Clin. Invest. 108, 981–990 (2001).
Ma, X. et al. High-level expression, purification and pro-apoptosis activity of HIV-TAT-survivin (T34A) mutant to cancer cells in vitro. J. Biotechnol. 123, 367–378 (2006).
Yang, D., Welm, A. & Bishop, J. M. Cell division and cell survival in the absence of survivin. Proc. Natl Acad. Sci. USA 101, 15100–15105 (2004).
Atkin, N. B. & Baker, M. C. Specific chromosome change, i(12p), in testicular tumours? Lancet 2, 1349 (1982).
Rodriguez, S. et al. Expression profile of genes from 12p in testicular germ cell tumors of adolescents and adults associated with i(12p) and amplification at 12p11.2-p12.1. Oncogene 22, 1880–1891 (2003).
Korkola, J. E. et al. Down-regulation of stem cell genes, including those in a 200-kb gene cluster at 12p13.31, is associated with in vivo differentiation of human male germ cell tumors. Cancer Res. 66, 820–827 (2006).
Chambers, I. et al. Nanog safeguards pluripotency and mediates germline development. Nature 450, 1230–1234 (2007).
Darr, H., Mayshar, Y. & Benvenisty, N. Overexpression of NANOG in human ES cells enables feeder-free growth while inducing primitive ectoderm features. Development 133, 1193–1201 (2006).
Singh, R., Letai, A. & Sarosiek, K. Regulation of apoptosis in health and disease: the balancing act of BCL-2 family proteins. Nat. Rev. Mol. Cell Biol. 20, 175–193 (2019).
Madden, D. T., Davila-Kruger, D., Melov, S. & Bredesen, D. E. Human embryonic stem cells express elevated levels of multiple pro-apoptotic BCL-2 family members. PLoS ONE 6, e28530 (2011).
Zhang, J. et al. Anti-apoptotic mutations desensitize human pluripotent stem cells to mitotic stress and enable aneuploid cell survival. Stem Cell Rep. 12, 557–571 (2019).
Ardehali, R. et al. Overexpression of BCL2 enhances survival of human embryonic stem cells during stress and obviates the requirement for serum factors. Proc. Natl Acad. Sci. USA 108, 3282–3287 (2011).
Watanabe, K. et al. A ROCK inhibitor permits survival of dissociated human embryonic stem cells. Nat. Biotechnol. 25, 681–686 (2007).
Milholland, B. et al. Differences between germline and somatic mutation rates in humans and mice. Nat. Commun. 8, 15183 (2017).
Kuijk, E. et al. Mutational impact of culturing human pluripotent and adult stem cells. Nat. Commun. 11, 2493 (2018).
Cervantes, R. B., Stringer, J. R., Shao, C., Tischfield, J. A. & Stambrook, P. J. Embryonic stem cells and somatic cells differ in mutation frequency and type. Proc. Natl Acad. Sci. USA 99, 3586–3590 (2002).
Petljak, M. et al. Characterizing mutational signatures in human cancer cell lines reveals episodic APOBEC mutagenesis. Cell 176, 1282–1294.e1220 (2019).
Kucab, J. E. et al. A compendium of mutational signatures of environmental agents. Cell 177, 821–836.e816 (2019).
Viel, A. et al. A Specific mutational signature associated with DNA 8-oxoguanine persistence in MUTYH-defective colorectal cancer. EBioMedicine 20, 39–49 (2017).
Bartkova, J. et al. DNA damage response as a candidate anti-cancer barrier in early human tumorigenesis. Nature 434, 864–870 (2005).
Gorgoulis, V. G. et al. Activation of the DNA damage checkpoint and genomic instability in human precancerous lesions. Nature 434, 907–913 (2005).
Cannan, W. J. & Pederson, D. S. Mechanisms and consequences of double-strand DNA break formation in chromatin. J. Cell Physiol. 231, 3–14 (2016).
Becker, K. A. et al. Self-renewal of human embryonic stem cells is supported by a shortened G1 cell cycle phase. J. Cell Physiol. 209, 883–893 (2006).
Becker, K. A. et al. Cyclin D2 and the CDK substrate p220NPAT are required for self-renewal of human embryonic stem cells. J. Cell Physiol. 222, 456–464 (2010).
Filipczyk, A. A., Laslett, A. L., Mummery, C. & Pera, M. F. Differentiation is coupled to changes in the cell cycle regulatory apparatus of human embryonic stem cells. Stem Cell Res. 1, 45–60 (2007).
Halliwell, J. A. et al. Nucleosides rescue replication-mediated genome instability of human pluripotent stem cells. Stem Cell Rep. 14, 1009–1017 (2020). Halliwell et al. show that human PSCs are subject to DNA replication stress that leads to DNA damage, mitotic errors and reduced population growth rates, but that these effects can be greatly reduced by culture in medium supplemented with nucleosides.
Simara, P. et al. DNA double-strand breaks in human induced pluripotent stem cell reprogramming and long-term in vitro culturing. Stem Cell Res. Ther. 8, 73 (2017).
Vallabhaneni, H. et al. High basal levels of γH2AX in human induced pluripotent stem cells are linked to replication-associated DNA damage and repair. Stem Cell 36, 1501–1513 (2018).
Akli, S. & Keyomarsi, K. Cyclin E and its low molecular weight forms in human cancer and as targets for cancer therapy. Cancer Biol. Ther. 2, S38–S47 (2003).
Bester, A. C. et al. Nucleotide deficiency promotes genomic instability in early stages of cancer development. Cell 145, 435–446 (2011).
Burrell, R. A. et al. Replication stress links structural and numerical cancer chromosomal instability. Nature 494, 492–496 (2013).
Frame, F. M., Rogoff, H. A., Pickering, M. T., Cress, W. D. & Kowalik, T. F. E2F1 induces MRN foci formation and a cell cycle checkpoint response in human fibroblasts. Oncogene 25, 3258–3266 (2006).
Pickering, M. T. & Kowalik, T. F. Rb inactivation leads to E2F1-mediated DNA double-strand break accumulation. Oncogene 25, 746–755 (2006).
Ahuja, A. K. et al. A short G1 phase imposes constitutive replication stress and fork remodelling in mouse embryonic stem cells. Nat. Commun. 7, 10660 (2016).
Halazonetis, T. D., Gorgoulis, V. G. & Bartek, J. An oncogene-induced DNA damage model for cancer development. Science 319, 1352–1355 (2008).
Jones, R. M. et al. Increased replication initiation and conflicts with transcription underlie cyclin E-induced replication stress. Oncogene 32, 3744–3753 (2013).
Cimini, D. et al. Merotelic kinetochore orientation is a major mechanism of aneuploidy in mitotic mammalian tissue cells. J. Cell Biol. 153, 517–527 (2001).
Chan, K. L., Palmai-Pallag, T., Ying, S. & Hickson, I. D. Replication stress induces sister-chromatid bridging at fragile site loci in mitosis. Nat. Cell Biol. 11, 753–760 (2009).
Naim, V., Wilhelm, T., Debatisse, M. & Rosselli, F. ERCC1 and MUS81-EME1 promote sister chromatid separation by processing late replication intermediates at common fragile sites during mitosis. Nat. Cell Biol. 15, 1008–1015 (2013).
Lukas, C. et al. 53BP1 nuclear bodies form around DNA lesions generated by mitotic transmission of chromosomes under replication stress. Nat. Cell Biol. 13, 243–253 (2011).
Lamm, N. et al. Genomic instability in human pluripotent stem cells arises from replicative stress and chromosome condensation defects. Cell Stem Cell 18, 253–261 (2016).
Maynard, S. et al. Human embryonic stem cells have enhanced repair of multiple forms of DNA damage. Stem Cell 26, 2266–2274 (2008).
Momcilović, O. et al. Ionizing radiation induces ataxia telangiectasia mutated-dependent checkpoint signaling and G2 but not G1 cell cycle arrest in pluripotent human embryonic stem cells. Stem Cell 27, 1822–1835 (2009).
Luo, L. Z. et al. DNA repair in human pluripotent stem cells is distinct from that in non-pluripotent human cells. PLoS ONE 7, e30541 (2012).
Hyka-Nouspikel, N. et al. Deficient DNA damage response and cell cycle checkpoints lead to accumulation of point mutations in human embryonic stem cells. Stem Cell 30, 1901–1910 (2012).
Adams, B. R., Golding, S. E., Rao, R. R. & Valerie, K. Dynamic dependence on ATR and ATM for double-strand break repair in human embryonic stem cells and neural descendants. PLoS ONE 5, e10001 (2010).
Adams, B. R., Hawkins, A. J., Povirk, L. F. & Valerie, K. ATM-independent, high-fidelity nonhomologous end joining predominates in human embryonic stem cells. Aging 2, 582–596 (2010).
Saretzki, G. et al. Downregulation of multiple stress defense mechanisms during differentiation of human embryonic stem cells. Stem Cell 26, 455–464 (2008).
Desmarais, J. A. et al. Human embryonic stem cells fail to activate CHK1 and commit to apoptosis in response to DNA replication stress. Stem Cell 30, 1385–1393 (2012). The study by Desmarais et al. demonstrates that human PSCs respond differently to DNA replication stress compared with somatic cells, as PSCs fail to activate the CHK1 checkpoint and, instead of arresting and repairing the DNA, PSCs undergo apoptosis.
Desmarais, J. A., Unger, C., Damjanov, I., Meuth, M. & Andrews, P. Apoptosis and failure of checkpoint kinase 1 activation in human induced pluripotent stem cells under replication stress. Stem Cell Res. Ther. 7, 17 (2016).
Hong, Y. & Stambrook, P. J. Restoration of an absent G1 arrest and protection from apoptosis in embryonic stem cells after ionizing radiation. Proc. Natl Acad. Sci. USA 101, 14443–14448 (2004).
van Echten-Arends, J. et al. Chromosomal mosaicism in human preimplantation embryos: a systematic review. Hum. Reprod. Update 17, 620–627 (2011).
Munné, S. et al. First PGT-A using human in vivo blastocysts recovered by uterine lavage: comparison with matched IVF embryo controls. Hum. Reprod. 35, 70–80 (2019).
Starostik, M. R., Sosina, O. A. & McCoy, R. C. Single-cell analysis of human embryos reveals diverse patterns of aneuploidy and mosaicism. bioRxiv https://doi.org/10.1101/2020.01.06.894287 (2020).
Brezina, P. et al. Genetic normalization of differentiating aneuploid human embryos. Nat. Preced. https://doi.org/10.1038/npre.2011.6045.1 (2011).
Kops, G. J., Foltz, D. R. & Cleveland, D. W. Lethality to human cancer cells through massive chromosome loss by inhibition of the mitotic checkpoint. Proc. Natl Acad. Sci. USA 101, 8699–8704 (2004).
Bolton, H. et al. Mouse model of chromosome mosaicism reveals lineage-specific depletion of aneuploid cells and normal developmental potential. Nat. Commun. 7, 11165 (2016).
Hardy, K. Cell death in the mammalian blastocyst. Mol. Hum. Reprod. 3, 919–925 (1997).
Haouzi, D. & Hamamah, S. Pertinence of apoptosis markers for the improvement of in vitro fertilization (IVF). Curr. Med.Chem. 16, 1905–1916 (2009).
Ihry, R. J. et al. p53 inhibits CRISPR-Cas9 engineering in human pluripotent stem cells. Nat. Med. 24, 939–946 (2018).
Schiroli, G. et al. Precise gene editing preserves hematopoietic stem cell function following transient p53-mediated DNA damage response. Cell Stem Cell 24, 551–565.e558 (2019).
Enache, O. M. et al. Cas9 activates the p53 pathway and selects for p53-inactivating mutations. Nat. Genet. https://doi.org/10.1038/s41588-020-0623-4 (2020).
Chagtai, T. et al. Gain of 1q as a prognostic biomarker in Wilms tumors (WTs) treated with preoperative chemotherapy in the International Society of Paediatric Oncology (SIOP) WT 2001 trial: a SIOP renal tumours biology consortium study. J. Clin. Oncol. 34, 3195–3203 (2016).
Walker, B. A. et al. A compendium of myeloma-associated chromosomal copy number abnormalities and their prognostic value. Blood 116, e56–e65 (2010).
Kilday, J. P. et al. Copy number gain of 1q25 predicts poor progression-free survival for pediatric intracranial ependymomas and enables patient risk stratification: a prospective European clinical trial cohort analysis on behalf of the Children’s Cancer Leukaemia Group (CCLG), Societe Francaise d’Oncologie Pediatrique (SFOP), and International Society for Pediatric Oncology (SIOP). Clin. Cancer Res. 18, 2001–2011 (2012).
Kraggerud, S. M. et al. Genome profiles of familial/bilateral and sporadic testicular germ cell tumors. Genes Chromosomes Cancer 34, 168–174 (2002).
Bown, N. et al. 17q gain in neuroblastoma predicts adverse clinical outcome. U.K. Cancer Cytogenetics Group and the U.K. Children’s Cancer Study Group. Med. Pediatr. Oncol. 36, 14–19 (2001).
Nguyen, H. T. & Duong, H. Q. The molecular characteristics of colorectal cancer: Implications for diagnosis and therapy. Oncol. Lett. 16, 9–18 (2018).
Robertson, G. P., Herbst, R. A., Nagane, M., Huang, H. J. & Cavenee, W. K. The chromosome 10 monosomy common in human melanomas results from loss of two separate tumor suppressor loci. Cancer Res. 59, 3596–3601 (1999).
Kimmelman, A. C., Ross, D. A. & Liang, B. C. Loss of heterozygosity of chromosome 10p in human gliomas. Genomics 34, 250–254 (1996).
Popat, S. & Houlston, R. S. A systematic review and meta-analysis of the relationship between chromosome 18q genotype, DCC status and colorectal cancer prognosis. Eur. J. Cancer 41, 2060–2070 (2005).
Damjanov, I. & Andrews, P. W. The terminology of teratocarcinomas and teratomas. Nat. Biotechnol. 25, 1212 (2007).
Steinemann, D., Göhring, G. & Schlegelberger, B. Genetic instability of modified stem cells - a first step towards malignant transformation? Am. J. Stem Cell 2, 39–51 (2013).
Valli, R. et al. Comparative genomic hybridization on microarray (a-CGH) in constitutional and acquired mosaicism may detect as low as 8% abnormal cells. Mol. Cytogenet. 4, 13 (2011).
Mayshar, Y. et al. Identification and classification of chromosomal aberrations in human induced pluripotent stem cells. Cell Stem Cell 7, 521–531 (2010).
Weissbein, U., Schachter, M., Egli, D. & Benvenisty, N. Analysis of chromosomal aberrations and recombination by allelic bias in RNA-Seq. Nat. Commun. 7, 12144 (2016).
Jackson, S. P. & Bartek, J. The DNA-damage response in human biology and disease. Nature 461, 1071–1078 (2009).
Bartek, J. & Lukas, J. Chk1 and Chk2 kinases in checkpoint control and cancer. Cancer Cell 3, 421–429 (2003).
Krokan, H. E. & Bjørås, M. Base excision repair. Cold Spring Harb. Perspect. Biol. 5, a012583 (2013).
Hoeijmakers, J. H. Genome maintenance mechanisms for preventing cancer. Nature 411, 366–374 (2001).
Lieber, M. R. The mechanism of double-strand DNA break repair by the nonhomologous DNA end-joining pathway. Annu. Rev. Biochem. 79, 181–211 (2010).
San Filippo, J., Sung, P. & Klein, H. Mechanism of eukaryotic homologous recombination. Annu. Rev. Biochem. 77, 229–257 (2008).
This work was funded in part by grants from the European Union’s Horizon 2020 research and innovation programme under grant agreement no. 668724 and from the UK Regenerative Medicine Platform, MRC reference MR/R015724/1.
The authors declare no competing interests.
Peer review information
Nature Reviews Molecular Cell Biology thanks Florian Merkle and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
COSMIC Cancer Gene Census database: https://cancer.sanger.ac.uk/census
- Age-related macular degeneration
A common cause of blindness in elderly people due to degeneration of the retinal pigment epithelium underlying the retina.
- Mosaic culture
A culture containing two or more genetically distinct cell types; for example, an original cell type and a genetic variant derived from it.
- Population bottleneck
A situation occurring during successive passaging of cell cultures in which a culture is derived from a very small number of cells from the preceding culture.
- Driver genes
Genes whose altered expression provides the main selective growth advantage associated with a particular genomic variant.
- Clonogenic assays
Assays in which clones of cells are grown out from isolated single cells to assess the properties of the different cells composing a mosaic culture.
An unbalanced genome caused by the presence of an abnormal number of chromosomes or fragments of chromosomes in a cell; it does not include abnormal numbers of chromosomes that are exact multiples of the haploid set of chromosomes (that is, 23 in human cells).
- Interstitial duplications
A type of chromosomal aberration in which a duplicated DNA segment is inserted in the same chromosome.
- Whole-exome sequencing
A method of sequencing all of the protein-coding regions (exome) in the genome.
Embryonic tissue that gives rise to all of the fetal tissues, including the germ line.
- Xenograft tumours
Tumours developing from cells transplanted to a host of a different species; in this Review, typically tumours produced by human cells in an immunodeficient mouse host.
- Primitive endoderm
Cells found in teratomas and closely resembling cells of the extraembryonic endoderm found in the pre-implantation embryo.
A discrete region of the genome that has been duplicated one or more times.
- Hitchhiker genes
Genes present on amplified or deleted chromosome segments with no effect on the growth advantage of the variant cell.
- Cell competition
Cell–cell interaction mechanism leading to elimination of cells that are viable in their homotypic environment in the presence of comparatively fitter cells.
A chromosomal rearrangement in which one whole arm of a chromosome is replaced by a complete copy of the other arm, resulting in a loss of the genes located on the first arm and duplication of the genes located on the other arm.
- Clinical grade
A loose and ill-defined term that identifies cell lines that have been developed and maintained in ways that will satisfy regulatory authorities for clinical application; it is commonly applied to pluripotent stem cell lines that have been derived according to good manufacturing practice.
Genomic changes involving the insertion or deletion of a sequence of one or more nucleotides.
- RB1–E2F checkpoint
Controls the entry into S phase and the initiation of DNA replication during the cell cycle; dependent on the retinoblastoma tumour suppressor protein, RB1, regulating expression of the transcription factor E2F.
- G1/S checkpoint
Also known as the restriction point, safeguards entry into S phase during the cell cycle, where DNA synthesis occurs.
About this article
Cite this article
Halliwell, J., Barbaric, I. & Andrews, P.W. Acquired genetic changes in human pluripotent stem cells: origins and consequences. Nat Rev Mol Cell Biol 21, 715–728 (2020). https://doi.org/10.1038/s41580-020-00292-z
Regenerative Medicine (2021)
International Journal of Molecular Sciences (2020)