Pluripotent stem cells, which give rise to every cell type, can acquire cancer-causing genetic mutations when grown in vitro. This finding has implications for the use of pluripotent cells in basic research and in the clinic. See Letter p.229
The production and culture of pluripotent stem cells, which can give rise to every cell type in the body, have great potential for drug screening, basic research and regenerative therapies. Genetic mutations are known1 to arise when pluripotent cells are propagated in vitro, but the nature and extent of these mutational events have been unclear. On page 229, Merkle et al.2 provide evidence that human pluripotent stem cells are prone to develop mutations in a gene that drives cancer development, TP53.
To drill down into the genetic alterations that occur in cultured pluripotent cells, the authors took a lead from cancer genetics, and searched for DNA sequences present only in subsets of cells (subclones) in a population. If such mosaic mutations confer a growth advantage, the subclonal population can expand over time, at the expense of chromosomally normal cells that proliferate more slowly. Such subclones could alter the properties of the cell population, and perhaps replace normal cells with those at higher risk of acquiring properties associated with cancer development.
Merkle and colleagues obtained 140 lines of pluripotent cells extracted from embryos, called human embryonic stem cells (hESCs). They cultured the cells under standard growth conditions and, taking serial samples, searched for base-pair changes across the protein-coding regions of the genome using a next-generation sequencing technique that allows simultaneous sequencing of millions of stretches of DNA. Within weeks, a fraction of cells in five of the lines harboured mosaic, single-nucleotide mutations in TP53 (Fig. 1). The observed mutations cause the amino-acid substitutions most frequently reported in adult cancer genomes3. These same mutations are observed in people who have Li–Fraumeni syndrome4, which confers a predisposition to multiple cancers, often at an early age.
Next, the authors mined published sequence data for RNA and protein-coding DNA from hESCs and human induced pluripotent stem cells (hiPSCs), which arise from differentiated adult cells that have been induced to adopt a pluripotent state. The fraction of these lines that harboured TP53 mutations was comparable to those in the authors' own analysis, and most were again mosaic. In a subset, there was evidence that the subpopulation of TP53-mutant cells had replaced the normal cell population. Furthermore, TP53 mutations arose independently in different laboratories that used cells from the same, unmutated source, underscoring the proclivity for TP53 mutations to arise that have the potential to cause cancer5.
Merkle and colleagues' results add to concerns raised by others that induced pluripotency can generate a spectrum of genetic abnormalities — not only single-nucleotide changes, but also deletion or replication of large chromosomal regions1. The authors found no recurrent mutations apart from the dangerous cluster in TP53, which in itself is highly informative. However, the moderate size of the study means that we cannot rule out the possibility of additional, less frequent acquired events in other cancer genes. Indeed, the study revealed an apparently random distribution of mutations across all other genes in the hESC lines — a pattern similar to cancer genomes6. These random mutations arise on the road to cancer, as a result of disruptions of key processes that maintain genomic integrity.
The changes that arise in mutational signatures in cancer cells over time correlate with distinct biological mechanisms such as ageing, or environmental exposures such as smoking7. To develop a more comprehensive picture of changes in pluripotent stem cells over time, serial samples of cultured hESCs and hiPSCs could be taken with greater frequency, and sequencing performed on whole genomes, rather than on protein-coding regions alone. This could provide information about the spectrum of acquired mutations, and perhaps about mutational signatures related to culture conditions. We might learn how artificial culture of hESCs puts stress on the genome, perhaps increasing the probability of a deleterious TP53 mutation. Comparison with cancer genomes could provide insights into how and why specific mutations arise, particularly during the induction of hiPSCs or during maturation of hESC progenitors in vivo.
The acquisition of mosaic TP53 mutations in pluripotent cells over time is reminiscent of the higher frequencies of acquired mosaic mutations that human genomes exhibit with age8. Perhaps pluripotent cells undergo ageing-like changes in culture — which would not be surprising, given the inherent stress of in vitro conditions. In molecular epidemiological studies, both single-base mosaic mutations and large replications or deletions have been associated with increased risk of disorders of ageing such as cancer, diabetes and neurodegenerative diseases8. Serial analyses could be useful as a surrogate for investigating the underlying mechanisms by which mosaic events occur in the ageing genome.
That Merkle and colleagues have uncovered a major issue in culturing hESCs — the emergence of dangerous TP53 mosaic mutations — should not deter the pursuit of regenerative medicine. But the results give pause for thought. In future, it will probably be necessary for researchers to carefully monitor the genetic history of induced pluripotent cell lines. The field will benefit from studies to identify additional recurrent events in TP53 or other genes, particularly as pluripotent cells move into clinical studies.
The authors of the current paper are to be commended for surveying as many cell lines as possible in a rigorous and reproducible way. But, unfortunately, a substantial fraction of published hESC lines were not available for analysis. Every effort, whether individual or collective, should be made to characterize hESCs. Even lines that have major abnormalities in chromosome number, or particularly harmful TP53 mutations, are of value for investigating key biological questions. But until we know more, these mutation-harbouring hESCs are not suitable for clinical use. A large international database reporting the genomic landscape of every hESC line, with an emphasis on serial analyses, would certainly benefit researchers.
A pressing question is whether screening of cells before and after they have undergone stressful procedures such as gene editing (which kill some cells, leading to a bottleneck that might cause the expansion of certain subclonal populations) should be limited to TP53. Probably not — at least, not until the landscape of acquired mutations is better defined. In the meantime, genome-wide surveys of serial samples of propagated pluripotent lines should be routinely carried out. Finally, it will be crucial to check pluripotent lines before clinical testing and, in the future, before their use in regenerative medicine9.