Thumbnail image of

In this edition of BioEssays two papers address the mechanisms of cancer. Cancer is widely (but not universally) regarded as a genetic disease – i.e. a disease of genes. This is true in both the familial and somatic senses of the term. It has long been recognized that the risk of developing cancer is not evenly distributed in the community, and that some families are particularly prone to the disease. For the past 30 years, it has become increasingly clear that cancer cells also acquire changes in the DNA code, some of which appear to be required for tumor formation. The evidence for this comes from various sources, including the use of increasingly sophisticated and powerful genomic tools. Perhaps the most important evidence comes from clinical trials, where it is becoming increasingly evident that the presence of (usually acquired) mutations in certain genes is predictive of responses to targeted therapeutics – the phenomenon known as “oncogene addiction.”

So much for the hardwiring of cancer. Over much the same period, it has also become increasingly apparent that epigenetic factors play a crucial role in carcinogenesis. The dependency of cancer cells upon epigenetic events was elegantly shown by Beatrice Mintz in the 1970s. Illmensee and Mintz showed that teratocarcinoma cells can contribute to mouse development following inoculation of early blastocysts, implying that their malignant potential was being “reset” by unknown processes during embryonic development 1. Almost 30 years later, Hochedlinger et al. demonstrated a similar phenomenon using melanoma cells, this time applying nuclear transplantation 2. In this astonishing study, cells containing the melanoma genome contributed widely to most tissues in the derived animals, suggesting that what distinguishes normal cells from cancer cells is to a large extent reprogrammable. Whatever else the gross genetic derangements in most cancer cells may mean, they seem unlikely to be grossly reversible.

All hardware needs software – which is where epigenetics comes in. Epigenetics may refer to anything that heritably codes for phenotype that is not contained in the DNA sequence itself. Examples of this are the chemical modification of DNA by methylation, or modification of histones by methylation or acetylation. It is known that cancer cells have markedly abnormal patterns of DNA methylation and histone modifications. Typically, it appears that the genome is for the most part hypomethylated, but focally hypermethylated. The precisely targeted methylation of CpG dinucleotides in the promoters of key tumor suppressor genes contributes to tumorigenesis. What is not yet clear is how these methylation templates are applied. The hypothesis article in this issue by Zuzana Jasencakova and Anja Groth (pp. 847–855) suggests that disrupted chromatin organization during replication may facilitate stochastic silencing of genes at points of fork stalling. The aberrant firing of origins of replication, and perhaps defects in the DNA repair machinery, may contribute to what the authors refer to as “replicative stress.”

The definition of epigenetic is growing like topsy, because it is emerging that the gap between the genotype and phenotype is a highly regulated space, including transcriptional and translational regulation via miRNA, ncRNAs, modulation of message stability, RNA editing, and so on. Another hypothesis article in this issue, by Adam Wilkins (pp. 856–865), addresses a potential role for retrotransposons in cancer initiation. As briefly summarized by Wilkins, the role of transposable elements in mammalian biology has been looming for what seems like decades – waiting for application to human disease. Transposable elements constitute an increasingly fascinating component of intergenic DNA. The latter accounts for ∼99% of the human genome, and is increasingly recognized as far from passive. Retrotransposons, which utilize an RNA intermediate, are activated by cellular stresses, including carcinogen exposure. Activated retrotransposons are hypothesized to affect transcription of other genes, and to promote genomic instability, both of which have oncogenic consequences. The epigenetic side of this process relates to the mechanism of repression of retrotransposon activity in normal circumstances, which is mediated in part by DNA methylation. As noted above, there is broad loss of methylation across the genome in cancer cells, particularly in intergenic regions. The loss of methylation is therefore postulated to result in the de-repression of transposable elements, with resultant oncogenic effects.

Key questions remain, as we should expect of any intriguing hypothesis. Evidence for the link between activation of retrotransposable elements and modification of specific genes that lead to cancer is circumstantial rather than unequivocal. It is not clear why activation of retrotransposable elements should result in cancer, rather than have neutral or fitness-reducing effects on individual cells. Of course, this is a general problem for (epi)genetic models of cancer, where it is assumed that natural selection favors particular configurations of epigenetic and genetic changes, which themselves arise in a stochastic manner. The hypotheses presented in this issue present interesting – and hopefully testable – concepts that link genetic and epigenetic processes in tumorigenesis. We commend them to you.


  1. Top of page
  2. References