Recently we have published the PrePrint of a new hypothesis on the roles of zinc and chromosome chaos (aneuploidy) in cancer. During discussions about this aneuploidy metal transporter cancer (AMTC) hypothesis, a few questions have come up, which we try to clarify below:
Question: Is the AMTC hypothesis compatible with “mainstream” cancer genomics?
Answer: Yes, this hypothesis can be seen as an extension rather than a contradiction to current models of cancer genomics. It does not contradict the important roles of established oncogenes and tumor suppressor genes and their associated driver mutations, although it raises the possibility that some of these mutations may interact with or compensate early changes in intracellular zinc and copper concentrations. As an example, the three most prevalent driver mutations in TP53 occur at positions R175, R248 and R273 (cBioPortal.org; all cancer types), all of which are zinc-ion binding sites. Overall, the AMTC hypothesis assigns a key role in carcinogenesis to aneuploidy, a phenomenon which is not explained satisfactorily by current models of cancer genomics.
Question: Often it is said that cancer is caused by mutations in TP53 or other cancer genes.
Answer: Well, there is no doubt that these established driver mutations play a key role – but they are not sufficient to cause cancer (e.g. Li, Solnik et al. 2000 in PNAS on Hahn, Counter et al. 1999 in Nature). An additional type of genetic mutation, aneuploidy, is present in tumor genomes and arguably may be required for carcinogenesis. In our model, the mutations in known cancer genes act in combination with aneuploidy, based on the observations that (i) essentially all malignant tumors are aneuploid, including tumors with somatic or inherited TP53 mutations and (ii) aneuploidy occurs early, typically before point mutations.
Question: What about hereditary types of cancer, such as congenital BRCA1 or BRCA2 mutations, leading to family histories of breast and ovarian cancers?
Answer: Again, this phenomenon is compatible with our hypothesis. Obviously, BRCA1 or BRCA2 mutations do not directly lead to malignant transformation, since most of the billions of cells in a carrier are healthy. But in case a hereditary tumor develops, it is always aneuploid, similar to non-inherited types of solid cancers. Thus, these mutations are certainly a predisposition to cancer and hereby accelerate its formation, but in addition, they seem to require one or several chromosome-level mutations, aneuploidy, in order to become malignant.
Question: If this idea was correct, why did the zinc transporter genes not show up as candidate driver genes in previous cancer studies?
Answer: This is indeed a key point. The short answer is that analysis of aneuploidy has rarely been used to suggest specific candidate genes and that our study is one of the first attempts. In previous studies based on similar datasets, the focus was on point mutations or alternatively, when copy number variation was taken into account (e.g. Zack et al., 2013), emphasis was on focal events including high level amplifications (many copies of a small genomic region) and deletions (zero copies of a small genomic region). These focal events are sometimes (possibly mistakenly) used to narrow down the much longer, often chromosome-arm long stretches of DNA that are found to be gained (resulting in three copies) or hemizygously lost (one copy). However, in our article we consider focal mutations and chromosome-arm mutations as separate events, which is consistent with their distinct background mutations rates as was observed (Mermel et al., 2011) and the low degree of sequence overlap between the two categories (e.g. Beroukhim et al. 2010). In a recent study focusing on aneuploidy (Davoli et al., 2013), candidate driver genes were pre-selected based on different types of point mutations, and hence, driver genes that exclusively act through aneuploidy and gene dosage, may have been overlooked. In short, we think that many established cancer driver genes (e.g. RB1, ERBB2, MYC etc.) function mainly through point mutations or focal copy number changes, others might function through a combination of point mutations in one copy and aneuploidy affecting the second copy of the gene (e.g. chromosomal loss including TP53 on chromosome arm 17p) – while certain metal transporter genes (e.g. SLC39A14, SLC39A1, SLC39A4, ATP7B etc.) act through aneuploidy in cancer.
Question: In cancer cells, many different metabolic changes occur, e.g. in the balance of glucose, redox or oxygen levels. Such changes may be seen as a consequence of carcinogenesis, so why should this be different with the balance of trace elements such as zinc and copper?
Answer: Several reasons make us think that the disruption of zinc and copper homeostasis indeed may represent a causal factor in carcinogenesis: (i) zinc is closely linked to several of the cancer hallmarks as discussed in the article (ii) there is experimental evidence that up- or down-regulation of several different iron, zinc and copper transporters drives a cell clone towards malignancy and (iii) locally induced zinc deficiency as a possible cancer-promoting selective force has been demonstrated experimentally. Taken together and in addition to the genomic analysis, these experimental observations provide good reasons why trace elements such as zinc and copper are causal factors in cancer development, likely in contrast to other metabolic changes.
Question: It seems that we have a chicken and egg situation- which comes first, the chicken or the egg? The hypothesis argues that dyshomeostasis of metals characteristic of cancer is the result of gene dosage effects of transporters due to gains and losses of chromosomes and chromosomal fragments. However, it is also suggested that the loss of metal homeostasis is a cause of the gains and losses of chromosomes (aneuploidy). Isnt this incongruent?
Answer: It is true that the metal dyshomeostasis comes in twice, but it still is not a chicken and egg situation: In a first step, external localized zinc deficiency in a tissue may come in as a selective force that kicks off the expansion of certain cells with gains or losses: such zinc deficiency was observed as cancer-promoting in mouse experiments, and speculatively, it is associated with many situations of increased cancer risk, e.g.: bacterial infection, obesity, old age, etc. This leaves a number of cells in the pre-cancerous tissue, that are genetically adapted to these low zinc conditions. In a second step, due to their chromosomal gains and losses, these cells show a deregulated intracellular trace element homeostasis – and so-to-say as an accidental side-effect, this may have an effect on the hallmark characteristics and contribute to the clonal expansion. Of course, there is not much experimental data on early chromosomal events, but the little data that exists, seems fully compatible.
Question: Aneuploidy is a late event, therefore it must be a consequence not a cause of cancer.
Answer: No, aneuploidy and expansion of aneuploid cell clones are early events in carcinogenesis. Aneuploidy is widespread in benign lesions, and develops progressively in full tumors and metastasis. Of note, aneuploidy and CIN (chromosomal instability) are not the same, e.g. not all aneuploid cells show CIN, but cells with CIN seem all to be aneuploid. For example, in pre-cancerous “benign lesions”, the first events turn out to be chromosomal mutations, hence aneuploidy. This has been known from CGH studies on benign lesions from the 1990s and more recently from resequencing studies, e.g. a studiy on breast cancer that analyzed the order of mutations in different patients, including both aneuploidy and point mutations (Newburger et al., 2013). Interestingly, based on single cell sequencing, there is a background level of randomly aneuploid cells even in healthy tissue (Knouse et al., 2014). This might provide the genetic diversity for an early selective force to act upon.
Question: The statistical analysis does not prove the AMTC hypothesis.
Answer: This is correct and we only claim that the genomic results are compatible with our hypothesis (by the way, as far as I understand, no statistic in the world has ever “proven” a biological model anyway). Conclusive evidence has to come from additional functional experiments and analyses, and we sincerely hope that scientists from different disciplines will help out. Furthermore, due to the similar behaviour of hundreds of neighboring genes in aneuploidy, it may be extremely difficult to identify individual driver genes purely based on statistical analysis of genomic data. Several ways to improve the analysis are sketched out in the Supplemental Notes and in our suggestions for additional experiments (text box B), so your contributions and improvents are most welcome! Nevertheless, we think that we have done a good job in identifying at least some of the important players in the aneuploidy of cancer, partially because we had a prior working hypothesis on metal transporters. Importantly, I would like to emphasize that in my opinion, the strength of the article is that it combines many different lines of observations (zinc in the hallmarks of cancer, gene dosage in fifteen types of cancer, experimental evolution of yeast clones, in vitro experiments on metal transporters, epidemiological observations etc.) – together they form a pretty convincing picture which certainly cannot be reduced to a simple p-value.
Question: Could it really be so simple and are these ideas actually new?
Answer: It is too early to say of course, but I wouldnt be surprised if zinc and aneuploidy turn out as a simple, unifying theme in cancer – it may have been a “blind spot” since not too many cancer scientists work on aneuploidy and even less scientists are familiar with the emerging field of zinc biology. Curiously, most of the observations are actually quite old and well-established, e.g. aneuploidy (at least since 1914), changes in metal concentrations in tumors (at least since the 1940s), the genomic position of transporter genes (at least since the human genome project) or the role of zinc and copper with regard to the hallmarks of cancer (last decades). Aneuploidy as an early adaptation to tissue zinc deficiency would be new, and second, the link between all these obervations would be new.
Question: Do you suggest that metal transporter genes are mutation hotspots in cancer genomes?
Answer: Not exactly, although we do suggest that as a result of mutation and clonal selection, metal transporter genes together with their encoding chromsome arms end up as mutated more often than other genes. In an evolutionary sense, mutations mainly happen as a stochastic, undirected process, meaning that all chromosome arms probably have similar chances of getting lost or being gained. However, most of these genomes in single cells are never seen – instead we only observe the one cell that becomes multiplied in millions or billions of tumor cells due to clonal selection. And its cancer cell genome indeed shows mutations of very specific chromosome arms, which according to the AMTC hypothesis are associated with gene dosage effects through gains or losses of certain metal transporters. The story is analogous to cancer genes such as TP53: Tumor cells often carry point mutations that code for certain protein domains of p53 (the protein that is encoded by TP53). But why do tumor cells often show such mutations in TP53, provided that point mutations are expected to occur randomly across the genome? Well, it seems safe to assume that during the evolution of each tumor with such a mutation, there have existed millions of mutated cells without major consequences (e.g. one mutation in each domain of every other protein-coding gene ) before one random mutation finally would affect a certain domain of TP53 and enable the tumor cell to out-compete the other cells.
Question: How do I cite this idea?
Answer: Just download a citation at bioRxiv.org for your favourite reference manager or copy/paste:
The disruption of trace element homeostasis due to aneuploidy as a unifying theme in the etiology of cancer. Johannes Engelken, Matthias Altmeyer, Renty B Franklin.
bioRxiv doi: http://dx.doi.org/10.1101/002105
Question: Do you plan to publish this idea in a peer-reviewed journal?
Answer: We haven´t decided yet. Since this idea needs to be tested by experiments and cannot be verified by peer-review alone, publication without further delay at bioRxiv seemed the best option. At some point in the future, we may still consider submitting an improved and reformatted version to an open-access journal.
Question: Do you plan to test this idea experimentally?
Answer: I have applied for funding to test the first step of the AMTC hypothesis. Essentially I would run a experiments of experimental evolution on cell lines and expect to recapitulate the early observed patterns of aneuploidy, e.g. gains of chromosome arms 1q and 8q. This might involve zinc deficiency in the Petri dish or some other selective force. Analogous to yeast experiments, generation time would have to be at least 100 generations, corresponding to several months of cell culture. Since I feel that this idea could be potentially relevant for many people, I do not want to hold back ideas. If it turns out real, this would be a gift for everybody. Therefore, we have described a number of possible experiments in the PrePrint, so that anyone interested can join in.