How many mutations does it take, and where?

It has been widely accepted for the past three decades that you require two-hits (Knudson’s Two-Hit hypothesis), in TSG to drive cancer. This was supported by the fact that most TSG are usually recessive, and encode loss-of-function mutations, and oncogenes are usually dominant, and encode gain-of-function mutations (Cooper and Hausman, 2007). Activation of one allele (cancer mutation) in a proto-oncogenes is sufficient to drive a dose-response, and both alleles need to be removed in the TSG to see an effect. However, as TSG mutations are recessive and can be passed in the germ-line, a single hit later in life is the second. However, many examples of haplo-insufficiency variation according to genetic background/environmental trigger exist (Rose and Bhattacharya, 2016). Furthermore, many TSGs have related proteins that carry out similar functions in their absence, meaning that loss of one allele may depend on the state of accessory factors.

High-resolution cancer genome sequencing will further our understanding of the genetic bottlenecks and regulatory processes in cancer. However, it is not as if we know less about

cancer genetics now than we did previously. The work of the 1970s-90s revealed a world of conserved mutations (sometimes called driver mutations) and structural genomic events to be associated with diverse forms of cancer. This conservation of cancer elements between people suggests that the genes responsible for cancer are under selective pressure and are highly important to other biological processes; otherwise, these genes would have found regions DNA less susceptible to large-scale DNA changes such as chromosomal losses and translocations. Small base-pair insertions, substitutions and deletions, amplifications of short DNA stretches, and microsatellite variations have all been found to aid tumour progression, including in colorectal cancer. Any type of mutation has the possibility to affect gene expression, function and/or regulation; demonstrating the many genetic pathways to cancer. Given the somatic heterogeneity observed between different somatic cells across the human body, mutations will also have cell type-specific effects (MacPherson et al., 2004). Considering the burden of reactive oxygen species in our cells, and the amount of carcinogens afflicting us daily, our body does a good job at preventing neoplasia in general. There is a lot of cellular turnover.

The number of the essential cancer cell mutations and the rate, varies between tumours and individuals. However, commonality between patients does exist, as shown by the work of Laura Wood and colleagues, who analysed exonic regions in colorectal tumours, and found that one group of genes were commonly mutated, and another group were also mutated, but at lower frequency (Wood et al., 2007). The authors called this pattern, mountains and hills, referring to driver and passenger mutations in the cancer; of which 15 and 60 were described, respectively. When such observations are coupled to the common structural changes described (i.e. BCR-abl), it tells us that some mutations are more oncogenic than others; and their frequency is also constrained by genetics and environment.

On top of such base-altering mutations are epigenetic marks, such as cytosine methylation, which adds a functional group to the DNA sequence that can regulate gene expression and DNA replication. Indeed, various methylation marks have been associated with cancer development. For example, hypermethylation of different tumour suppressor genes (or intergenic regions) leads to their inactivation, and cancer development (Kulis et al., 2013). Thus, the identification of endogenous genes and proteins and exogenous factors mediating cancer-associated methylation changes require further exploration; as do other epigenetic marks present in cancers, like the increased deacetylation of histones H2 and H3 that predispose to tumorigenesis. Another, partly epigenetic mechanism involved in tumorigenesis involved microRNA silencing. In humans and other vertebrates, microRNAs are potent regulators of gene expression, and many important regulatory microRNAs are found to be heavily methylated and silenced in cancers – allowing erroneous gene expression to proceed (Garzon et al., 2009).

Importantly, in cells, the effects of many diverse forms of genomic lesions are often self- amplifying and additive. For example, mutations in DNA proof-reading enzymes often lead to mutation accumulation with normal division (with more mutations accumulating with every cellular generation), and accumulating additional mutations in putative oncogenic regions, can

both further a tumour’s capacity to rapidly divide. This can be thought of as a chain reaction, in which one error leads to ever more errors, and disease. When cancer cell clonal expansion is considered, the proportion of cells dying due to the accumulation of deleterious variants is unknown. It is likely that along the oncogenic path, the mutation burden can also terminate cancer.

The mutations themselves, however, are not the be all and end all of tumour development. Other factors are often able to propagate or inhibit cancer progression in tandem with deleterious variants. In many scenarios, mutated cells require the action of growth promoters to cue proliferation. Oestrogens are well known mediators of breast and endometrial cancer in women (Travis and Key, 2003). Progesterone can be used to antagonise the effects of oestrogen clinically. It all depends on how the cancer cell can exploit the environment in which it finds itself. Ovarian cancers might arise more frequently when mutations in endometrial cells lead to constitutive steroid hormone receptor expression, driving cellular growth and proliferation. The gene-environment interaction again rears its head.