Cancer Genetics and Somatic Mutation


By the early 2000s, many oncogenes had been identified in humans and databases were available cataloguing the DNA sequences for particular oncogenes obtained from biopsied tissues of different tumor types. Most analyses of these databases simply examined frequencies of somatic mutations at particular sites in different tumors.

Codon-based models have been applied to analyze DNA substitutions observed between individuals, or species, allowing inferences about positive, or negative, selection using sequence samples from populations, or via phylogenetic comparisons of sequences from different species. This is done by examining the dN/dS ratio, the rates of nonsynonymous versus synonymous substitution, a ratio greater than one indicates the nonsynonymous substitutions are favored (positive selection) and a value less than 1 indicates they are disfavored (negative selection). We decided to try applying a similar approach using a codon-based model to analyze somatic mutations in oncogenes from tumor samples [1]. This work was done in collaboration with Simon Ro, a PhD student in the group, and Ziheng Yang. New molecular techniques and sequencing methologies also spurred our interest in the process of somatic mutation in healthy cells. Simon Ro and I subsequently collaborated on several studies aimed at estimating patterns and rates of somatic mutation in vivo [2,3].

The papers

[1] Z. Yang, S. Ro, B. Rannala. 2003. Likelihood Models of somatic mutation and codon substitution in cancer genes. Genetics 165: 695–705. Download

The method developed in this paper incorporates a new codon-based model to estimate the relative rate of substitution (fixation of a somatic mutation in a cancer cell lineage) of nonsense vs. missense mutations in different functional domains and in different tumor tissues. In population genetics and phylogenetics the DNA substitutions we observe in coding regions are typically either synonymous or nonsynonymous (missense) mutations. Other types of substitutions, such as deletions causing frameshifts, or nonsense mutations are so deleterious that they are not observed. With oncogenes severe mutations can be beneficial for a tumor if, for example, they make a tumor-suppressor gene such as TP53 non-functional. Thus, it is necessary to consider a broader range of variants in modeling oncogene somatic variation and the evolutionary codon substitution models cannot be applied to analyze tumor gene mutation data without modifications. A novel finding of our analysis of the TP53 tumor mutation database was that the ratio of the rates of nonsense to missense substitutions is significantly lower in the DNA-binding and transactivation domains (ratios near 1) than in structural domains such as the linker, tetramerization (oligomerization), and proline-rich domains (ratios exceeding 100 in some tissues), implying that the specific amino acid sequence may be less critical in structural domains (e.g., amino acid changes less often lead to cancer).

[2] S. Ro, B. Rannala. 2004. A stop‐EGFP transgenic mouse to detect clonal cell lineages generated by mutation. EMBO Reports 5: 914-920. Download

This paper describes a novel transgenic mouse that we developed to allow clonal cell lineages to be traced in virtually any tissue. It was intended as a system for stem-cell lineage studies. A green fluorescent cell lineage is generated by a random mutation at an enhanced green fluorescent protein gene that carries a premature stop codon. Each mutation is unique ensuring clonality. The transgenic system allows detection of mutations and stem-cell fate mapping in the epidermis using live mice, as well as in the kidney and liver post-mortem. Cell lineages that descended from single epidermal stem cells were found to be capable of generating three adjacent corneocytes using the system, providing evidence for horizontal migration of epidermal cells between epidermal proliferative units (EPUs), in contrast to the classical EPU model.

[3] S. Ro, B. Rannala. 2005. Evidence from the stop‐EGFP mouse supports a niche‐sharing model of epidermal proliferative units. Experimental Dermatology 14: 838-843. Download

The classical model of epidermal proliferative units (EPUs) postulates that each EPU is composed of a single column of corneocytes plus epidermal cells directly below the column and is maintained by a single stem cell within the unit. Using the stop-enhanced green fluorescent protein (stop-EGFP) transgenic mouse system, we previously showed epidermal stem cell clonal lineages could produce multiple adjacent corneocytes (i.e. epidermal cells belonging to multiple adjacent EPUs), contradicting the classical EPU model. In this study, we replicate our earlier findings using untreated stop-EGFP mice and relying on spontaneous mutations to generate clonal cell lineages. We propose an alternative to the classical EPU model to explain the dynamic nature of epidermal proliferation. Our niche-sharing model of EPUs allows epidermal cells to horizontally migrate among EPUs, so that multiple stem cells cooperatively maintain a larger proliferative compartment.