In silico p53 mutation hotspots in lung cancer.

For cancer one of the primary aims of molecular epidemiology is to identify the endogenous or exogenous cause of mutations within a gene. Regarding exogenous mutagens, many mutation data have become available via in vitro and in vivo mutation assays and become publicly available through mutation databases such as the Mammalian Gene Mutation Database (http://lisntweb.swan.ac.uk/cmgt/index.htm). One particular mutation assay incorporates the bacterial supF tRNA gene which allows selection of mutations at virtually all nucleotides. We have developed an algorithm called LwPy53 that utilizes mutation data from supF that can be used to predict chemically induced hot-spots along the p53 gene. The prediction is based on a number of parameters: the mutability of supF dinucleotides after treatment with a mutagen of interest; DNA curvature along the p53 gene; the selectability of a mutation along the gene; the likelihood of a site being within a nucleosome. We applied LwPy53 to exons 5, 7 and 8 of p53 using benzo[a]pyrene diol epoxide (BPDE)-induced mutation data for supF to obtain a predicted BPDE G-->T transversion spectrum after hypothetical treatment with BPDE. The resulting predicted mutation distribution reveals strong mutation hot-spots at codons 157, 248 and 273 that correlate with known BPDE adduct hot-spots within p53. The predicted BPDE spectrum strongly resembles the G-->T mutation spectrum compiled from known lung cancer mutation data from smokers and further supports evidence that BPDE contributes to the overall smoking-related mutation distribution in lung cancer. The algorithm shows how BPDE target sequence specificity and DNA curvature both shape the overall mutation distribution.