In 2013, a ground-breaking paper by Ludmil Alexandrov and colleagues at the Wellcome Trust Sanger Institute, used genomic sequencing data and mathematical modelling to demonstrate the presence of specific mutational signatures in the pattern of mutations found in cancers[i]. Some of these signatures are associated with endogenous processes, some bear similarities to environmental exposures such at tobacco smoking, whereas others have no apparent etiology. These findings proposed that mutational signatures could be used to understand mutagenic mechanisms involved in carcinogenesis.

Our laboratory is investigating the induction of mutations in vivo using the MutaMouse transgenic rodent model that contains the lacZ gene integrated in the DNA of each cell[ii]. Following chemical exposure, the lacZ gene can be recovered from any tissue, packaged into phages, and infected into bacteria that, under selective conditions, will allow the formation of plaques only if the lacZ gene has been inactivated by a mutation. By sequencing these mutations, we can rapidly and inexpensively obtain insight into the mutagenic mechanisms of chemicals.

A few years ago, at a Gordon Conference on Mutagenesis, I listened to Prof. David Phillips present his laboratory’s experiments deriving mutational signatures in human and mouse cells exposed to benzo(a)pyrene (BaP), a component of tobacco smoke. The results were published together with other chemicals demonstrating the presence of mutational signatures in the mutation profiles of many environmental agents[iii]. Since our laboratory had recently published the mutation profile of BaP obtained by sequencing the lacZ gene[iv], I wondered whether we could use the lacZ gene and its ~3,000 bases to derive mutational signatures. This seemed like a good matchup: we could pair our in vivo mutation patterns from specific chemicals to known mutational signatures, and potentially link previously unknown etiologies to an exposure.

I discussed the idea with my collaborator Carole Yauk, and two members of our group, Marc Beal and Matt Meier, ran with it. Together with our team (including Jason O’Brien, Clotilde Maurice and Danielle LeBlanc) we gathered all our lacZ sequencing data and other published lacZ mutation data resulting in a panel of 10 mutagens with different modes of action and potency. First, we took into account the differences between the lacZ gene and the human genome in the occurrence of the 96 possible trinucleotides that form each signature. Then, we used the most recent Catalogue of Somatic Mutations in Cancer (COSMIC) database[v] to create of a parallel set of signatures that represented how each signature would look like if the lacZ gene was used instead of the human genome. Finally, we developed a workflow that integrated existing algorithms for decomposing mutation profiles into signatures to work with the lacZ sequence (Figure 1).

To our delight, we found that the several mutation profiles of the 10 agents were associated with the expected COSMIC signatures[vi]. For example, the mutation profile of BaP obtained by sequencing the lacZ gene was associated with signature 4, which is found in lung tumors of smokers; the mutation profiles of UVB and sunlight radiation were associated with signatures observed in human skin cancers. We also found that some signatures with unknown etiology contributed to the mutation profile of some of the agents in our study, providing a possible causative factor for these signatures.

[i] Alexandrov LB, et al. Signatures of mutational processes in human cancer. Nature 500, 415-421 (2013).

[ii] Meier MJ, Beal MA, Schoenrock A, Yauk CL, Marchetti F. Whole genome gequencing of the MutaMouse model reveals strain- and colony-level variation, and genomic features of the transgene integration site. Sci Rep 9, 13775 (2019).

[iii] Kucab JE, et al. A Compendium of Mutational Signatures of Environmental Agents. Cell 177, 821-836 e816 (2019).

[iv] Beal MA, Gagne R, Williams A, Marchetti F, Yauk CL. Characterizing Benzo[a]pyrene-induced lacZ mutation spectrum in transgenic mice using next-generation sequencing. BMC Genomics 16, 812 (2015).

[v] Alexandrov LB, et al. The repertoire of mutational signatures in human cancer. Nature 578, 94-101 (2020).

[vi] Beal MA, el al.  Chemically induced mutations in a MutaMouse reporter gene inform mechanism underlying human cancer mutational signatures. Comms Biol (2020)