An in-depth computational analysis is helping to unravel the molecular mechanisms by which smoking increases the likelihood of cancer. The researchers uncovered several distinct smoking-related mutational signatures, including one thought to be the direct consequence of replicating carcinogen-damaged DNA, and another that may result in a normal clocklike mutational process being sped up.

An in-depth computational analysis led by Los Alamos National Laboratory (LANL), NM, and the Wellcome Trust Sanger Institute in Cambridge, UK, is helping to unravel the molecular mechanisms by which smoking increases the likelihood of cancer.

The researchers analyzed 5,243 exome and whole-genome sequences from 17 tumor types for which smoking is a known risk factor, including lung, bladder, and pancreatic cancers. These specimens encompassed smokers and nonsmokers, and total numbers of different somatic mutations such as single-base substitutions and copy-number variations were compared between the two groups. Smokers had considerably more single-base substitutions, the team found.

Previous work by lead author Ludmil Alexandrov, PhD, and others had shown that single-base substitutions have their own fingerprint—a distinct pattern of bases immediately before and after the one that's altered, depending on what caused the change. In the current study, the researchers observed that five of these mutational signatures had a higher incidence among smokers, including one mainly characterized by C-to-A changes.

“Signature 4 was elevated in lung adenocarcinoma, larynx cancer, and esophageal cancer—all tissues that are directly exposed to tobacco smoke,” says Alexandrov, a research fellow at LANL. Noting that it resembles a mutation pattern produced experimentally by treating cells with the carcinogen benzo[a]pyrene, he adds that “this signature is likely an immediate consequence of tobacco carcinogens entering cell nuclei, where they induce DNA damage that's then misreplicated.”

Signature 5, predominantly characterized by substitutions of T with C or vice versa, was elevated in smokers across nine different cancers. It is thought to contribute to the normal accumulation of mutations over time in all cells, and “we think smoking somehow speeds up this clocklike mutational process, although we have yet to figure out the exact mechanism,” Alexandrov explains. His team also traced the cause of two additional molecular fingerprints in lung adenocarcinoma to overactive DNA editing by APOBEC enzymes, which they hypothesize may be switched on by smoke-induced inflammation.

Although differences in DNA methylation between smokers and nonsmokers with cancer have been reported, the researchers were unable to corroborate these findings: Overall methylation levels in both groups were similar across the 17 cancers they assessed. However, “it's possible that smoking does change the methylome, and this difference is simply obscured by other, bigger neoplastic events in tumor evolution,” Alexandrov says.

To Paul Brennan, PhD, head of genetics at the International Agency for Research on Cancer in Lyon, France, “the discovery that smoking results in distinct mutational signatures is exciting but not entirely surprising, given the large number of known tobacco carcinogens.” The challenge, he says, will be “going beyond tobacco smoke and generating similar connections for other lifestyle and environmental factors.”

Alexandrov, who credits LANL's supercomputing resources for enabling his team to speedily sift through thousands of cancer genome sequences—each of which contained several hundred gigabytes of unstructured data—agrees. He's now using the same approach to better understand mechanisms underlying the influence of obesity, sex, and geographic location on cancer risk. –Alissa Poh