NEW YORK (GenomeWeb News) – Researchers from Genentech, Affymetrix, and Pennsylvania State University reported online today in Nature that they have used mismatch repair detection technology and complementary approaches to find somatic mutations in several human cancers.
"We [and others] are trying to build a master catalog of all the changes in cancer," senior author Somasekar Seshagiri, a molecular biology researcher at Genentech, told GenomeWeb Daily News.
"Our focus has been to try to drill down a little bit in terms of thinking about drug targets and being able to turn this information into a useful therapeutic," he added, explaining that the team is particularly interested in getting a more detailed view of cancer sub-types.
By focusing in on roughly 1,500 coding genes in hundreds of breast, lung, ovarian, prostate, and pancreatic tumors, the team found nearly 2,600 somatic mutations, including more than 2,400 mutations that have not been reported in the past. As expected, the sorts of mutations and mutation frequency varied by cancer type, though many of the affected genes tended to fall in cancer-related pathways.
"Clearly the sub-types have a distinct set of mutations, though they may have some commonality," Seshagiri noted.
He and his colleagues used a combination of mismatch repair detection technology and Affymetrix tiling arrays to detect somatic mutations in roughly four million bases of DNA in 441 tumors, including 183 breast tumors, 134 lung tumors, 58 ovarian tumors, and 58 prostate tumors and eight pancreatic tumors.
The mismatch repair detection method — developed by Affymetrix researcher Malek Faham and described in a 2007 Nature Methods paper — relies on screening human DNA in bacteria to find mutations, Seshagiri explained. "The bacteria switches antibiotic resistance based on the presence or absence of mutations."
The researchers specifically focused on genes suspected of playing a role in cancer or related processes, analyzing the coding exons and neighboring splice sites for 1,507 genes.
Overall, the team found 2,576 somatic mutations affecting 967 of these genes. Most of the mutations — some 95 percent — were new, they noted, and were not reported in past cancer studies or in the Catalogue of Somatic Mutations in Cancer, or COSMIC, database.
Within the group of genes that were frequently mutated in their analyses, the researchers detected several protein kinase genes and genes coding for G-protein coupled receptors.
Even so, the frequency and nature of the mutations — as well as the sorts of genes affected — generally varied by tumor type.
"[T]he set of significantly mutated genes varied across each tumor type and subtype," they wrote, "indicating the complexity of the genetic mechanisms underlying carcinogenesis."
For example, although each tumor carried an average of 1.8 predicted protein-altering changes per million bases, the mutation rates were much higher for some cancer sub-types, particularly lung adenocarcinomas and squamous carcinomas. In contrast, the team detected far lower mutation rates for other cancers, including prostate cancers.
Meanwhile, the team's statistical analyses suggested 77 genes were significantly more likely to be mutated in the tumor types tested, including 19 genes in the lung squamous carcinoma and 18 genes in the lung adenocarcinoma samples.
"I think the consensus that's emerging in the field is that cancers are pretty diverse," Seshagiri said. "You're going to get so many different combinations [of somatic mutations] and that's essentially what we find."
When they used Roche 454 sequencing to look at a sub-set of the newly detected mutations in another 50 lung tumors, the team found that mutations in some of the same genes turned up with comparable frequency in these samples as well, Seshagiri noted.
The researchers also used the Agilent 244K array to do comparative genomic hybridization analyses for most of the samples, integrating this data with mutation information to narrow in on potential tumor suppressors or oncogenes.
"In cases where the genes were amplified and mutated, there was a very good chance that they are probably oncogenes," Seshagiri explained. "Where they're deleted and mutated they're likely to function as tumor suppressors."
Among the 35 statistically significant genes identified in this integrated copy number and mutation analyses was the G-protein coupled receptor sub-unit gene GNAS.
For their subsequent experiments, the team homed in on another mutated gene in the G-protein coupled receptor pathway — the G-protein sub-unit gene GNAO1 — as well as a JNK signaling pathway gene MAP2K4, looking at the effects of expressing mutated versions of these genes in cell lines.
The team is currently surveying more cancer samples and sub-types to continue cataloguing cancer-related somatic mutations. By identifying such mutations and teasing apart affected pathways, they hope to not only learn more about the biological changes associated with specific cancer sub-types, but also to uncover candidate pathways for cancer treatments.
"Understanding the set of changes at the individual patient level will enable personalized treatment," they concluded. "Furthermore, patient selection based on tumor mutational profile and genomic alterations for clinical drug testing will be critical for successful development of new treatments."