By Monica Heger
Sequencing both the exome and the transcriptome in cancer samples could help researchers zero in on variants that affect allele-specific expression and loss of heterozygosity, which could ultimately aid in the search for tumor suppressor genes, according to a team of researchers from the J. Craig Venter Institute, the Ludwig Institute for Cancer Research, and other institutions.
According to Ewen Kirkness, an investigator in the genomic medicine group at the J. Craig Venter Institute and an author of a proof-of-principle study outlining the approach, the results provide a shortlist of genes expressed from only one allele. This could be useful in the study of cancer genomes because loss of expression from a single allele has frequently been observed for tumor suppressor genes.
"Haploinsufficiency can have serious consequences if the genes in question are tumor suppressors," Kirkness said. The study, published last month in Genome Biology, also demonstrates that "we should consider more than point mutations and structural rearrangements in tumor cells. Events leading to haploinsufficiency should also be assessed in studies of tumor genomes."
While Kirkness' team is not the first to use the two techniques in combination to study cancer genomes, Mark Rubin, a professor of pathology at Weill Cornell Medical College who has used transcriptome sequencing to identify novel gene fusions in prostate cancer (IS 11/2/2010), said he expects the approach to become more common as it proves to yield clinically relevant results.
"I think this will be one of many papers that will show this type of data," he said. "It's a good way to strengthen the data because you can prioritize variants in genes that are affecting transcription."
The researchers sequenced both the exome and the transcriptome of a breast cancer cell line and a lymphoblast cell line from the same individual. For the exome sequencing portion, they used a NimbleGen capture array and Roche's 454 sequencing with Titanium chemistry. They achieved a median target coverage of 16-fold for both cell lines, covered 94.8 percent of the breast cell line, and 96 percent of the lymphoblast cell line, and called 13,102 SNVs and 14,219 SNVs, respectively.
For the transcriptome sequencing, the team used the Illumina Genome Analyzer, generating 14 gigabase pairs of sequence from the breast cancer cell line and 13.6 gigabase pairs from the lymphoblastoid line.
Comparing the data from the exome sequencing to the transcriptome sequencing, the team identified loss of heterozygosity in 403 genes in the breast cancer cell line, and in one gene in the lymphoblastoid cell line. Additionally, they found 86 and 50 genes, respectively, with allele-specific expression events. Included in the genes identified are known tumor suppressor genes such as BRCA1, MSH3, and SETX. Additionally, the results provide a "short list of novel candidates for the study of tumor suppressor activities," the authors wrote.
Other groups have recently begun using transcriptome sequencing to study disease, including a group at Boston University who is studying lung diseases(IS 10/19/2010). However, using transcriptome sequencing in conjunction with exome sequencing, or whole-genome sequencing, could be an even stronger approach because transcriptome sequencing does not call variants as reliably as exome sequencing, Rubin said. "So, what they're doing is building a solid foundation with exome sequencing … Then they can ask the question, 'How does this [variant] affect transcription?' And that might give us insight into whether this is functionally relevant."
A group from the BC Cancer Agency, for instance, used a combination of whole-genome and transcriptome sequencing to help guide patient treatment (IS 9/28/2010). And researchers from the Dana Farber Cancer Institute are starting a project to use exome and transcriptome sequencing to study drug response in tumors (IS 9/28/2010).
Rubin said that while the study found important results, it is more of a "starting point" because the candidate genes will first have to be validated in actual patient tumor samples, as opposed to cell lines.
Also, the approach, while useful, has important limitations. The level of a particular transcript may not yield any insight into the function of the mutated transcript or gene, Rubin said. Additionally, "the interesting mutations may not be associated with transcripts because they may be preventing the transcripts or genes from being expressed."
Nevertheless, he added, combining the two techniques helps pare down the glut of data generated by exome or whole-genome sequencing, and should allow researchers to focus in on variants associated with gene expression.
The detection of putative tumor suppressor genes could lead to important insights about the biology of cancer, and Kirkness said that he plans to use a similar approach to study additional cancer cell lines.
In terms of implications for drug development, Rubin said that tumor suppressor genes tend to be hard to target, because it is more difficult to design drugs that turn a gene on, rather than knock a gene out. But he said that knowing the tumor suppressor genes could help identify a pathway in which there would be other drug target candidates.
Have topics you'd like to see covered in In Sequence? Email the editor at mheger [at] genomeweb [.] com.