NEW YORK (GenomeWeb News) – Through whole-genome sequencing of several matched tumor-normal pairs, researchers from the US, Israel, and Spain have identified a recurrent gene fusion that turns up in some colorectal cancers.
As they reported online in Nature Genetics yesterday, the group sequenced the genomes of nine colorectal tumors and matched normal tissue samples. In the process, they identified a few dozen genes that harbored possibly deleterious substitutions, small insertions, or deletions in two or more of the tumor genomes, along with 11 rearrangements predicted to produce fusion proteins.
When they screened 97 other colorectal tumors, the team found that one of the fusions, which involved the genes VTI1A and TCF7L2, showed up in three more colorectal cancers.
"Our results provide no evidence for high-frequency recurrent translocation, such as those that are seen in prostate adenocarcinoma," co-corresponding author Matthew Meyerson, an oncology and pathology researcher affiliated with the Dana-Farber Cancer Institute, the Broad Institute, and Harvard Medical School, and co-authors wrote.
"However, the discovery of the recurrent VTI1A-TCF7L2 fusion in [three percent] of colorectal cancers shows that functionally important fusion events occur in this disease," they added, "and suggest that further structural characterization will likely identify additional new recurrent rearrangements."
Past genetic studies of colorectal cancer, including exome-sequencing studies, have pointed to several genes and pathways that tend to be altered in the disease, the team explained, noting that the tumors also tend to show chromosomal and/or microsatellite instability.
In the current study, researchers looked at the genome-wide mutation patterns in this type of cancer by sequencing the genomes of nine colorectal tumor samples, all collected prior to pre-chemotherapy or radiation treatment, to an average of 30.7 times coverage using paired-end 101 base Illumina GAII reads.
They used a similar strategy to sequence the genomes of matched normal tissues sampled near the tumor sites from each of the nine patients to 31.9 times coverage, on average.
In their ensuing analyses of the cancer genomes, the study authors tracked down 137,968 potential somatic mutations, including 712 that appear to cause non-synonymous substitutions, insertions, or deletions within protein-coding sequences. Of the 521 apparently deleterious mutations they were able to test by mass spectrometry-based genotyping, the team found that 439 of these were authentic somatic changes.
Based on the patterns detected in the nine tumor samples, the team estimated that each colorectal tumor genome contains around five mutations per million bases of sequence.
The rate seems to be slightly higher — about 6.7 mutations per million bases — in the sequences between genes than it is in the introns of genes, they reported, where it's expected to be closer to 4.8 mutations per million bases, they reported.
The exons of genes had the lowest predicted mutation rate, at around 4.2 mutations per million bases overall. And for every million bases of exon sequence, the researchers estimated that there were roughly 3.1 non-synonymous mutations in the tumors.
Consistent with past studies, the researchers noted, mutation rates tended to be higher overall in parts of the genome where cytosine and guanine are found together than in non-CpG sites.
While they detected relatively few small insertions and deletions affecting protein-coding sequences, the team did detect two-dozen genes affected by either indels or non-synonymous substitutions in at least two of the tumors. These mutated genes included notorious cancer-related genes such as TP53 and KRAS as well as genes previously implicated in colorectal cancer, including NRAS, SMAD4, and PIK3Ca.
Even so, the study authors explained, many of these genes did not have statistically significant mutation rates because of the relatively small number of samples tested. "Large sequencing projects will be needed to identify a fuller set of genes with significant recurrent mutations," they wrote, "such projects are now being carried out under The Cancer Genome Atlas."
Along with these mutations, researchers detected 675 possible somatic rearrangements in the colorectal cancers, an average of 75 per tumor tested. Of these, 11 rearrangements — two inter-chromosomal and nine intra-chromosomal — were predicted to produce the sorts of "in frame" coding changes expected to produce fusion proteins.
One of the in-frame fusions involves a chromosome gene called TCF7L2, the researchers reported, which codes for a transcription factor called TCF4 that was previously implicated in colorectal cancer-related carcinogenesis.
Though the fusion does not appear to produce a TCF4 binding domain used to interact with beta-catenin, an apparent partner for TCF4 in this cancer-promoting process, the researchers decided to look at the fusion between TCF7L2 and another chromosome 10 gene, VTI1A, in more detail. Indeed, when the team screened for the fusion transcripts in a panel of 97 primary colorectal cancer samples, they found the VTI1A-TCF7L2 fusion in three more colorectal cancers.
The team's follow-up experiments in a colorectal cancer cell line containing a VTI1A-TCF7L2 fusion similar to that identified in the study indicated that the fusion seems to contribute to cancer cell growth.
"When coupled to the recent report of TCF7L2 mutations in colorectal cancer and evidence that TCF4 can also have tumor suppressive functions in colorectal neoplasia," the researchers wrote, "these data suggest additional complexity regarding the function of [beta-catenin] and its cooperating factors in colorectal cancer."