By Monica Heger
While nearly half of all prostate cancers contain gene fusions involving ETS transcription factors, there is still wide heterogeneity in clinical outcomes of the disease, and the underlying mechanism of oncogenesis remains unknown. Hypothesizing that additional rearrangements could be affecting outcomes, researchers from Weill Cornell Medical College and Yale University recently sequenced the transcriptomes of 25 prostate cancer samples, seven of which contained ETS gene fusions and 18 that did not, identifying novel gene fusions and new potential disease pathways.
Steven Jones, associate director at the Genome Sciences Center at the BC Cancer Agency, who has used transcriptome sequencing in conjunction with whole-genome sequencing to determine the best course of treatment for a patient with a rare cancer (IS 9/28/2010), said that the study was a good example of how RNA-seq "has been able to very quickly discover a potential new mechanism in prostate cancer oncogenesis."
Somewhat surprisingly, the novel gene fusions were primarily found in tumors already harboring the common gene fusions. The team is now studying those novel fusions to piece together their functions and pathways, and to determine whether they make good candidate drug targets. Additionally, they are working with the Broad Institute to sequence the transcriptomes and whole genomes of around 100 prostate cancer samples.
In the study, published in Genome Research last week, the team sequenced the transcriptomes of 25 prostate cancers and three normal prostate tissue samples on the Illumina Genome Analyzer using paired-end sequencing. Read lengths varied from 36 base pairs to 54 base pairs and insert libraries increased from 250 base pairs to 400 base pairs over the course of the study. In total, the team generated over 1 billion reads, about 600 million of which were mappable.
To search for gene fusions, they used a computational tool they developed called FusionSeq, which accounts for noise and artifacts to identify only high-confidence chimeric RNA transcripts. Using that method, the team identified seven new gene fusions, which they validated using fluorescence in situ hybridization and RT-PCR.
Of the seven new gene fusions, two involved ETS factors and genes known to be involved in androgen regulation. The other five were rare and involved genes that had not previously been implicated in gene fusions. In addition, they were all found in samples that also contained a common, known fusion between TMPRSS2, a gene involved in androgen regulation, and the transcription factor ERG.
Of the five non-ETS fusions, none of the genes are regulated by androgen. However, one has previously been identified as an oncogene, one as a tumor suppressor, and one of the fusions involves a potential target for which there are already drugs.
Rubin said he was surprised by the results. "We expected that we would identify rearrangements involving tumors that did not already have a known rearrangement," said Mark Rubin, a professor of pathology at Weill Cornell Medical College and a senior author of the paper. "But the novel rearrangements that we found were in tumors that already had a known rearrangement. It suggests that that type of tumor may be susceptible to rearrangements." Additionally, Rubin said it also hints at a separate subtype in the samples without gene fusions.
BC Cancer Agency's Jones added that the while the results may suggest that the known, recurrent mutation is making the tumor susceptible to additional rearrangements, they also suggest that there is an additional driving mutation that is causing chromosomal instability in the first place.
"The rearrangements could be a smoking gun in terms of a loss of chromosomal stability in a way that's going to allow rearrangements to occur," he said.
Increasingly, researchers have been using transcriptome sequencing to study cancer, including Arul Chinnaiyan's group from the University of Michigan, which identified rare, druggable gene fusions in the RAF kinase pathway in prostate cancer using transcriptome sequencing (IS 6/8/2010), as well as groups from the BC Cancer Agency and Boston University (IS 10/19/2010).
Jones said that the combination of paired-end transcriptome sequencing and the team's FusionSeq algorithm allowed the researchers to quite easily filter through the noise and artifacts.
He said that while whole-genome sequencing would have also allowed the team to detect the fusions, they "wouldn't have been able to confirm that a mature transcript was present within a cell, or confirm how precisely it was spliced. That's where RNA-seq really helps."
Rubin added that RNA-seq is an efficient way to discover chimeric fusion transcripts. "Genome sequencing, including exon sequencing, does not provide information regarding which fusions are actively transcribed."
He added that RNA-seq and whole-genome sequencing are good complementary approaches. "Having insight into whole-genome sequencing data, while more costly and difficult to interpret, allows us then to go to RNA-seq, where we have the ability to do more samples and build support for our findings."