By Monica Heger
Researchers from Helicos and Massachusetts General Hospital Cancer Center have used amplification-free digital gene-expression profiling to discover potential new biomarkers for cancer.
The researchers, who described the study in Science last week, used the Helicos platform to sequence the transcriptomes of mice pancreatic tumors, and found an abundance of a particular type of repetitive, non-coding RNA transcript, known as a satellite repeat, which is linked to gene silencing and chromosome maintenance and could potentially be used as a biomarker for early diagnosis of cancer.
The digital gene-expression method, which the partners developed last year in order to perform transcriptional profiling in circulating tumor cells (IS 6/27/2010), allowed them to discover the previously undetected transcripts because it provides "a quantitative and sequence-specific measure of highly repetitive sequences that are excluded from traditional analytic programs," the authors said in the paper.
David Ting, co-lead author on the paper and a researcher in Daniel Haber's lab at MGH, said the team was sequencing the transcriptomes of primary pancreatic tumors from mice as part of its work to develop a method for sequencing circulating tumor cells. In the process, they discovered that the majority of sequences they were discovering did not match the known transcriptome.
"We wanted to figure out where it was coming from," Ting said. When they aligned them to the genome, they discovered that the discrepant sequence was coming from the major satellite repeats. "That was a big surprise," Ting said.
Satellite repeats are large repetitive areas that are found near the centromeres and telomeres. They don't have a promoter or normal gene architecture, so most researchers didn't think those areas were transcribing, he added.
The Heliscope is the only sequencer at the cancer center, said Ting, and the group originally decided on that technology in order to study circulating tumor cells. "They are rare cells, and since we don't have much material, we didn't want to skew the expression profile," he explained.
Ting added that the Helicos protocol for sequencing the transcriptome turned out to be amenable to discovering these satellite repeats because it does not have an amplification step, and it also sequences native RNA. As a result, it offers advantages over microarrays, which rely on previously annotated transcripts, as well as other analytical methods that exclude repeat sequences.
The Helicos system "has the resolution to count individual transcripts," Ting said. While he acknowledged that he wasn't sure whether other sequencing platforms would have produced the same results, he noted that most sequencing software is designed to mask repeats, because they are generally thought to be unimportant.
In the study, the team first sequenced a mouse pancreatic tumor with an activated KRAS oncogene and the loss of the tumor suppressor gene TP53. They found that 47 percent of the transcripts sequenced mapped to the major mouse satellite, a 100-fold increase over normal tissue levels. Analyzing additional mice tumors, the team discovered increased satellite expression in seven out of nine pancreatic tumors, two out of three colon tumors, and two out of two lung tumors.
Next, the team wanted to determine whether the finding held true for human tumors. They used the digital gene-expression method to analyze the transcriptomes of 15 human pancreatic tumors, and found a median 21-fold increase in expression of satellite transcripts compared with normal pancreatic tissue. Increases in satellite transcripts were also present in human tumors from the lung, kidney, ovarian, and prostate cancer.
Interestingly, the different tumors showed different expression levels of different types of satellite transcripts. The satellite transcript that showed the greatest difference in expression between normal and tumor tissue was HSATII, which is undetectable in normal human pancreas, but comprised just over 10 percent of the satellite reads in the tumor, a 131-fold increase in expression. HSATII was also present at high levels in lung, kidney, ovarian, and prostate cancers, "indicating that this may be a shared feature of a variety of carcinomas," the authors wrote.
Other satellite transcripts were lower in tumor tissue compared to normal. The most abundant class of normally expressed satellite, ALR, showed a 43-fold increase in expression in the pancreatic tumors.
They then tested the technique on clinical samples, taking fine-needle aspirates of pancreatic masses. In 10 of 10 cases that had pancreatic cancer at the time of resection, HSATII-positive cells were identified, suggesting it could be a useful biomarker.
These results suggest that "satellite expression occurs fairly early in tumorigenesis, so has implications for early diagnosis," Ting said.
Additionally, the researchers also examined other genes to see whether their expression correlated with increased expression of the satellite transcripts. Ting said they identified around 300 genes, mainly related to neural and stem cells, that behaved similarly to the satellite transcripts.
"It points to something going on in the tumor that is turning it into a stem cell," Ting said.
However, Ting said that these genes would have to be studied further before any conclusions were made. He said the team's next steps were to study additional tumors from pancreatic, lung, kidney, ovarian, and prostate cancers to confirm their results and figure out why the satellite transcripts are being expressed in tumors. They will also try to determine whether the satellites can be detected in circulating tumor cells.
Additionally, they will continue to pursue how satellites could be used as biomarkers for early cancer diagnosis.
"Currently, it's challenging to look at a few cells and make that diagnosis," Ting said. "Maybe the satellites will help clinicians say this is cancer or not, but that's still something that a larger study needs to address."
Have topics you'd like to see covered in In Sequence? Contact the editor at mheger [at] genomeweb [.] com.