Skip to main content
Premium Trial:

Request an Annual Quote

MGH, Helicos Develop Amplification-Free DGE Method for Analyzing Rare Cells


By Julia Karow

Researchers at Helicos BioSciences and Massachusetts General Hospital Cancer Center have jointly developed an amplification-free digital gene expression profiling method on the Helicos platform that allows them to analyze as few as several hundred cells at a time.

The method, called low-quantity digital gene expression, or LQ-DGE, was published online last week in Nature Methods. Helicos said it is currently assessing the technique for its "potential to advance molecular diagnostic assays for patients suffering from breast and ovarian cancer."

David Ting, a researcher in Daniel Haber's lab at MHG and one of the authors of the report, told In Sequence that his team's original motivation for developing the method was to enable transcriptional profiling in circulating tumor cells that have been enriched from a patient's blood. Standard microarray techniques have failed for this because the cells are rare and only a few hundred to a few thousand are typically captured.

Haber and his colleagues also wanted to avoid amplification, he said, because they found that many amplification kits distort the true transcriptional profile, even in cell lines.

Ting and his colleagues developed the method in collaboration with Helicos, and most of the sequencing was done on the sequencer that Helicos placed at MGH last year.

"We believe that we have overcome a major technological barrier to investigating transcriptional profiles of rare cells," he said. The researchers are now using the method to study circulating tumor cells in mouse and human samples. Other groups at the MGH Cancer Center, Ting said, are interested in applying it to analyzing cancer stem cells.

For LQ-DGE, the scientists first capture mRNA, with poly-A-tails, from cell lysates using poly-T-coated Helicos flow cells. Because the RNA does not need to be isolated first, it is in its native state, Ting noted. They then use reverse transcriptase to generate cDNA on the surface of the flow cell, add a guanine tail for priming, and sequence it on the Helicos platform.

Initially, the researchers tested their method on two cell lines — one derived from a pre-cancerous pancreatic lesion in an engineered mouse, the other a malignant pancreatic ductal adenocarcinoma cell line. They chose these because "we wanted to resolve transcriptional differences between very similar cells," Ting explained.

They analyzed between 250 and 16,000 cells, containing an estimated 2.5 nanograms to 1.6 micrograms of RNA. Even 250 cells "generated sufficient usable reads for digital gene expression profiling," they noted in the paper, and measurements from 1,000 cells were "highly reproducible," although the cell lysis method had some effect on the results.

Using one cell line, they also compared LQ-DGE on 1,000 cells with Helicos' standard DGE method, which starts with isolated RNA and requires several million cells to get enough RNA, and found that the two datasets correlated well. They did find some differences, Ting said, which were expected because even within a population from the same cell line, gene expression can vary.

To see whether LQ-DGE is able to pick up differentially expressed genes, the researchers compared profiles from the two cell lines and found that more than 2,000 genes with greater than 10 transcripts per million reads were expressed at more than a two-fold difference. They validated results for several of these genes using real-time PCR and found good agreement between the two methods.

In addition, in a proof-of-principle experiment, they tested their method on formalin-fixed, paraffin-embedded cancer samples, which contain degraded RNA, comparing them with matched frozen tumor samples, and found "fairly concordant" expression profiles. This application is "not our primary focus in our lab," Ting said, but people "who want to do transcriptional profiling from FFPE would find this an interesting method to use."

[ pagebreak ]

The main advantage of LQ-DGE, he said, is that it does not require RNA to be isolated or amplified, thus avoiding biases. However, the method "may still suffer from common cDNA synthesis artifacts," the authors wrote, such as spurious second-strand formation and reverse transcriptase-related biases due to RNA structure.

Also, only 15 percent to 25 percent of the templates give rise to reads that can be aligned to reference sequences, the authors noted, a problem that could be solved by multipass sequencing, or sequencing the same template repeatedly after melting away the newly synthesized strand. According to the paper, the researchers have already shown proof-of-concept for this, and Ting said the challenge will be to melt off the strands efficiently without damaging the template.

Cell preparation also needs to be optimized in order to apply the method successfully to analyzing rare cells. Stem cells, for example, are commonly isolated by a flow sorter, Ting said, in which they "undergo some rough conditions," so the quality of the RNA may be compromised.

Ting and his colleagues are currently applying LQ-DGE to study circulating tumor cells in a mouse model of pancreatic cancer and in human clinical samples from a variety of cancers. This could help them understand what role these cells play in metastasis, and enable them to discover biomarkers and new therapeutic targets, he said.

Helicos said in a statement that it is "assessing techniques described in this publication for their potential to advance molecular diagnostic assays for patients suffering from breast and ovarian cancer."

Patrice Milos, the company's chief scientific officer, told In Sequence by e-mail that "the work is currently focused on the ability to work with smaller and smaller cell quantities to address the quantitative expression profile of circulating tumor cells by directly capturing nucleic acid derived from the patient to allow one the potential for patient monitoring."

Ting said he and his colleagues are currently not involved in Helicos' efforts to improve diagnostic assays for breast and ovarian cancer, but he said he believes the method could be useful for transcriptional profiling of diagnostic biopsies, where only small numbers of cells may be available.

According to Azim Surani, a professor of physiology and reproduction at the Gurdon Institute at the University of Cambridge who was not involved in the publication, the method could potentially be used for high-throughput analysis and is "relatively fast" to generate results, which "could be useful for diagnostic purposes."

He said in an e-mail message that the ability to carry out all reactions on the surface of the flow cell is an advantage because they could potentially be automated, while current RNA-seq methods require multiple pipetting steps in tubes. He also agreed that the lack of PCR amplification is an advantage.

Surani and his colleagues recently published a single-cell transcriptome sequencing, or RNA-seq, method that uses the Applied Biosystems SOLiD platform (IS 7/13/2010). LQ-DGE cannot currently analyze single cells but requires hundreds of cells, so "if the cells are indeed rare, they may not detect them by the LQ-DGE system," he said.

He also pointed out that his group's RNA-seq method can generate between 1 and 3 kilobases of sequence per transcript, whereas LQ-DGE only produces short sequence tags. However, earlier this year, Helicos published another method, called low-quantity RNA sequencing, or LQ-RNA-seq, that allows researchers to sequence entire transcripts. That method starts with isolated RNA, though, and does not generate the cDNA on the flow cell.

Earlier this year, another group, at the Wellcome Trust Sanger Institute, published a transcriptome sequencing method for the Illumina Genome Analyzer that, like the Helicos LQ-DGE, generates the cDNA on the flow cell and does not include a PCR amplification step (IS 1/19/2010), however, it starts with RNA rather than cell lysate.

Ting said a future goal is to develop the LQ-DGE method further to work with even fewer cells, down to single cells. "For circulating tumor cells and stem cells, being able to do single-cell full transcriptional profiling would be a tremendous advance," he said. "I think that's a lofty goal, but you have to have goals. It's something we would want to do in the future, and we would hope we can do with Helicos."

The Scan

Cell Atlas of Human Lung Development Gives View of Developing Airway

Researchers have generated a cell atlas of human lung development, which they report in Cell.

Study Finds Costs of Genome Sequencing May Limit Utility in Routine Care

Researchers report in the European Journal of Human Genetics that genome sequencing for rare disease diagnoses currently has similar benefits as less expensive exome analysis.

Study Suggests Nursing Mother's Diet Can Impact Offspring's Gut Microbiome

A new Cell Host and Microbe paper finds that mice whose mothers were fed a low-fiber diet during nursing experience lasting microbiota dysbiosis and increased obesity.

Study Links Genetic Risk for ADHD With Alzheimer's Disease

A higher polygenic risk score for attention-deficit/hyperactivity disorder is also linked to cognitive decline and Alzheimer's disease, a new study in Molecular Psychiatry finds.