Researchers at Applied Biosystems and the University of Cambridge have used ABI’s SOLiD sequencing system to profile gene expression in single stem cells, one of the first times that researchers have used second-generation sequencing to analyze single cells.
“It strikes me as really exciting work [and] as an important direction to go because we know that in any dish of, say, stem cells, the cells are not the same,” said Ken Kosik, a researcher at the University of California, Santa Barbara whose lab studies neurons and stem cells.
The work was described for the first time in September at two research symposia in New Jersey and Connecticut that were organized by ABI in collaboration with academic institutions. Kosik, who gave a talk at one of the symposia, has recently begun using the SOLiD system but has not yet applied it to single-cell analyses.
The ABI-Cambridge study resulted from a longstanding collaboration between Kai Lao, a principal scientist at ABI, and Azim Surani, a professor of physiology and reproduction at the University of Cambridge, on single-cell gene expression. In prior studies, they had used real-time PCR and microarrays, according to Lao, which are “good tools for expression profiling of known transcripts.”
In order to discover new transcripts and transcript variants, they started using SOLiD sequencing in May, studying single mouse embryonic stem cells that were manually extracted from embryos at the four-cell stage. The goal of the project, which used embryos that had specific genes knocked out, is to understand gene regulation by maternal microRNAs in mature oocytes, according to Lao.
While the Cambridge researchers generated conditional knockout mice, isolated the embryos, and amplified the RNA, ABI researchers constructed the sequencing libraries, performed the SOLiD runs, and analyzed the data.
To prepare the RNA for sequencing, the researchers needed to remove abundant RNAs, such as rRNA, from the messenger RNA. “However, there didn’t appear to be any established way of doing this for a single cell without jeopardizing the limited quantity of material present,” Lao told In Sequence via e-mail. He and his colleagues developed a new method that involves selectively amplifying RNAs with poly-A-tails, which are mainly mRNAs and some non-coding RNAs. The challenge, he said, was to amplify 500 to 2,000 base pairs of cDNA from picograms of RNA template.
The researchers also had to modify the library construction protocol “to ensure it was robust and reproducible,” he said.
Following sequencing on the SOLiD system, about a third of the reads mapped to 15,000 RefSeq transcripts with more than 10 hits, according to Lao. “We are examining the rest of the millions of SOLiD reads to see if they map to alternative splicing or to non-coding RNAs,” he said.
“It strikes me as really exciting work [and] as an important direction to go.”
The approach “not only allows us to discover many novel transcripts that have been overlooked,” especially low-abundance transcripts, “but also to get a quantitative estimate of their abundance in the cell by the frequency with which the sequence occurs in the SOLiD reads,” he said. Lao and his colleagues plan to publish their results after analyzing more ES cells with conditional knockouts in the Dicer and Ago2 genes, he said.
“The study has shown the feasibility of single-cell next-generation sequencing,” said Brent Graveley, an associate professor of genetics and development biology at the University of Connecticut Stem Cell Institute, who gave a talk at one of the symposia and heard Lao speak. “While the data I saw was quite convincing, further work remains to be done to determine how reproducible the assay is,” he said.
The main advantage of studying transcription in single embryonic stem cells rather than ES cultures is that the cells are “notorious for their variability,” according to Graveley, whose lab uses Illumina’s Genome Analyzer to study gene expression in cultures of human ES cells.
By analyzing gene expression in multiple individual cells, “you get an idea of which genes are expressed in a stable manner and which genes are expressed in a stochastic or variable manner,” he said.
However, single cell studies are “more technically difficult and much more expensive” because more replicates are required, he added.
“The challenges are going to be both in the sample preparation steps and then, post sequencing, getting used to possibly very dramatic differences in transcript levels [between] cells next each other [that are] presumably doing the same biochemistry,” noted Bob Setterquist, director of scientific operations for expression analysis at ABI. He is working with another, undisclosed researcher who has worked on single-cell expression profiling. “We have been profiling RNA mixtures from millions of cells for so many years that researchers somewhat naively think all the cells are acting as one.”
But ABI is not the only company that believes its second-generation sequencing technology can be used to study minute amounts of RNA. "We know of several research groups, including some of our close collaborators, who are developing methods to study very low input amounts for both small RNA and mRNA sequencing applications using the Illumina GA system," said Gary Schroth, senior director for expression applications R&D involving sequencing at Illumina, in an e-mail.
Schroth added that he believes the sample prep method developed by the ABI-Cambridge team would likely also work on the Illumina sequencer. “There is nothing platform-specific about this application," he said.
Several aspects of the method are already available to SOLiD customers, according to Lao, such as a protocol for low DNA input.
Other researchers with limited clinical material to study could also benefit from the approach, and Lao said he and his colleagues are planning to establish more collaborations with cancer and neurobiology researchers “to help optimize this protocol as well as explore molecular signatures by cell type and stages of cell development.”
They also plan to use “other” technologies from ABI to “explore limited sample analysis for small RNA and whole-transcriptome analysis,” he said, but did not elaborate.