By Monica Heger
A number of research teams have recently taken steps toward the goal of single-cell sequencing, though they face a number of obstacles that must be addressed before the approach moves into common practice.
Sequencing the genomes of single cells as opposed to bulk tissue could yield insights important to a range of applications such as monitoring cancer progression, in vitro diagnostics, and neurology and stem cell research. For instance, sequencing single cancer cells could overcome the shortcomings of tumor heterogeneity and help pinpoint driver mutations that spurred the initial development of a tumor, which mutations led to metastasis, which were involved with cancer progression, and those associated with resistance to therapy.
Sequencing single cells is tricky, though, and only recently have techniques advanced to the point where it is feasible, although there are still many hurdles to overcome. There are two main problems when it comes to sequencing single cells: Whole-genome amplification protocols do not amplify the genome in an unbiased manner, and PCR steps in library preparation introduce additional biases and errors.
"The main hurdle is amplification bias," said Tomer Kalisky, a postdoctoral researcher in Stephen Quake's lab at Stanford University who recently co-authored a review article in Nature Methods about the importance and challenges of sequencing single cells. "You have to amplify [the genome] so much that you get things that you're not sure are real or not."
Nicholas Navin, an assistant professor at the MD Anderson Cancer Center who reported on the sequencing of 100 single cells from a breast cancer tumor at last year's Biology of Genomes meeting at Cold Spring Harbor Laboratory (IS 5/18/2010), added that another problem with amplification of the genome from just one cell is that it doesn't always cover the entire genome.
"When you amplify DNA from a single cell, you can't amplify the whole genome," he said. "Instead, you get a random amplification of about 10 percent of the genome."
Navin is also working on developing improved methods to sequence single cells. His team is addressing the problem by trying to reduce the amount of input material necessary to make sequencing libraries. Most methods require both a whole-genome amplification step and an additional PCR reaction to get enough material to construct a sequencing library.
He believes future methods for sequencing single cells will focus on either developing better amplification methods or on better detection methods to enable less starting material for sequencing.
He said that transposase-based library prep methods, such as one recently developed by Epicentre Biotechnologies, could have applications for single-cell sequencing because they are faster and require less starting material than other approaches (IS 12/21/2010).
Additionally, the method is able to fragment the DNA and add adapters in the same reaction, reducing the chance for contamination by transferring to a different tube, he said.
Navin has also tested methods by both Rubicon Genomics and Sigma-Aldrich. Each of those has their advantages and disadvantages, he said. In unpublished work, he has found that Rubicon Genomics' PicoPlex method tends to produce better coverage across the genome — covering up to about 20 percent, compared to between 5 and 10 percent by Sigma-Aldrich's method. However, Sigma-Aldrich's method seems to be better for analyzing copy number because its amplification steps do not introduce biases.
BGI researchers are also developing a method for single-cell sequencing. Earlier this month, BGI Americas CEO Xun Xu reported on early results of the effort at the American Association for Cancer Research meeting in Orlando. To amplify the whole genome, the researchers do two rounds of amplification. In the first round, they use a method called multiple displacement amplification, based on an isothermal amplification process that does not require PCR or thermal cycling.
[ pagebreak ]
The MDA step is then is followed by degenerate oligonucleotide primed PCR, which uses a single primer with defined sequences at its 5' and 3'-ends plus a random hexamer sequence between them, instead of pairs of target-specific primer sequences that are used in traditional PCR.
Xu said that the BGI team has been using the method to sequence the exomes from cancer samples. Following the amplification rounds, they do an exome capture step and then sequencing. The researchers are currently sequencing single cells from five different cancer types, Xu reported.
In a sample from renal cell carcinoma, the researchers sequenced the exomes from 400 cells to about 20- to 40-fold coverage, and were able to cover around 85 percent of the targeted region, Xu said.
He reported that the team found that the majority of mutations in the renal carcinoma cancer cells were of medium frequency. They also found that low-frequency mutations tend to appear in key cancer genes that are associated with resistance to therapy. In one example, the team found a mutated tubulin-beta gene, which has been previously reported to be resistant to therapy in breast cancer patients, in only six cells from a patient.
Xu said that the finding could help shed light on cancer progression, with "high-frequency mutations possibly happening in the early stage of cancer progression, while the low-frequency mutations happen in the later development stages."
Aside from cancer research, single-cell sequencing could have important implications for in vitro diagnostics, autoimmune disease, and stem cell research, said MD Anderson's Navin.
For instance, it is standard practice for current in vitro fertilization protocols to use FISH analysis to test individual cells for disorders like trisomy 21. However, eventually it would be possible to do a whole-genome sequencing of that single cell to screen for even more diseases, Navin said.
Jiaqian Wu, a researcher at Stanford University who is developing a single-cell transcriptome sequencing method, said she is most interested in applying the method to neurological research. Within the brain, there are many types of different cells in a small area that need to be "characterized in a more refined way than what we're doing now — taking a lump sum of different types of cells," she said.
However, she said, methods to sequence transcriptomes from single cells have problems with sample contamination and amplification. Because the transcriptome is even smaller than the genome, methods are very sensitive to contamination.
"Even without any input material, the kits amplify stuff," she said, making it necessary to always run a control sample with no input material to determine how much noise will be generated.
Additionally, amplification is necessary, which introduces bias. Methods that use Taq polymerase-based amplification tend to have a bias toward either the 3' or 5' end, and also do not have uniform coverage across GC content. Additionally, "it's really important to have a full-length transcript so you can actually analyze the untranslated regions, alternative splicing, and new exons," she said.
Wu is working on a method that will tackle two problems specific to transcriptome sequencing: maintaining the correct ratios of expressed transcripts and amplifying very long transcripts of up to 10 kilobases.
She did not want to share details of the method because it is still in the early stages of development, but she noted that the technique, when developed, should have important clinical implications.
"A lot of the research into single-cell sequencing is driven by clinical interest," she said. Due to next-generation whole-genome sequencing, researchers are now able to look at areas that they weren't able to look at in detail before, and "now it's very exciting to incorporate that into the single-cell level."
— Bernadette Toner contributed reporting for this article from the AACR conference.
Have topics you'd like to see covered by In Sequence? Contact the editor at mheger [at] genomeweb [.] com.