NEW YORK (GenomeWeb) – A team led by researchers at the Babraham Institute and the European Bioinformatics Institute in Cambridge, UK, has developed a method to sequence the methylomes and transcriptomes of single cells in parallel.
In a paper published today in Nature Methods, the researchers reported details of the approach, called scM&T-seq, and described using it to profile the transcriptomes and epigenomes of mouse embryonic stem cells, finding both known and new associations.
The new method is a combination of two previously developed methods, for single-cell bisulfite sequencing and for combined single-cell genome and transcriptome sequencing. In 2014, Babraham researchers published an approach for single-cell bisulfite sequencing, dubbed scBS-seq, and last year, a team at the Wellcome Trust Sanger Institute and their colleagues published Genome and Transcriptome Sequencing, or G&T-seq.
"It was a very obvious thing to do, to see if we could combine G&T-seq with our bisulfite sequencing [method] in order to get epigenetic and transcriptome information for the first time from the same single cell," said Gavin Kelsey, a principal research scientist at the Babraham Institute and one of the authors of today's publication.
The new method will be especially useful to help researchers understand how gene expression is regulated by epigenetic decisions during mammalian embryonic development, he said, as well as to study the heterogeneity of cell populations in cancer.
For G&T-seq, RNA and DNA from a single cell are physically separated so they can each be profiled on their own. For scBS-seq, the DNA is further bisulfite-converted to be able to detect DNA methylation.
Combining the two methods required the team to optimize the recovery of DNA from single cells, Kelsey said, making sure that the bisulfite treatment converted the DNA well and that as little DNA as possible was lost in the process. "We have got it up to a pretty good state now, where we can gain information from, on average, 10 to 20 percent of the genome at the level of single CpGs for DNA methylation," he said.
To validate scM&T-seq, the team applied it to mouse embryonic stem cells and compared the methylation results to previously generated standalone scBS-seq data. "We found, after the optimization of the methods, that indeed the DNA methylation data using the combined approach was just as good as using the standalone bisulfite sequencing," Kelsey said.
"Ultimately, the ambition of an experiment is going to be limited by how much one can afford to sequence," he said, noting that at the moment, the researchers can analyze about 20 single cells per lane on an Illumina HiSeq.
Scientists are already applying scM&T-seq in a variety of studies, for example to see how the link between methylation and transcription varies throughout stem cell differentiation. They are also using it to study mammalian preimplantation embryos, Kelsey said, to see at what stages of embryonic development methylation and transcription are coupled, to what extent methylation status regulates the function of enhancer elements, and what impact this has on how cell lineages are allocated in the developing embryo.
There are still some technical improvements to be made to scM&T-seq, he said, for example increasing the percentage of mapping reads from bisulfite sequencing, which could be achieved by optimizing the purification of DNA after the bisulfite reaction to suppress undesired sequences.
Scaling the method to larger numbers of single cells is another goal, in particular for bisulfite sequencing. Over the last couple of years, the researchers have already improved the throughput: while in 2014, they were only able to do scBS-seq on a dozen or so cells per week, they have now optimized many steps of the process and can profile a plate of 96 cells in a matter of days, Kelsey said.
In the future, scBS-seq could also be combined with single-cell parameters other than the transcriptome, he suggested, for example with chromatin conformation mapping.
There are "ongoing discussions" with a company to commercialize the method, he said, but there is not much intellectual property associated with scM&T-seq because it combines existing methods, "although there is a lot of interest in the market in having a whole suite of single-cell methods that could be put together."