By Julia Karow
Scientists at Pacific BioSciences have demonstrated that they can directly analyze several types of DNA methylation during single-molecule real-time sequencing, obtaining sequence and methylation data at the same time.
In a proof-of-principle paper published in Nature Methods this week, they showed that by measuring polymerase kinetics, they were able to detect and distinguish between three types of methylation in synthetic DNA, and identify methylation in genomic DNA. By sequencing the same molecule several times over, they could also detect methylated bases in single DNA molecules. Unlike other methods, theirs does not require methylated DNA to be prepared in any special way.
PacBio, which is scheduled to ship instruments to its first early-access customers this summer (IS 2/23/2010), plans to make methylation sequencing available to customers in mid-2011. Since submitting the paper, the researchers have seen evidence that the method also works for other types of DNA modification, and that it can be used for de novo methylation sequencing.
One of the biggest benefits of the new method, which PacBio first revealed publicly last fall (IS 9/22/2010), will be the ability to obtain base sequence data and methylation information simultaneously, even if a researcher did not set out to study methylation. "I suspect that the most important consequences of this technology will be the unanticipated discoveries in the field of epigenomics," said Steve Turner, PacBio's chief technology officer.
And unlike sequencing of bisulfite-treated DNA — one of the most common methods for measuring cytosine methylation today — the company's technique can distinguish between 5-methylcytosine and 5-hydroxymethylcytosine. According to the firm, it is also less expensive and time-consuming than bisulfite sequencing.
Others, such as Oxford Nanopore Technologies, have also shown that they can distinguish between methylated and unmethylated cytosine (IS 2/24/2009), but have so far not demonstrated that they can generate DNA sequence data at the same time.
PacBio's method relies on measuring the duration of and intervals between pulses of fluorescence that occur when a labeled nucleotide is incorporated by the polymerase. These metrics, the researchers found, differ for each type of nucleotide, and are also affected by the sequence context.
In the paper, the researchers first studied synthetic oligonucleotides about 200 bases in length that had deliberately placed 5-methylcytosines, 5-hydroxymethylcytosines, and N6-methyladenines, and showed that each of them produced distinct kinetic signatures.
They then went on to show that by sequencing a single oligonucleotide repeatedly, they were able to determine which adenine was methylated. For methylated cytosine, they wrote, "enhancements of kinetic sensitivity will likely be required."
[ pagebreak ]
Finally, moving on to genomic DNA, they sequenced a 3.7-kilobase fragment from the C. elegans genome that had been cultured in an E. coli strain that methylates adenines and, again, were able to identify the methylated bases.
All the work was done on a prototype instrument with a low throughput, Turner noted, which limited them to analyzing bacterial-type modifications. Next on the list, he said, is more complex mammalian DNA with cytosine modifications.
Because the method is still in early development, no collaborators outside of PacBio have had access to it yet. The company is currently developing algorithms to process the data, as well as a comprehensive database of kinetic reference data from unmethylated DNA to account for sequence context, so researchers will not have to sequence control DNA each time. They are also modifying the polymerase enzyme to make it more sensitive to methylation. The plan is to make methylation sequencing available to customers in mid-2011.
PacBio is also working on extending the method to other base modifications besides methylation, and might develop polymerases optimized for each type. "The technique is extremely sensitive, and we have tested a variety of different types of modified nucleobases," Turner said. "Our experience so far is that every single modified nucleobase that we have put into the system has produced a detectable signal."
The authors also note in the paper that the approach could eventually be useful for de novo detection of methylation in the absence of a reference database, "by tabulating expected kinetics over a suitable number of contexts or by taking advantage of heuristics that embody the observed trends in SMRT sequencing kinetics." Indeed, since submitting the paper, the researchers have found that the relevant sequence context is "small and predictable enough" for de novo methylation sequencing, Turner said.
According to Michael Freitag, an assistant professor of biochemistry and biophysics at Oregon State University, the main advantages of PacBio's methylation detection method are that it is direct, requiring no special preparation of methylated DNA, and that it has the potential to generate much longer reads than other technologies, such as the Illumina and 454 platforms he has worked with. "But there are lots of unknowns still," he said, including how error rates will be quantified. "That will become more obvious once they release a machine."
In the future, Freitag said, it is conceivable that the platform could be used for diagnostic applications that require the methylation status of DNA to be determined, for example in cancer. "This would be a great tool for that," he said, because it is probably going to be faster and less expensive than bisulfite sequencing.
And in addition, PacBio's method could enable scientists to learn more about the mechanisms of DNA polymerase enzyme kinetics, a field he said has laid dormant for a long time. "Their paper is interesting even from a basic science point of view — clearly not your average 'methods' paper," he said.