Skip to main content
Premium Trial:

Request an Annual Quote

Proof-of-Principle Study Suggests Single-Molecule Sequencing Possible by Mechanical Means


By Andrea Anderson

This article has been updated from a previous version to correct Vincent Croquette's principal affiliation.

A French and American research team has developed a hairpin DNA-based method that appears to have potential for single-molecule DNA identification and sequencing.

The approach, described in a proof-of-principle study online this weekend in Nature Methods, relies on measuring subtle changes in DNA hairpin length — a departure from most single-molecule sequencing methods, which either incorporate fluorescently labeled nucleotides or track the electrical signals associated with specific bases. Using this system, the team showed that they could find sequences of interest and sequence short stretches of single-molecule DNA by hybridization or sequencing-by-ligation strategies.

"This is a new platform, completely different from what has been done so far," the study's senior author Vincent Croquette, a biophysics researcher at Laboratoire de Physique Statistique, Ecole Normale Supérieure, in Paris, told In Sequence.

Croquette and colleagues are forming a startup called PicoSeq that will initially focus on manufacturing an apparatus for using hairpin DNA methodology in DNA identification, but they eventually aim to develop a sequencing instrument as well.

The approach is still in its infancy, though, and more work will be needed to not only scale up the process, but also to get a better sense of the achievable read lengths and throughput.

In an e-mail message to IS, University of California, Santa Cruz, biophysics laboratory co-director Mark Akeson said he thinks the technology may be useful.

"This is a nice piece of work from groups with solid reputations in biophysics and molecular biology," noted Akeson, who was not involved in the study.

"My one caution, which I stress within our group as well, is that any new approach to DNA sequencing practiced by an academic laboratory needs to be compared with what companies like Illumina, Pacific Biosciences, or Ion Torrent will be offering five years from now," he added.

"Today's commercial specs are not the correct measure — we target 100X better. For example, assuming a device capable of a 1,000 nucleotide read length for sale today, a competitive academic device would be demonstrably capable of 100,000 nucleotides so that when it goes commercial in five years it will be competitive."

Akeson and fellow UCSC researcher David Deamer collaborated with Harvard University researchers Daniel Branton, Jene Golovchenko, and George Church and the National Institute of Standards and Technology's John Kasianowicz on solid-state nanopore methods licensed by Oxford Nanopore in 2008 (IS 8/12/2008).

Croquette said his team's methodology grew out of magnetic tweezer and optical tweezer experiments originally concerned with studying DNA elasticity and related physical questions.

"At the beginning, we were just pulling on DNA, trying to measure the elasticity of the molecule," he said.

As the work progressed, the team began to see their DNA hairpin-based approach as a potential stepping stone to sequencing systems that don't require fluorescence detection or DNA amplification.

Hairpins and Beads

Generally speaking, the method involves fixing one end of a DNA hairpin to a solid surface and the other end to a magnetic bead.

In the new Nature Methods study, for example, researchers created hairpins by ligating a sequence of interest between a DNA loop and a DNA fork labeled with biotin on one side and the plant steroid digoxigenin on the other. The hairpins can then be tethered to a glass cover slip at one end using an antibody targeting digoxigenin and to a streptavidin-coated magnetic bead at the other.

A magnetic force then pulls on the bead, drawing the DNA hairpin apart. With the hairpin open, researchers can toss oligonucleotides at the DNA sequence in the middle as they ease off on the magnetic force holding the structure open.

When oligos do hybridize to the sequence of interest, it prevents the hairpin from snapping closed again. The bead's distance from the surface relative to where it is when the hairpin is fully open or closed — as measured by a magnetic tweezers device — offers information about where this oligo-induced blockage is happening.

Since they could track bead position with a resolution of almost one nanometer, nearing the single-base level, the researchers suspected it would be possible to tweak this system to track down sequences on a single molecule of DNA — or even determine its sequence.

In their earliest attempts to do sequencing with this system, the team tried a single-molecule version of the chain terminator strategy used for Sanger sequencing, Croquette explained.

For those preliminary experiments, not described in the new paper, researchers relied on the T4 polymerase, an enzyme that they found was active on hairpin DNA stretched straight by magnets, but which processively chewed up newly synthesized DNA when this magnetic force disappeared.

By using magnetic bead position to track DNA extension and termination at a specific base on a single molecule over many cycles, the team hoped to determine every position for each of the four bases.

That approach "nearly worked," but was thwarted by a lack of single-base resolution in the system, which was a particular problem for homopolymers, Croquette explained. "Since we have one nanometer resolution, we just barely see one base. So if we have two "A" bases right after another, we don't distinguish them."

The team had more success when it attempted to marry its hairpin DNA system to other sequencing methods, specifically hybridization and sequencing-by-ligation strategies.

On the hybridization side, the researchers took advantage of the hairpin blockages that occur when oligos hybridize to sequence in the hairpin, using magnetic bead position to determine when and where each oligo hybridized to the hairpin.

"If your oligo matches the sequence, you will see a blockage," Croquette said. "You know where the oligo is hybridizing with nearly single-base-pair resolution."

Flushing enough oligos through the system allows for sequencing by hybridization, he explained, pointing to a 2008 study by Swedish researchers who sequenced the Escherichia coli genome by hybridization using 512 fluorescently labeled, tiling oligos.

Croquette said it is theoretically possible to do the same thing at the single-molecule level using the hairpin DNA system rather than fluorescence to track hybridization.

But because it would require running hundreds of oligos through the system, that is not something that the researchers have demonstrated directly at this point, he added, explaining that all of their experiments so far have been done by hand.

"If you want to detect 512 oligos, you need a special machine, because if you do that by hand you will introduce a bubble and kill your sample at some point," Croquette said, adding, "I think automating the system to look at more oligos is not a major difficulty."

For their proof-of-principle experiments, the team got around this issue by taking advantage of a DNA conversion method developed for nanopore sequencing. That strategy, described by Boston University's Amit Meller and colleagues in Nano Letters in 2010 (IS 5/25/2010), involves swapping out each individual nucleotide in a DNA sequence of interest for a characteristic eight-base sequence.

"We have prepared such a molecule and then sequencing is very simple," Croquette said. "You just need four oligos."

By converting a 31 nucleotide DNA fragment to a 248 base sequence in which each of the original bases was represented by eight bases, he and his co-authors showed that they could sequence this stretch of DNA by adding four nucleotides to their DNA hairpin system.

The team's sequencing-by-ligation method, meanwhile, is somewhat akin to a single-molecule version of that used in the Life Technologies' SOLiD platform, Croquette said.

Again, though, the researchers focused on measuring bead position rather than fluorescence as bases were ligated into the fragment of interest — an approach that they said shows promise for getting around the problem of desynchronization, which can happen when bases elude detection in fluorescence-based systems that follow DNA polymerization or ligation.

Using their single-molecule sequencing-by-ligation method, the researchers demonstrated that they could sequence eight consecutive bases.

Another potential application for the hairpin DNA system — and one that the researchers are hoping to commercialize in the nearer term — is for identifying and even counting sequences of interest based on the characteristic blockage patterns associated with multiple oligos.

In the Nature Methods paper, for instance, the researchers reported that they could identify three different DNA molecules ranging in size from 89 base pairs to more than 1,200 base pairs using characteristic blockage patterns for two hybridizing oligos. They also demonstrated a ligation-based approach that used neighboring oligos to find sequences of interest.

Thinking Big

Though the experiments reported at this point have all been done by hand using a small-scale research system, those involved in the study are optimistic that both their DNA identification and sequencing methods can be scaled up and automated.

Whereas the hairpin system described in the new study relied on the same 50-bead system that the researchers used for prior studies of DNA-related enzymes, for example, Croquette emphasized that 50 beads are "not sufficient, by far, to do DNA sequencing."

The team is currently working to scale up the system so it can track more beads at the same time.

Last year, a group from the Netherlands reported using 450 beads in parallel in a similar magnetic tweezer system. Croquette noted that his team is shooting for high-density arrays containing even more beads — on the order of 1,000 beads per field of view — but said more work is needed to determine the actual limit.

The group is also working on ways to systematically convert DNA libraries to hairpin DNA. The sample preparation is relatively simple in theory, Croquette said, though the yield is not yet known.

"You would have to prepare a library, which we have not done, where you take the human genome, break it into pieces, and convert it into hairpins," he said. "What we want is one crocodile end on one end and a loop at the other end … and we are working on that."

The achievable read lengths, either by hybridization or sequencing-by-ligation systems based on hairpin DNA, are still unclear as well. So far the team has shown that it can sequence 31 bases by hybridization and eight nucleotides consecutively through sequencing-by-ligation when doing manual buffer changes.

Based on the current detection limit of the system, Croquette estimates that the theoretical read length limit for most of the sequencing applications is around 1,000 bases, since fluctuations in longer molecules can create noise that interferes with the ability to accurately resolve magnetic bead position.

Researchers are already reporting reads of a thousand bases or longer on PacBio RS. That company expects to see its read lengths stretch even further — to an average of 2,500 to 3,000 bases, on average — with its new C2 chemistry and software (IS 11/1/11).

Croquette argued that if the hairpin-based system does make the leap to become a full-fledged sequencing system, it may offer an accuracy advantage over single-molecule platforms that are based on fluorescence. While fluorescence detection is less of an issue for second-generation systems that use PCR amplification, he explained, bleaching or missed bases are more problematic when looking at individual DNA molecules.

Whereas raw read error rates for Helicos and PacBio single-molecule machines have been around 4 percent and 15 percent, respectively, the French-led team reported error rates on the order of 1 percent and believes it will be possible to achieve even higher accuracy.

Nanopore-based instruments announced by Oxford Nanopore at the Advances in Genome Biology and Technology meeting last month and expected later this year are said to have read lengths as long as 100,000 bases and error rates of around 1 percent (IS 2/21/2012).

Although he is a bit more optimistic about the potential benefits of nanopore sequencing, Croquette said that there would still be situations in which looking at a single DNA molecule in a DNA hairpin would be advantageous, since it can be interrogated over and over.

Croquette and his colleagues are in the process of forming their startup PicoSeq, which aims in the long term to develop a sequencing instrument, though Croquette conceded that that the sequencing market is "very, very aggressive."

"We plan first to do identification with, we hope, a high-throughput apparatus," he explained. If that goes as planned, he said the next step would be improving the resolution of that instrument and trying to tailor it to sequencing.

Have topics you'd like to see covered in In Sequence? Contact the editor at anderson [at] genomeweb [.] com.