In a step toward visualizing the sequence of long DNA molecules directly under an electron microscope, ZS Genetics has demonstrated that it can identify single labeled bases in DNA using annular dark-field scanning transmission electron microscopy.
For the proof-of-concept study, appearing online in Microscopy and Microanalysis last week, the researchers labeled one of the four bases of DNA, showing that EM-based DNA sequencing might be achievable in principle, and ZS Genetics remains committed to the goal of developing the technology commercially.
But the North Reading, Mass.-based company faces steep competition from Oxford Nanopore's nanopore strand sequencing, which also promises to read off the sequence of long DNA molecules directly using equipment that is less costly than an electron microscope. In the meantime, another ZS Genetics competitor, Halcyon Molecular, appears to have given up on EM-based sequencing, according to former collaborators and industry experts.
ZS Genetics has been working on EM-based sequencing since it was founded in 2005 (IS 3/22/2007) and has raised $5 million from private investors. The firm is currently seeking additional funding to develop and commercialize its technology.
Originally, the firm had planned to generate its first DNA sequence data in 2008 (IS 5/27/2008), but both the recession, which hampered fundraising efforts, and the intricacies of the technology have slowed it down, according to William Glover, the firm's founder, CEO, and vice president of research and development.
"When you work with nanotechnology, individual molecules, the details are really important and not obvious when you start out," he said. "It's really been the details that we've been working on."
Glover acknowledged that the path to DNA sequencing is still long, but his team's proof-of-concept work is showing the way. "We now know how the pieces fit together, we know the approach," he said. "That's not the same thing as having finished the hard work. There is a lot more to be done."
For its published study, ZS Genetics collaborated with researchers at the School of Engineering and Applied Sciences and the Center for Nanoscale Systems at Harvard University and with scientists at the Hubbard Center for Genome Studies at the University of New Hampshire.
In order to increase the contrast of DNA, so it can be seen under an ADF-STEM, the scientists labeled all the thymine positions in the molecule by replacing them in a PCR reaction with a dUTP nucleotide that has a single mercury atom attached to its base.
They used two types of DNA: single-stranded M13, a 7.2-kilobase viral genome, and a 3.3-kilobase synthetic DNA with a repeating pattern of Ts, a pair of thymines separated by one other base that repeats every 12 base pairs.
Following the labeling reaction, they mounted the DNA on a thin supporting substrate using a method that separates, linearizes, and may partially stretch the molecules.
After vacuum-drying the samples, they performed ADF-STEM imaging, using an aberration-corrected Zeiss STEM.
In M13 DNA, individual heavy atoms were "clearly visible", according to the researchers, and detection events were "mostly distinguishable from background fluctuations, but not perfectly."
For the synthetic DNA, within a 180-base pair segment, they were able to see 17 out of 30 predicted labels, spaced apart as expected by the known pattern of Ts. One reason labels were likely missing is because of a heating step prior to imaging.
The researchers concluded that in order to use the approach for sequencing DNA, they need to reduce the loss of label, be able to identify additional base types, and increase the fraction of bases labeled within each molecule.
According to Glover, one of the next steps will be to label all four bases – either individually using the same type of label, or simultaneously using different labels. He said the company has made "good but not complete progress" on this and will reveal its results in a scientific publication.
In addition, he and his colleagues "have been able to figure out how to make this fast enough so it's practical rather than a mere curiosity," details of which they also plan to publish. Right now, Glover said, the process is manual and "quite slow," but their goal is to automate it, so it will take approximately two days in total.
One potential challenge, which the researchers have not quantified yet, is the local elongation of DNA attached to the substrate, and how much it varies. "The sequence data comes from both the identification and the physical position [of the bases], and if local elongation were unpredictable, then it would be very challenging to determine local sequence," Glover said.
Several experts agreed that challenges still lie ahead for ZS Genetics as it looks to develop its technology. While the resolution and sensitivity of electron microscopy is good enough to visualize individual DNA molecules and bases, labeling and sample prep techniques are trailing behind, according to Stephen Pennycook, who heads the scanning transmission electron microscopy lab at Oak Ridge National Laboratory in Tennessee. "I think the microscopy side to it is a solved problem, but the chemistry side of it clearly is not a solved problem at this stage," he said.
Tackling these problems includes "making sure you have everything labeled accurately and then enough different labels on different base pairs that you can put all the data together," he added.
Pennycook said that the cost of a high-end STEM is currently high − on the order of several million dollars – but he said it could come down by a factor of 10 or so if an instrument was designed specifically for DNA sequencing, and if demand for it was strong.
Scan time should also not be a bottleneck, he said, because once the DNA is laid out on the sample reliably, it could be scanned rapidly and automatically. "But getting the DNA stretched out and labeled reliably, I think that's where the tricky bit is," he said.
Pennycook said he used to collaborate with Halcyon Molecular, another startup company, based in California, that had been pursuing EM-based DNA sequencing.
He and several DNA sequencing experts told In Sequence that they had heard the company stopped developing the technology earlier this year, following Oxford Nanopore's announcement that it was commercializing strand nanopore sequencing.
At the time, ONP said it had sequenced the 48-kilobase phage lambda genome as a single read (IS 2/20/2012), but the firm has not made those data public yet. It also said that one of the systems it is developing, a disposable device called MinIon, would sell for less than $1,000.
"Strand sequencing is now there, thanks to Oxford Nanopore," said Stuart Lindsay, director of the Center for Single Molecule Biophysics at Arizona State University's Biodesign Institute, whose lab is developing an electron-tunneling-based nanopore reader for DNA sequencing. And while ONP has not published its results in a peer-reviewed journal yet, he said a paper by academic researchers published earlier this year (IS 3/27/2012) shows strand sequencing works in principle.
ZS Genetics and its colleagues, Lindsay said, "have shown some correlation between positions of bases in the sequence, with the incorporation of modified Us, and it looks good, there is some electronic signal, but it's not sequencing, it's one base."
"Halcyon has been there and done that, and Halcyon has completely dropped within weeks of Oxford Nanopore making their announcement," he said.
Halcyon executives, scientific advisors, and investors did not return requests for comment on the status of the company.
ZS Genetics, in the meantime, remains undaunted. "Our objective is that we would be equal to or better than all of the second-generation [sequencing] platforms in each of the major categories of quality, overall cost, and speed," Glover said. "That's where we hope to get to, but we have a lot of work to do to get there."