A team of scientists on both coasts of the United States is getting closer to developing nanopores that can identify haplotypes one molecule at a time.
The research, which is funded by the National Institutes of Health and DARPA, and is part of an R&D collaboration with Agilent Laboratories, may significantly reduce the cost of haplotyping, according to experts.
“We want high-speed direct sequencing of DNA that can be applied to haplotyping or genotyping that is relatively reagent-free, and [doesn’t rely on] the complexities of labeling chemistries,” said Carl Myerholtz, Agilent’s manager of molecular systems. “We’ve been happy with what we’ve seen.”
Specifically, Myerholtz has been happy with research performed by Daniel Branton, professor of molecular and cellular biology at Harvard University, and colleagues at the University of California, Santa Cruz. One goal of their research is to be able to sequence a long strand of DNA by sending it through a 10-9-meter hole at 1 billion bases per second.
While that ability is still a number of years away, Branton said the technology can have a strong impact in haplotyping. He has already designed solid-state nanopores big enough to thread a DNA molecule through, but small enough to send the bases by only one at a time. The nanopores are constructed by directing ion beams at solid-state membranes made of substances such as silicon nitride, in order to narrow existing pores through lateral transport, or create a pore from a cavity through sputter erosion.
“DNA is a long, floppy, molecule,” Branton told attendees of a genomics conference earlier this year. “But as it goes through the nanopore it has to proceed in single file order. We asked ourselves what kind of probe we could apply to detect the differences in bases.”
Branton initially thought he could measure differences in bases by sending a beam of ions across the pore. While this worked to distinguish between groups of bases, the size of the nanopore limited the number of ions that could be sent across the pore. This, in turn, limited the requisite sensitivity for measuring distinctions down to the individual base.
“We could either slow down the DNA molecule to count the ions, or use an alternative probe,” Branton said. “Our preference right now is to do it with tunneling,” which would send enough electrons across the pore to distinguish differences down to single base pairs while maintaining the high-throughput rate of one base per millisecond. The technique also allowed the sequencing of DNA strands without the need for amplification, which is another costly component to haplotyping and genotyping.
Though his arsenal of electron-tunneling nanopore sequencers in parallel may be able to sequence an entire human genome in three hours, Branton thought a more realistic application would include sequencing specific polymorphic regions of human chromosomes.
About three years ago, Branton got a call from Agilent Laboratories, a unit of Agilent Technologies, which said it had an interest in helping him develop his nanopore technology. Today, Branton co-leads a 15-person team with Harvard physicist Gene Golovchenko.
“We have until now been doing a lot of work with a prototype pore, which is a protein that inserts itself into a lipid bilayer,” Branton told SNPtech Reporter in a telephone conversation this week. “We’ve learned a lot about how DNA moves in pores, and how to derive information from the traversal of the DNA through such pores. We’re now beginning to apply this to solid-state pores whose fabrication methods we recently developed. With Agilent’s help, we’re now beginning to make those nanopores on a routine basis. We’re still a few years away.”
For its part, Agilent is bringing to the table its strengths in automation and reproducibility. Branton’s lab has “done an excellent job in pioneering synthetic pores,” but since the pores are currently made one at a time by a single instrument developed by physicist Golovchenko, “we’re going to bring our engineering skills to make it more robust,” said Agilent’s Myerholtz.
Myerholtz said this open-ended collaboration will investigate whether it is worth investing additional time or money into the platform — indeed, this partnership “is one of the longest-term research programs” underway at Agilent. The company also expects to employ some of its microfluidics technologies as well as its experiences in DNA chemistries.
“There are so many problems to overcome on the physics and electronics because we’re working on dimensions that are orders of magnitude smaller than current IC technology,” Myerholtz said. “But I’ve seen lots of promising early results.”
To be sure, “there’s many different possible applications” for the technology, said Branton. “There are very few methods for high-speed haplotyping that are available because it requires knowing particular sequences in areas of SNPs that may be far removed from each other on a single chromosome.”
In fact, research that seeks to read individual molecules has created a kind of burgeoning cottage industry with the likes of Solexa and US Genomics. The nascent science has also attracted big genomic names like Craig Venter, who is a fan of single-molecule methodologies like US Genomics’ GeneEngine, and who last summer was appointed to the company’s scientific advisory board.
“Current methods, like array technologies, can give you information [about] which SNPs are in the genome, but they make it hard to know what the haplotype is,” Branton said. “Nanopore is just … excellent for high-speed haplotyping, which has much greater predictive value to help researchers predict outcomes for certain medications, or anticipating treatment outcomes, and simply knowing SNPs.”
One of the first “practical applications” of the technology will be in research labs, Branton explained. In fact, when SNPtech Reporter spoke with Branton, he was in the process of submitting for publication a paper that details how nanopore characterization can be used to detect impurities, chemical degradation, or states of phosphorylation on DNA “in a very extraordinarily long range of different molecular sizes,” he said. “That kind of research use will probably come into being in a year or two. Haplotyping, probably three or four years.”
“This approach is extremely promising — but also some ways in the future,” said David Altschuler, director of the medical and population genetics department at the Whitehead Institute. “We are doing … these [haplotype] analyses on a macroscopic scale when in fact we’re reading molecules. And if there were some way to read individual molecules, you can certainly make the speed and cost arbitrarily low,” said Altschuler, who also represents the United States in the International HapMap Project. “I’m very optimistic in the long run.”
However, independant concerns exist. “There is a physical reason to assume you can never get all the data, so it’s always going to be somewhat statistical,” said a genomics insider familiar with the technology. “This is its Achilles’ heel … and it is a fundamental issues that I worry about.
“When you’re doing single-molecule detection, it’s impossible to achieve perfect efficiency; if you have a unique quantum event that happens on a single molecule, there’s physical constraints that limit the ability to detect it,” said the expert, who asked not to be named. “You’ll never detect that 100 percent of the time. If you’re trying to haplotype or sequence and you’re trying to read many bits of information off each molecule, ... it would be impossible to get to the point that you can get all bits of information off all molecules.
“To my mind, this problem has not been solved with single-molecule methods yet — even though it is a very attractive approach,” he said.