News FocusGenome Sequencing

Search for Pore-fection

See allHide authors and affiliations

Science  04 May 2012:
Vol. 336, Issue 6081, pp. 534-537
DOI: 10.1126/science.336.6081.534

At long last, nanopore sequencing seems poised to leave the lab, promising a new and better way to decode DNA.

Threading the pore.

With nanopore sequencing, single molecules of DNA will be deciphered as they pass through a tiny channel.


In a packed Florida conference center 3 months ago, Clive Brown introduced an audience of scientists, engineers, and biotech analysts to a device resembling an oversized thumb drive. He promised it would decipher almost a billion DNA bases in 6 hours and sell for $900. As backing for that claim, Brown described how Oxford Nanopore Technologies, where he is chief technology officer, had used a prototype to decode the genome of a virus in a single pass of a complete strand of its DNA. “There was an audible gasp from the audience,” recalls Oxford Nanopore's CEO, Gordon Sanghera.

If Oxford Nanopore's claims and promises are borne out—and some scientists remain skeptical—the company is set to achieve the first commercialization of a long-awaited and oft-doubted technology called nanopore sequencing. The technology, based on protein pores so tiny that 25,000 of them can fit on the cross section of a human hair, could be the next big thing in genome sequencing and analysis.

Although they've gotten much cheaper and smaller in recent years, machines that read DNA and RNA still usually cost hundreds of thousands of dollars, take up entire lab benches, and require much upfront and postsequencing processing to generate a genome. Nanopore sequencing could change all that. This new technology “really requires you to think about things in a completely different way,” says Elaine Mardis, co-director of the Washington University Genome Institute in St. Louis.

As the tweeters and bloggers in Brown's audience went wild, sending missives out onto the Internet, David Deamer, a biophysicist at the University of California, Santa Cruz (UCSC), and Harvard University cell biologist Daniel Branton sat in the front row, beaming. In 1996, 7 years after Deamer initially had the idea, they had publicly proposed that threading DNA through a tiny pore and monitoring changes in the current going through the pore could yield a more direct, faster way to sequence genomes. Yet until the Florida meeting, no one had claimed success in reading DNA as it moved through a pore, leaving many to wonder whether the technology would ever pan out. “Over the years, the number of people who truly believed in nanopore sequencing you could probably count on your two hands,” says Mark Akeson, a molecular biologist at UCSC. “Now both companies and academics are seeing [evidence] that this stuff actually works. This technology is going to really take off.”

Sequencing gold rush

Over the 2 decades that nanopore sequencing has lingered backstage, many other advances have greatly reduced the cost and increased the speed of reading the strings of adenines, guanines, thymines, and cytosines that compose strands of DNA. Whereas that first human genome sequence cost an estimated $1 billion to complete, the all-inclusive price at a high-throughput sequencing center today is about $18,000, and a few companies are promising costs approaching $1000 per genome. The pace has also quickened. It took 3 years at the turn of the century to produce a draft of a human genome; the same can now be done in a week. Since the human genome sequence was completed in 2003, researchers have decoded hundreds of genomes of plants, animals, cancer cells, and even ancient humans, proving that sequencing is a valuable tool for biomedicine and all sorts of other disciplines, from ecology to anthropology. Researchers are calling for 10,000 vertebrates to be sequenced, for example, and physicians may soon routinely order up a patient's genome sequence for diagnostic or preventive purposes.

Nanopore sequencing has not been part of this revolution. Instead, it was an appealing idea for which every aspect needed to be developed. When they first considered the concept, Deamer and Branton didn't have an appropriate pore or a way to control DNA's flow through such a pore, and they didn't know for sure that they could distinguish the different bases on a strand of nucleic acid. Ever so slowly, they and a handful of others have made advances on all those fronts, with several key publications in the past 2 years signaling progress, not just with protein pores but also with solid state ones (see sidebar, p. 536).

Oxford Nanopore has promised to sell its new protein-pore sequencers by the end of the year, and if those machines pan out, it could set off another genomics revolution, many scientists predict. “Current sequencing has an awful lot of complications that just go away with nanopore sequencing,” says Stuart Lindsay, a physicist at Arizona State University, Tempe.

Nanopore sequencing should require little upfront preparation beyond isolating an organism's DNA, and even that might be done away with in some applications. In contrast, current approaches require that the DNA be copied many times over and, typically, labeled with a fluorescent tag that can be read by an optical sensor. Such preparation takes time and money and erases any of the chemical modifications that result in the epigenetic control of gene expression—something researchers increasingly want to know about that nanopore devices may be able to read.

Furthermore, current sequencers work by decoding many short stretches of DNA—typically 200 bases or so—and that information has to be painstakingly pieced together. Nanopore technology can read much longer stretches of DNA: At the February meeting, Brown reported decoding a 48,000-base genome of a bacteriophage, a virus that infects bacteria, by first linking the ends of the two strands of its DNA, then threading the entire genome, first one strand and then the other, through a pore in one pass. “That really stunned the audience,” Deamer says.

While no scientist outside of Oxford Nanopore has reported seeing the prototype sequencers Brown bragged about in Florida, the company says it will eventually have an 8000-pore version—many pores will be needed to sequence genomes much larger than a phage's DNA. With 20 of these machines, it should be possible to reveal a human genome sequence in 15 minutes. “You don't have to wait 2 weeks to do the assembly; you are watching it on the fly,” Akeson says. “If [nanopore sequencing] works, there's not going to be anybody in genomics who is not using the device in some fashion.”

That's a big “if,” Mardis notes. “It's such a beautiful possibility, but there are many technical hurdles to getting it to actually produce sequence data.”

Not just an idea

Deamer began trying to jump those hurdles almost 25 years ago, long before Oxford Nanopore formed. In 1989, Deamer was working on the origins of life and was struggling to figure out how to get the molecule adenosine triphosphate (ATP) across a lipid membrane to supply energy to enzymes trapped inside his synthetic “cell.” He quickly realized that his theoretical solution—to insert a channel of some sort into the membrane—had other possibilities. If ATP could squeeze through, so might DNA. And as DNA crossed the channel, he reasoned, it would alter the ion flow through the channel. Finally, if the changes in this hypothetical channel's ionic current differed with each of DNA's bases, then that could open up a whole new way of sequencing. At the time, the idea seemed fanciful even to Deamer. For starters, he recalls, “there was no pore available.” Nevertheless, he sketched out his idea in a lab notebook.


If it works, this device could enable DNA sequencing to be done from a laptop.


Deamer also shared the scheme with Branton, and they approached Harvard about patenting it. They weren't the only ones thinking along those lines: They discovered that a colleague, geneticist George Church, had independently come up with a similar plan to sequence DNA using a pore from a bacteriophage. The three of them decided to join forces and eventually filed for a patent together. The chief missing ingredient was still a big enough pore. Church moved on to other sequencing projects, but Deamer kept an eye out for a way to make his idea reality.

Nanopore dreamers.

After David Deamer (top right) sketched out nanopore sequencing in 1989, he teamed up with Daniel Branton (bottom right).


Deamer learned about α-hemolysin, a protein that Staphylococcus aureus uses to bust open red blood cells. John Kasianowicz, a researcher at the National Institute of Standards and Technology (NIST) in Gaithersburg, Maryland, was testing pores formed by this protein as biosensors for toxic heavy metals in solution. Working with Hagan Bayley, now at the University of Oxford in the United Kingdom, he had embedded an α-hemolysin pore in a membrane and applied a voltage to produce an electrical current of potassium and chloride ions through the pore. Sensitive electronics measured the ion flow. Kasianowicz hoped the heavy metals would bind to the pore and alter the ionic current in distinctive ways.

“It occurred to me that this pore might in fact be large enough” to allow strands of RNA or DNA to move through, Deamer says. In 1993, he went to NIST with some RNA to test the concept. Because DNA and RNA are negatively charged, they would be pulled through the pore. As Deamer suspected, as the strand of RNA passed the narrow point of the pore, it interfered with the ion flow, changing the current. “We immediately got huge numbers of signals from the recorder,” indicating that the RNA was blocking the pore's ionic current as it threaded its way through, Deamer says.

In 1996, he, Kasianowicz, and Branton published a paper in the Proceedings of the National Academy of Sciences, in which they reported that they could unravel a coiled nucleic acid so that its bases move through the pore single file. They could tell the length of a strand of DNA going through the pore by the amount of time the ionic current signal was altered. In the article, they suggested that this approach could also be used to sequence DNA. “That was the pioneering paper,” says Henry White, a chemist at the University of Utah in Salt Lake City.

Even with a potential pore in hand, Deamer and his colleagues realized that DNA's bases were whipping through the channel too fast to be identified. One solution was to harness another protein to latch onto the DNA and control its movement through the pore. Reza Ghadiri of the Scripps Research Institute in San Diego, California, was also interested in nanopore sequencing and had taken the first steps toward controlling DNA movement using a polymerase, an enzyme that copies DNA by ratcheting a DNA strand along base by base, like a sprocket moving the links of a bicycle chain, as it adds the complementary base. Independently, Akeson, working with Deamer, started buying and testing various polymerases from different species and other proteins. The first ones he and his colleagues tested quickly fell off the DNA, only briefly moving the strand through the pore. After many years of trying, in 2010, they discovered that a polymerase from a phage called ϕ29 would move long stretches of DNA, one base at a time, at a reasonable pace through α-hemolysin.

Yet although the ϕ29 polymerase slowed the DNA down as desired, the stem of the mushroom-shaped pore was so long that more than a dozen bases were passing through at any one time, creating a fuzzy ionic current signal at best. The signals weren't distinctive enough to tell one base from another. They needed a different pore.

A better pore

Fortunately, Jens Gundlach, a gravitational physicist at the University of Washington, Seattle, had heard about nanopore sequencing and was intrigued enough to move into biophysics. In 2003, Gundlach started looking into alternatives to α-hemolysin. A literature search yielded no promising candidates, but then he saw in Science a pore in a different bacterium with a potentially better geometry—it was shaped like a funnel—for getting a strong ionic current signal. Called MspA (for Mycobacterium smegmatis porin A), this channel has a single narrow section long enough for just four bases. The natural MspA had limitations, however: The constricted section carries a negative charge, making it hard for the similarly charged DNA to get through.

Gundlach and his colleagues tweaked MspA's gene, changing the protein so that the constricted part of the pore was neutral, and produced the modified pore by expressing the altered gene in bacteria. They had also added some positive charges at the pore entrance to enhance the inflow of DNA. When DNA was suspended in the modified pore, the signal for each base was almost 10 times stronger than the signal for immobilized bases in α-hemolysin, Gundlach's team reported in 2010. A sequencer using this unnatural MspA in theory “could resolve in much finer detail the DNA strands,” Akeson says.

Perfecting pores.

Cross sections of the MspA (left) and α-hemolysin (right) pores show their different geometries.


But each base still zipped by in a microsecond, 1000 times faster than could be read. And the only way Gundlach could slow them down required modifying the DNA itself, an impractical solution. So last year, he and Akeson joined forces. “The ϕ29 polymerase provides a mechanism to move the DNA through the pore at a reasonable speed,” about one base every 30 milliseconds, Gundlach says. With the combination of Gundlach's pore and Akeson's polymerase, nanopore sequencing finally made its public debut, at least in the academic sense. Gundlach and his colleagues reported at the Florida meeting and online 25 March in Nature Biotechnology that they had could distinguish the bases in six DNA strands ranging from 42 to 53 bases long. “This is the first paper where somebody has actually [read] DNA,” says chemist Geoffrey Barrall, president of Electronic BioSciences in San Diego, California, which is also developing nanopore sequencing technology. (Oxford Nanopore has yet to publish a scientific paper on the phage genome sequencing Brown described in Florida.) Gundlach says he has since tested longer stretches of DNA.

A company is born

While Deamer and the other U.S. researchers were struggling to make nanopore sequencing a reality, Bayley was modifying the α-hemolysin pore with a different primary goal in mind: sensing devices. He had started looking at α-hemolysin in the 1980s to learn how water-soluble proteins made it through membranes, but he got interested in engineering pore proteins for biotechnology. Bayley envisioned pores that would help kill tumor cells or detect metals, sugars, and other proteins, and he had been modifying this pore for these different applications, making much progress. In 2005, he started a company to commercialize these biosensors.

About the same time, the push for the $1000 genome (Science, 17 March 2006, p. 1544) had resulted in a new U.S. National Human Genome Research Institute (NHGRI) program for technology development. Bayley decided to apply and see what his modified α-hemolysins could do with respect to sensing DNA. Since he knew that he could make pores that could distinguish mirror versions of the same molecule, he was confident he could distinguish DNA's bases. With Ghadiri, he got an NHGRI grant and eventually published that the pore could tell DNA's building blocks apart when they were in solution. He decided to pursue the idea of feeding individual bases through the pore and started looking into using an enzyme that would break off each base as the DNA entered the pore. In 2008, his company, now renamed Oxford Nanopore Technologies, stepped up its efforts in nanopore sequencing, first pursuing the idea of reading cut-up bases and later following the path others had taken, decoding long, intact DNA strands.

Oxford Nanopore went after a better pore in earnest, developing a high-throughput approach toward testing and modifying potential protein candidates. The company licensed technology developed and patented by Bayley, Deamer, Branton, Akeson, and others. Because the natural lipid bilayers of the cell membrane originally used to hold the pores are not very stable, the company developed a polymer alternative that could withstand exposure to blood or pollutants. And Oxford Nanopore has its own proprietary motor protein to control the DNA's flow through the pore. The company won't disclose any details yet but says it will have data and machines for academics to evaluate in the coming months.

One challenge, the company acknowledged in Florida, is getting the error rate down from its current 4%. Academics concur that errors are a problem. As a polymerase ratchets along, it sometimes backtracks so that a base is read twice; other times, the base gets through the pore without being read. One can compensate for these random errors by sequencing each DNA strand multiple times. With the pore setup developed by Akeson and Gundlach, they can read the DNA as it is first pulled down through the pore and then again as it is pulled up and turned into double-stranded DNA by the polymerase. In theory, one could repeat those two steps with the same strand as many times as needed. With its sequenced viral genome, Oxford Nanopore showed it could tackle the problem by connecting the DNA's two complementary strands so that each is sequenced, the second providing an accuracy check on the first.

Neither approach solves another problem, accurately reading long stretches in which the same base is repeated, a not-infrequent occurrence in genomes. But nanopore proponents point out that this repetitive DNA is difficult for all sequencing techniques.

How solvable this and other problems are and whether they can be overcome in the coming months is not yet clear. Oxford Nanopore has a good reputation, but some rival sequencing companies and genome experts won't believe the sequencers work as advertised until they can test one. They caution that nanopore technology has been “coming” for so long that it's hard to believe the hurdles are finally overcome. However, “if they could pull it off,” Mardis says, “it would be a complete game changer.” And Deamer and Branton's smiles may get even wider.

View Abstract

Stay Connected to Science

Navigate This Article