This article has been updated to clarify that John Nelson referred to the processivity of specific exonucleases, not all exonucleases.
Oxford Nanopore Technologies said this week that the firm's detector technology possesses sufficient accuracy and reliability to become part of a commercial DNA sequencer, after company researchers and their partners at the University of Oxford showed they can reliably identify five different DNA building blocks, including methylated cytosine, as they pass through a protein nanopore.
The research, conducted last summer and published online on Sunday by the company and its academic collaborators in Nature Nanotechnology, represents another step in Oxford Nanopore's development of a label-free single-molecule sequencer.
The company received an $18 million equity investment from Illumina last month (see In Sequence 1/13/2009) and appears to have successfully engineered a protein nanopore into a reliable sensor for DNA bases. However, it still needs to couple this detector with an exonuclease that can work under the same experimental conditions in order to generate a complete DNA-sequencing system.
The Nature Nanotechnology paper is noteworthy because it is the first to show that researchers can detect unlabeled single DNA bases "to a confidence level that is appropriate for a highly competitive commercial sequencing system," Oxford Nanopore CEO Gordon Sanghera told In Sequence by e-mail last week.
He added that while the work described in the article was completed last summer, it is "still highly significant" because it demonstrates the detector works and because the company showed it was able to distinguish at least some of the bases in combination under conditions that are compatible with an exonuclease.
"This is impressive work — particularly the fact that the nanopore not only can distinguish sequence at high accuracy, but that it is also able to distinguish methylated cytosine," Andre Marziali, an associate professor of biophysics and director of engineering physics at the University of British Columbia who has been working on using nanopores for genotyping, told In Sequence by e-mail.
"This is exceptionally promising, but it will remain to be seen whether the combination of read length, read rate, and parallelization is competitive with other single-molecule [sequencing] technologies," he said. He also pointed out that the error rate resulting from nucleotides failing to enter the nanopore detector is not clear yet from this work.
"I think it's a really nice piece of work," said John Nelson, a principal scientist at GE Global Research. "They definitely demonstrated that they have a cool little nanopore in a patch clamp that can read something like 35 bases a second" with single-base resolution.
However, he pointed out that under the experimental conditions that allowed the researchers to resolve all five DNA bases — including methylated cytosine, the so-called fifth base — most exonucleases do not perform well, and it will be "a significant challenge" to either make the enzyme work at higher salt concentrations, or to achieve better base resolution at lower salt concentrations.
Since generating the data for the paper, Oxford Nanopore has developed the technology further but is not yet ready to comment on the anticipated error rate, read length, or degree of multiplexing of its commercial system. According to a company spokesperson, read lengths will be "long" and the company will talk more about scaling up its nanopore chip — through in-house development and an external collaboration — "in due course."
"As you can imagine, we have been working on improving [the] performance [of the nanopore sensor] even further, as well as progressing its integration into our overall system, including the exonuclease and arrayed chip," Sanghera said.
[ pagebreak ]
He did not elaborate on how far along the company is, but said that almost 60 company employees are working on the nanopore and enzyme chemistry, chip and instrument engineering, and informatics elements.
Also, the company has "significant collaborations" with several undisclosed external partners underway to "complement" its in-house R&D, he said, adding that "you will hear more in due course" about these partnerships as well as development timelines.
So far, Oxford Nanopore has not disclosed any commercial partnerships other than its strategic alliance with Illumina. However, the company has been collaborating with Philips Research and a number of academic partners as part of an EU-funded research consortium (see In Sequence 12/9/2008).
The Nature Nanotechnology paper builds on prior research by Hagan Bayley, the company's academic founder, that was published in 2006, in which the researchers showed that they were able to use alpha-hemolysin protein nanopores with a small adapter molecule inside the pore to distinguish in principle the four nucleoside monophosphates, or dNMPs, that make up DNA. According to the company, "at that time, the accuracy was not high enough to compete with existing sequencing technologies."
One of the problems with the previous approach was that the cyclodextrin adapter was not attached to the nanopore covalently, so there were periods during which the adapter was absent and bases could not be observed continuously. The researchers have now linked the adapter covalently, allowing them to detect 98 percent of all dNMPs passing through the pore at a data acquisition rate of 20,000 Hz.
Through mutagenesis of the protein pore, they have also optimized the position of the adapter within the barrel of the pore, which has improved their ability to discriminate the different bases. The percentage of binding events that could be assigned to each base with near-100 percent confidence ranged between 97.9 percent and 99.99 percent, depending on the dNMP and the salt concentration, either 800 mM or 400 mM.
In addition, the scientists were able to distinguish, at 400 mM salt, methylated dCMP from the other four bases, a potentially important capability for epigenomic studies, since up to 5 percent of cytosines in mammalian genomes are methylated.
They also found that at a voltage where the bases can be discriminated well, a "very high" proportion of dNMPs translocates to the other side of the nanopore after binding to it, decreasing the chance that the same dNMP will be recorded twice.
Importantly, the scientists were also able to distinguish three DNA bases at a time at lower salt concentrations that are compatible with an exonuclease: At 200 mM salt on one side of the nanopore that contained E. coli exonuclease I, and 500 mM on the other side, they were able to record dNMPs released from oligonucleotides made up either of G, A, and C; or of G, T, and C.
The authors note in their paper that unlike most other enzymes, exonuclease I can work under high salt conditions as well, "which may allow data acquisition in high salt for optimal base discrimination." They did not show data supporting this. In addition, the enzyme's salt tolerance, processivity, digestion rate, and stability could be improved by enzyme engineering, they write.
They also outline in their article how an exonuclease might be coupled to the nanopore detector in order to produce a complete sequencing system. To bring the enzyme close enough to the nanopore, it could be attached either chemically or through genetic fusion. Further, the active site of the exonuclease must be aligned with the entrance to the nanopore, so the detector does not miss any dNMPs.
Also, the exonuclease might need to be slowed down — either by mutating it or by varying the operating temperature or other conditions — so there is sufficient time for a base to be recorded before the next base enters the nanopore. Certain exonucleases, the researchers add, are highly processive, allowing for long read lengths of a sequencing system.
According to GE's Nelson, exonuclease I "is not known for being a processive enzyme," and neither are other commercially available types of the enzyme that work at the salt concentrations used by the Oxford researchers in their paper. But it might be possible to make the enzyme more processive by, for example, fusing it to another protein that binds to single-stranded DNA. He also said it would be "perfectly reasonable" to slow the enzyme down by mutating it.
"I don't think that any of it is unachievable," he said. "It's just going to take work to find the conditions that give them the resolution of all five bases in a condition that allows the enzyme to work."