A recent simulation study in the Journal of Chemical Physics is pointing to system tweaks that may be needed to improve error rates and stretch out read lengths in proposed nanopore-based sequencing schemes that use an exonuclease enzyme to methodically lop off individual nucleotides before each of these DNA bases enters the pore.
Researchers from the National Institute of Standards and Technology, the National Institutes of Health, Virginia Commonwealth University , and Wheaton College relied on a combination of computer simulations and analytical theory to delve into the details of exonuclease/nanopore sequencing, looking in particular at ways in which diffusion influences individual base capture by the pore.
Their models also consider interactions within the pore — particularly between the liberated mononucleotides and adapter molecules that help in identifying bases within the pore.
"[W]e study the critical parts of this system's performance, determine the probability that a given length polynucleotide can be accurately sequenced with the technology, and suggest how the method might be implemented," corresponding author John Kasianowicz, a physical scientist with NIST's CMOS Reliability and Advanced Devices group, and his co-authors wrote.
Based on the exonuclease-based nanopore data reported in the past, the group concluded that while the idea of combining exonuclease enzymes with nanopores "has significant merit" as a sequencing strategy, "several aspects of the technique need improvement."
In particular, results of the new analysis suggest that researchers need to find ways to get more cleaved nucleotides to enter a nanopore, along with somewhat better binding between bases and molecular adapters in the pore. In the absence of such advances, authors of the study argued, too many bases will be missed or identified incorrectly, leading to short read lengths and/or a need for extensive parallelism.
"The available data and analytical results suggest that the proposed method will be limited to reading [fewer than] 80 bases," they wrote, "imposed, in part, by the short lifetime each nucleotide spends in the vicinity or the detection element within the pore and the ability to accurately discriminate between the four mononucleotides."
Still, the team stopped short of ruling out the possibility of developing exonuclease/nanopore-based DNA sequencing systems. Rather, they argued that the type of analyses done for the current study could prove useful for guiding future developments within the exonuclease/nanopore arena.
"The bottom line is [that] the ability to discriminate between the four bases is pretty good — maybe not good enough for sequencing yet, but I think maybe if you worked hard you could fix that," Kasianowicz told In Sequence.
A fairly wide range of approaches have been described by those interested in harnessing nanopore properties and using them to produce new DNA sequencing systems.
Whereas some have been pursuing solid-state nanopores, others have been working to exploit and improve on protein pores already found in nature, such as those produced by the Staphylococcus aureus protein alpha-hemolysin or a Mycobacterium smegmatis protein called MspA.
Likewise, there have been various ideas about to best use the pores for sequentially reading DNA bases.
Most of the methods discussed so far involve electrical detection of nucleotides based on current changes that occur in a system as each of these bases moves through the pore. Strand sequencing approaches have focused on ways of following these changes as DNA strands get threaded through the pore, whereas the method scrutinized in the current study involves enzymatically nipping each base off of the DNA strand with an exonuclease enzyme before the mononucleotides enter the pore.
In this exonuclease/nanopore system, Kasianowicz said, "you're literally slicing off one base at a time and hopefully retrieving every one of those sliced mononucleotides, reading them, and not missing them to make the proper identification."
"Of course, the devil's in the details," he added.
A desire to understand those details and their implications for sequencing prompted the new analysis, which considered issues related to doing exonuclease/nanopore sequencing with nanopores based on alpha-hemolysin pores.
In the past, researchers affiliated with the UK-based sequencing company Oxford Nanopore Technologies have presented research related to both exonuclease and strand-sequencing approaches based around alpha-hemolysin nanopores, though the commercial GridIon and MinIon instruments that the firm announced last year are strand sequencing systems (IS 2/21/2012).
Oxford Nanopore declined to comment on the new exonuclease/nanopore paper or whether its conclusions have implications for any of the nanopore sequencing strategies the company is currently exploring.
For their analyses, Kasianowicz and his colleagues brought together data generated through past exonuclease/nanopore experiments described by themselves and others, including information presented by researchers from Oxford Nanopore and the University of Oxford in a 2009 study in Nature Nanotechnology (IS 2/24/2009).
Based on their simulations and analytical models, he and his colleagues determined that some bases likely escape detection in the exonuclease/alpha-hemolysin pore system owing to diffusion of cleaved bases away from rather than into the pore.
Their results suggest that the probability of each base being captured is currently around 80 percent. Consequently, Kasianowicz said, the probability of correctly reading multiple bases consecutively goes down dramatically as sequences get longer.
His group's analyses suggest that the maximum read lengths possible using the exonuclease/nanopore strategy as proposed currently is less than 80 bases, while the chance of accurately reading a 100-mer is "vanishingly small" when just 80 percent of cleaved bases get captured.
At least some of these lost bases could theoretically be captured by subtly altering the positions of the exonuclease and pore proteins relative to one another, the researchers predicted.
While diffusion does seem to influence exonuclease/nanopore-related base detection, though, the team's models don't support the notion that there are molecular forces slapping bases away from the pore before they can enter, as some had suspected.
Instead, the analysis suggests that a lower-than-anticipated detection rate described in exonuclease/nanopore systems so far is primarily related to the kinetics of the interaction between cleaved DNA bases and the beta-cyclodextrin adapter used to nab and help distinguish between mononucleotides inside the pore.
A 2009 study by the University of Oxford's Hagan Bayley and his colleagues demonstrated that it was possible to distinguish between all four DNA bases using a mutant alpha-hemolysin nanopore outfitted with beta-cyclodextrin.
"It turns out that if you have a weak binding constant, that means that [the mononucleotides] don't spend much time on the site," Kasianowicz said. "So they bind, but the kinetics don't work out right."
"If it stays bound for a unit of time and our detector can't see that fast, then you're not going to see the event," he explained. "And we think that's what was going on."
The new analyses have implications related to the amount of voltage that can be applied to exonuclease/nanopore systems in their current form, too.
While increasing the voltage applied to the system is expected to nudge more cleaved bases into the pore, the researchers reported, that would also abbreviate the amount of time each mononucleotide spends interacting with the beta-cyclodextrin in the pore, making it trickier to get stable, measurable interactions.
"You gain from the voltage on the outside, but you pay the price on the inside," Kasianowicz said.
"On the one hand, you want to go as fast as possible to read a billion bases quickly," he added. "On the other hand, you need to go slow enough to read them reliably."
Issues related to DNA translocation speed through the pore have come up in nanopore sequencing studies that involve intact DNA strands, too. There, a variety of approaches have been proposed for slowing the DNA molecule's movement through the pore — from polymerase-based ratcheting systems (see IS 9/28/2010, IS 2/21/2012, IS 3/27/2012) or magnetic tweezers (IS 4/21/2009) to systems with altered viscosity, salt concentrations (IS 1/24/2012) or temperature gradients (IS 12/11/2012).
In the case of the exonuclease/nanopore system, predictions from the new study suggest very careful voltage selection will be needed to get bases into the pore without lowering the detection rate — unless researchers come up with ways to get more bases into the pore without ramping up voltage, enhance interactions between bases and molecular adapters in the pore, or both.
"If you can draw the active site closer to the pore, you increase the chance of it going in," Kasianowicz said. "And if you can simultaneously increase the residence time of the mononucleotides on the molecular adapter, that will allow you to use more voltage on the outside."
To that end, Kasianowicz and his colleagues have applied for funding for research aimed at finding alternative adapter molecules that might hold DNA bases within the pore somewhat longer.
The team is also pursuing its own nanopore sequencing strategies in parallel. Along with exonuclease/nanopore-based approaches, the group is collaborating with researchers at Columbia University and nanopore sequencing startup Genia Technologies to chase an alternative method.
The latter approach relies on a polymerase enzyme to incorporate tagged nucleotides into a strand complementary to the DNA strand of interest.
The idea there is to design tags with properties that are easily taken up by the pore and produce non-overlapping signals, so that DNA sequences can be determined using an enzyme that chops off the tags at the edge of the nanopore.
"If you make the appropriate sized tag, you may be able to cleave and read [the tag] immediately and, therefore, increase your chances of capturing what you want to read," Kasianowicz explained.
The group has a ways to go to show that that theoretical approach is possible, he added, since that tag-based strategy has not yet been tested experimentally.