By Monica Heger
This article was originally published March 26.
Overcoming two major hurdles that have plagued researchers developing nanopore sequencing devices, a team from the University of Washington has demonstrated both single-base resolution and control of DNA movement through the pore.
Building on previous work, the University of Washington team, led by Jens Gundlach, has coupled the Mycobacterium smegmatis porin A, or MspA, pore with an automated system for ratcheting DNA through a nanopore to read known DNA sequences between 42 and 53 bases long.
Gundlach previously demonstrated that MspA would be an ideal pore for nanopore sequencing due to its size and shape (IS 8/24/2010). In this most recent study, published this week in Nature Biotechnology, he combined the MspA pore with a method for controlling DNA movement developed recently by Mark Akeson's lab at the University of California, Santa Cruz.
Akeson's method involves incorporating a "blocking oligomer" onto the template DNA strand, which must first be unzipped as the DNA moves through the pore, after which synthesis is initiated and the DNA moves back through the pore in the opposite direction (IS 2/21/2012).
The study marks the first demonstration "that ion current levels caused by single-nucleotide movements of (basically) unmodified DNA can be read," Gundlach told In Sequence.
Oxford Nanopore Technologies, which is planning on launching a nanopore sequencing system this year, has not released details of the nanopore it is using or the method it uses to control translocation (IS 2/21/2012).
The company told In Sequence this week that its in-house chemistries and methods differ from those that have been published by academic groups, but declined to comment further.
When combined with the MspA pore, Akeson's ratcheting method "worked pretty much straight out of the box," Gundlach said. "There were a few technical issues we had to learn about the phi29 [DNA polymerase, which controls the ratcheting through the pore], but once those were resolved, it worked like a charm," he said.
The two labs have been collaborating for about a year. Both groups are supported by the National Human Genome Research Institute's $1,000 Genome grant program.
The next step, Gundlach said, is to refine the base-calling algorithms and to make additional improvements and optimizations that would enable the development of a commercial device. He said that he has not yet licensed the technology to a commercial entity.
"What we have here is the innermost core of a new sequencing [device]," he said. While he said that the study demonstrates that "it is going to work," commercialization will require the researchers to "improve it in every way."
To make the system, an MspA pore was established in a lipid bilayer. As described by Akeson's group, a template DNA strand was annealed to a 23-nucleotide primer, followed by either a 15-nucleotide or 29-nucleotide blocking oligomer with a 3' end consisting of seven abasic sites and a three-carbon spacer. The abasic sites serve to demarcate the completion of the "unzipping" step.
One of the advantages of the MspA pore is the size of its constriction site, said Gundlach. It is small enough that only one base at a time can fit inside the site, which helps reduce noise and increase resolution as compared to the commonly used alpha-hemolysin protein pore. The smaller size also maximizes the current flowing through the pore, he said.
Bases are identified by the current level generated from the base in the constriction site of the pore. Different bases generate different current levels.
Even though only one base is in the constriction site at a time, surrounding bases impact the current, said Gundlach. The base coming into the constriction site and the base leaving the constriction site have the biggest impact, while the bases immediately adjacent to those play a minor role.
Gundlach said that in order to read the bases, his team designed an algorithm that essentially measures the difference between the current generated from the base moving in and the base moving out of the constriction site, he said.
As described in the current Nature Biotech paper, the researchers tested a "block homopolymer" DNA template, which contained all four bases in short homopolymer sections. Before adding the necessary nucleotides for synthesis, they ran the DNA through the pore and recorded the current patterns that would be consistent with unzipping the blocking oligomer.
As expected, the abasic sites caused a spike in the current as they passed through the pore. These were included to serve as a marker of the end of the unzipping step. The team confirmed that when translocation was complete, the phi29 enzyme was unable to initiate synthesis, since there were no nucleotides available.
Next, the team performed five separate experiments using block homopolymer templates between 80 and 91 nucleotides in length, and included all the necessary nucleotides to enable synthesis. They observed that different current levels corresponded to the different bases, and also that the current pattern generated from unzipping the oligomer was identical to the current pattern generated from synthesis.
Additionally, the time it took for the DNA to pass through a pore as the blocking oligomer was being unzipped was significantly longer than the time it took for it to pass through in the opposite direction, since it is being forced "against its natural direction," said Gundlach.
In the case of unzipping, the DNA moves at a rate of about 1 nucleotide every 100 milliseconds. Going in the other direction, it moves at a rate of about 1 nucleotide every 30 milliseconds.
Current levels lasted on average for about 28 milliseconds, and the current levels showed differences of up to 40 picoamperes.
Gundlach said that was a good speed and resulted in "nicely resolvable current levels." As the system is optimized, though, he expects that it could be made to run even faster.
The team next ran a series of experiments testing DNA templates with known sequences, including a sequence that contained a tri-nucleotide repeat of cytosine, adenine, and thymine, except for one triplet where a thymine was substituted for guanine. The system was able to detect the substitution.
Additionally, the team tested the system on four different heterogeneous DNA templates, where the nucleotides were not in any specific pattern.
The team demonstrated that the system could successfully read the random DNA templates. "For each experimental sequence, we obtained consistent current patterns with distinct features corresponding to individual nucleotide steps as DNA passed through MspA's constriction," the authors wrote.
Gundlach said that he has not estimated what the system's accuracy profile would look like since he is still working on developing a better base-calling algorithm. However, he said one challenge would be calling bases in homopolymer regions. The team's experiments indicate that the system was most likely to miss a base change in homopolymer regions.
"When you have more than four of the same nucleotide in a row, you don't quite know if a new one moved in or out," he said.
Akeson has also indicated that this will be an issue, and said that his team is working on developing a method to count the number of bases as they pass through the pore.
Gundlach's team also observed that the unzipping step was more likely to miss base changes than the synthesis step, particularly at the beginning or end of unzipping.
Despite these challenges, Gundlach said the system has "the potential of being a very good reader with very high confidence and quality."