NEW YORK (GenomeWeb) – A group of researchers from the Technion–Israel Institute of Technology and Boston University has demonstrated that a solid-state nanopore can be used to subtype pathogens based on genomic characteristics.
Not a full-fledged sequencing device, the device analyzes single molecules in order to detect SNVs, insertions, or deletions in order to determine characteristics like drug resistance of a particular pathogen.
Led by Amit Meller, a professor of biomedical engineering at the Technion-Israel Institute of Technology, the team published the proof of principle work in PLOS One recently.
"Instead of asking what the genome sequence is of the pathogen, you know it's one out of a series that have already been sequenced," Meller told GenomeWeb, and "you need to differentiate between genomic types, which is a much simpler question to ask."
Meller said his goal was to create a nanopore device sensitive enough to distinguish subtypes of pathogens and to identify important characteristics of pathogens, such as genes that confer drug resistance or increase virulence.
The group harnessed nanopores' abilities to distinguish between DNA fragments of different lengths and to determine whether one or two fragments are present in order to ask simple yes or no questions, explained Allison Squires, lead author of the study and a postdoctoral researcher at Stanford University.
The researchers built a solid-state nanopore device out of silicon with a diameter of 4 nm and a length of 7 nm. They ran the device in two different detection modes. In one mode, the device detects the specific length of the DNA fragment based on translocation dwell time. In this mode, researchers can detect whether a large insertion or deletion is present in the DNA fragment.
In the second mode, DNA samples are first exposed to a restriction enzyme that cuts at a specific SNV site or short indel. Then, when the DNA translocates through the nanopore, the device just has to distinguish between cleaved and uncleaved sites to determine whether the variation is present.
The goal of the proof-of-principle study was simply to demonstrate that the device could detect these two classes of variations: SNVs and small indels or large insertions and deletions, said Meller.
In the study, the researchers tested their method on Mycobacterium tuberculosis and Streptococcus aureus.
First, they determined the parameters of the nanopore by analyzing samples containing mixtures of DNA fragments of known lengths. Nanopore translocation analysis involves measuring the amount of current blockage as DNA moves through the pore as well as the amount of time each molecule spends in the pore, also called dwell time. For DNA sequencing applications, researchers have determined that different nucleotides result in a different current blockade, which means they produce a different electrical signal. Distinguishing between these different signals enables them to sequence DNA. However, because DNA moves so quickly through pores, there is a lot of noise.
Meller and his team simplified this process by not attempting to read every single base, but by trying to figure out the length of a DNA molecule and whether there are one or two fragments. The team analyzed the same characteristics — dwell time and current blockade — but did not need to distinguish between minute differences in blockade signatures and dwell time.
To determine the length resolution of the nanopore, the researchers analyzed DNA fragments of 100 bp, 200 bp, 900 bp, 1,000 bp, and a combination of 100 bp and 900 bp DNA fragments.
The team determined that the nanopore could easily distinguish insertion and deletion events several times larger than the base length. For insertions or deletions of 100 bp in length, when the base DNA fragment is also 100 bp, the molecule only has to be analyzed a few times to distinguish the event with greater than 95 percent confidence. But, for longer base DNA fragments, of about 900 bp, "several hundred events" are required to differentiate between the 900 bp fragment and the 1,000 bp fragment at greater than 95 percent confidence.
In order to evaluate smaller variations, like SNVs, the researchers developed a second mode of analysis that used restriction enzymes to cut the DNA at a specific variation. When they did this, they found that the nanopore readily distinguished between a sample of two DNA fragments that were 900 bp and 100 bp and a sample that was one contiguous 1,000 bp DNA fragment.
By applying Bayesian statistics, the researchers found they were able to provide an estimate of statistical confidence associated with each classification. They then used these statistics to determine how many translocation events they would need to correctly call cut versus uncut DNA. They found that after just a "few tens of events from an unknown test sample," the probability of a correct call increased rapidly.
At typical nanopore collection rates, less than a minute would be sufficient to obtain a confident result," the authors wrote.
Next, the team tested the device on different model pathogens—two strains of M. tuberculosis, one in which an SNV in the parC gene causes antibiotic resistance and another in which an SNV in the mazG gene affects virulence; and methicillin-resistant S. aureus, which is caused by large insertions or deletions often from whole genes from related bacteria or phages. In addition, MRSA also has a fair amount of genetic diversity, often exhibiting frequent small mutations and indels, in contrast to M. tuberculosis, which has little genetic diversity between strains.
For MRSA, the team selected two well-characterized isolates of the same strain: USA300-FPR3757 and USA300-HOU-MR. Although the two isolates are both resistant to methicillin, they have many other small indels, SNVs, and variations in plasmids and mobile elements that give them a different spectra of antibiotic resistance.
For M. tuberculosis, the team analyzed two closely related strains that differ only in a few SNVs and indels, but one is virulent and one is not.
For each pathogen, the researchers focused on specific genes. The researchers first used PCR to target and amplify the genes of interest from each strain and then used restriction digestion to focus in on the specific variations.
In the non-virulent M. tuberculosis strain, for example, restriction digest created a 942 bp fragment from the mazG gene, but in the virulent strain, restriction digestion created two fragments of 621 bp and 321 bp.
Similarly, for MRSA, after restriction digestion, either one single fragment 885 bp long of the parC gene remained in the strain that was multi-drug resistance, or if the strain was only methicillin resistant, two fragments were created. The team demonstrated that they were able to discriminate between the two strains with greater than 99.5 percent confidence after 80 translocation events.
Meller said that the team is now working on a number of steps to optimize the device. For instance, he said, in the proof of principle, the researchers used PCR. But for a commercial device, he said he would like to eliminate PCR since the goal is to have a device that would be fast and amenable to being deployed in the field or in situations where getting a rapid answer — in about an hour — is critical.
"We're working hard on demonstrating that we can do it without PCR," he said.
A second goal is to develop a device that can characterize not just one or two different variations that distinguish one strain from one other strain, but to be able to analyze many different variations that can distinguish one strain from an unknown sample.
Squires said that in order to assess more than one variation from a single genome at the same time, a device would consist of multiple nanopores, each one assessing a different variation. Those nanopores would run in parallel, she said.
For this to happen, the researchers would also need to automate the process, likely with the use of microfluidic chips so that multiple restriction digestion reactions could run in parallel, Squires said.
A commercial device would then likely include a pre-set "panel of critical sites" for point-of-care pathogen diagnostics, the researchers wrote. Squires added that it would also be possible to make a device without built-in panels so that researchers could design their own.
Meller did not provide a timeline for commercialization, but said he thought the group could be able to solve the problems of automation, PCR, and being able to analyze multiple variants from one sample within a few years.
In addition, last year, his group published a method for optical detection of nanopore translocation. The approach is very different, but Meller said it could be complementary to the electrical detection he used in this recent study.