By Aaron Krol
February 29, 2016 | NanoString Technologies, a Seattle company with a small but comfortable niche in automated genetic analysis, is preparing to make the leap into DNA and RNA sequencing. The company revealed its novel sequencing process in a poster at the Advances in Genome Biology and Technology meeting, held in Orlando, Fla. earlier this month.
NanoString’s proposed sequencer is closely related to its current line of nCounter instruments, which target short genetic sequences with optically barcoded hybridization probes. In an nCounter experiment, a set of probes designed to bind with a specific group of genetic elements―for example, segments of genes that are commonly overexpressed in cancer―is mixed with a DNA or RNA sample. Each probe carries a unique sequence of fluorescent tags, the optical barcodes that are imaged to learn which genetic targets are present in the sample, and in what numbers.
The new sequencing process, named Hyb and Seq, uses very similar probes. But instead of just spotting key genes and variants, a Hyb and Seq experiment can read huge volumes of barcodes to recover the complete sequence of a DNA molecule.
NanoString is not trying to sprint to the front of the pack in sequencing, a competitive field where the market leaders are increasingly able to produce whole human genomes cheaply and quickly. Hyb and Seq, by contrast, can only be used for targeted gene panels―focused experiments where users already have some idea what they’re looking for. But as the market for DNA information widens, there is more room for technologies that fill interesting niches, and Hyb and Seq has several unique properties that could give it an edge for NanoString’s core applications in cancer testing.
“It’s not for whole genomes,” says Joe Beechem, Senior Vice President of Research and Development. “It’s very much a targeted cancer panel we’re developing.”
In a Hyb and Seq experiment, single-stranded sample DNA molecules are immobilized on a flowcell and inserted into the instrument. (For its pilot studies, NanoString is using a modified nCounter SPRINT instrument, with added 3D-printed components.) The DNA does not have to be amplified or prepared with enzymes―the sequencing process works directly with the native molecules. Once the DNA is on the flowcell, a mix of over 4,000 different probes is added, at low concentrations that ensure probes will attach to the DNA more or less one at a time.
The probes in Hyb and Seq are much shorter than in NanoString’s gene expression assays, targeting sequences of just six DNA letters; together, they cover every possible six-letter combination of the DNA bases A, T, C, and G. After a probe hybridizes to the sample DNA, the sequence it targets can be read optically. The barcodes in Hyb and Seq consist of four fluorescent molecules, one for each DNA base, which attach sequentially to the probe and are imaged one at a time.
Reading the barcode reveals only that a certain six-base sequence appears somewhere on the target molecule. On its own, this is a very small amount of information: the instrument doesn’t know where on the molecule this sequence occurs, and six bases isn’t enough to say much about what region of the genome a DNA fragment comes from. But the optics and chemistry cycle very quickly, so in a short time, the system picks up a lot of random six-base sequences. Because Hyb and Seq is used for targeted assays, with only a limited number of gene regions represented in the sample, NanoString’s software can rapidly eliminate regions that don’t have the right combinations of six-base sequences in them.
“You quickly know what gene you’re on,” says Beechem. At that point, additional sequence fills in gaps and covers genetic variants. “Then it’s just a matter of covering all the areas you want to sequence… It’s like a mini-assembly of the individual genes in your panel.”
“A Highly Accurate Readout Very Quickly”
Hyb and Seq will never be able to take on tasks like de novo sequencing, but for assays covering a few key genes, it has some intriguing advantages.
First, there’s sample prep. “It’s almost zero,” says Brad Gray, President and CEO of NanoString. Most sequencing technologies require sample DNA to be copied through the time-consuming process of PCR, producing huge volumes of DNA fragments to read in parallel. But Hyb and Seq looks at single molecules, so the only prep involved is to pull targeted gene regions out of the sample―in the case of cancer assays, typically DNA from a formalin-fixed, paraffin-embedded (FFPE) tissue biopsy. It takes just a few minutes to mix sample DNA with targeted capture probes and insert it into a NanoString flowcell for sequencing.
“The fact is that [sequencing] is still not a clinically friendly technology for a lab tech to run,” says Beechem. “There is still an unmet need for a radically simplified workflow and hands-on time.”
Hyb and Seq also has the unique property that the sample DNA is not damaged or altered. Barcoded probes can be repeatedly attached and washed off, so getting higher coverage to correct errors is as simple as resequencing the same DNA molecule over and over again. Beechem calls it “sequence until,” because the system can be programmed to keep running until it reaches a certain accuracy threshold, rather than for a specified number of cycles.
This could be especially useful when working with FFPE samples, where DNA is typically scarce and badly damaged. NanoString is designing its first demonstrations of Hyb and Seq for exactly this type of sample, which usually yields DNA fragments no more than three or four hundred bases long. “It’s a very natural extension of the business we’re already in,” says Gray. “The company mission is built around getting biomarker information out of tumor tissue, which can be used for research and ultimately in the clinic.”
Hyb and Seq is far from commercialization―Gray says a market-ready version, with a dedicated instrument, isn’t in the cards until at least 2017. But the earliest test runs suggest the system is workable, at least for sequencing small genetic variants on short DNA fragments. The rate of miscalled bases in each read is less than 3%, and while NanoString is still reviewing the data it has, Beechem says that so far there’s no indication of biased errors that would prevent that rate from improving with resequencing.
“Those are the most accurate single-molecule, single-pass error rates that have ever been measured, by a large factor,” he says. “And the fact that we can reread [DNA molecules] gives you a highly accurate readout very quickly.”
For all this early promise, NanoString is planning to enter one of the most competitive parts of the sequencing market. Cancer genomics is one of the few applications of DNA sequencing that can be used meaningfully in the clinic today, and companies are hurrying to launch comprehensive assays that small hospital labs can run in-house. Illumina’s line of TruSeq tests, and Thermo Fisher’s AmpliSeq panels, cover much the same ground as NanoString’s proposed assays. And QIAGEN recently released the GeneReader specifically for the cancer market, running nothing but targeted gene panels. (Unlike with Hyb and Seq, this is a deliberate choice and not a technical limitation.)
Hyb and Seq will be fascinating to molecular biologists for its distinctive approach to getting sequence data, but whether it’s a leap forward for NanoString will depend more on mundane factors like speed and cost. Still, the minimum of hands-on time is a novelty that other technologies will have a very hard time matching, and the single-molecule process lends itself to tumor biopsies in a way that the competition does not. For now, this latest entrant into sequencing is a curiosity, but it’s a curiosity well worth keeping an eye on.