NEW YORK (GenomeWeb) – Researchers at The Broad Institute of MIT and Harvard and Cambridge University have published a validation of an integrated, high-throughput microfluidic sample preparation system for genomics.
The researchers claim that the device can reduce DNA input 100-fold — from more than 1 million cells required to fewer than 10,000 — compared to high-throughput liquid handling robotics and electrowetting-based digital microfluidics.
In the study, published last month in Nature Communications, the device was used to perform whole-genome shotgun sequencing of Mycobacterium tuberculosis and bacterial colonies in soil, as well as to generate nearly 400 clinical Pseudomonas aeruginosa libraries.
"Sample prep is a little bit unsexy, but we see it as a key technical and economic bottleneck," said Paul Blainey, a Broad bioengineer and corresponding author on the study. For example, in applications like microbial isolate sequencing it might cost about $10 to generate all the data for a draft genome, but the sample prep can cost hundreds of dollars through a service. "As sequencing keeps getting cheaper, very quickly all NGS applications are going to become sample prep limited," he said.
On the other hand, the device Blainey's lab has developed with a large contribution from post-doc Soohong Kim can perform three essential elements — extraction, reaction steps, and clean-up steps — that make up the basic operations sufficient to get through prep for RNAseq, ATAC-seq, ChipSeq, and other NGS-based methods. "We see a huge need coming for that, as well as for human genomic and epigenomic NGS assays," Blainey said.
In addition, the automated microscale extraction component of the device is quite novel, Blainey said, emphasizing that the group "showed good performance at very low inputs for some tough bugs" such as pseudomonas species in soil isolates and lab-cultured TB.
Technologies typically used to prepare samples for sequencing workflows include standard manual benchwork and large liquid-handling instruments that require substantial investment and custom development of robotic scripts. Illumina had a microfluidics-based solution called NeoPrep, which was launched in 2015, but that technology is scheduled to be discontinued. Oxford Nanopore, meanwhile, recently began an early-access program for its system, called VolTrax.
But, "Most of the microfluidic systems out there do only a tiny part of the workflow," Blainey noted. NeoPrep, for example, required inputting sheared genomic DNA, so there was significant initial processing off-chip, he said, as well as tens of thousands of dollars of investment in an acoustic shearing instrument. "One of our big interests was having the chip do, essentially, all the work," he added, and the study provides an example of putting cells directly onto the chip and completing the entire workflow to generate a library, with the exception of one off-chip PCR the particular reaction happened to require.
Furthermore, "Despite the potential for cost reduction and input reduction, we're very much holding the line on data quality, and that is not something we were interested in compromising on," Blainey said.
Blainey noted that even for something "relatively simple" like microbial whole-genome sequencing, there is typically a separate liquid handling robot for each of the eight to 10 steps. "Our sample-in, library-out microdevice eliminates all of that, which is a practical limitation on turnaround time for high-throughput workflows currently," he said.
The device is unique in that it combines common microfluidic motifs in a new way, Blainey said. Specifically, it uses a rotary reactor as well as a unit that is essentially a "leaky valve" modified to trap beads and cells that acts as a reversible mechanical filter and is less prone to clogging that typical filters.
The device also has a high density of reaction wells — 96 reactors in a chip that is smaller than a business card, Blainey said. And, because it is multilayered with a vertical passage, or via, it can have input and output to individual units without crossing channels. The alternative would have been to put additional external access ports throughout the device, which would have limited the possible density of reactors.
The rotary reactor, leaky valves, layering, and vias also enables on-chip cleanup reactions, which had been a stumbling block for previous NGS application technologies, Blainey said. Namely, the valves allow for pull-down reactions onto beads, and, in the study this modality was used for DNA extractions and DNA clean-ups.
"That capability hasn't been commercialized before in this sort of device," Blainey said, noting that electrowetting-based systems, like NeoPrep, are capable of performing clean-up steps, but they are much lower-density systems.
The unique microarchitecture can also drive "a new paradigm in designing microfluidic systems for different workflows," Blainey said. In previous methodologies, putting a three-step kit workflow on a chip would require three chambers or reactors, with each chamber in the chain being successively bigger than the last in order to dilute out inhibitors. "It was basically a big ugly workaround for the fact that there has been lacking a practical way to do clean-ups," he said. And changing to a four-step workflow then required a redesign of the microfluidic device.
"With our device, there is only one reactor — we do a reaction, then we clean stuff up, then we put the products back in that reactor for the next step," Blainey said. Thus, the system can handle a process that has one step or 10 steps without any chip redesign needed.
The device has been the subject of ongoing conversations between Broad's Office of Strategic Alliances and Partnering and different industry groups, Blainey said, but so far there are no definitive commercial plans. However, commercialization remains the ultimate goal. His team is also using the device to sequence samples obtained from a clinical trial of a MRSA erradication protocol, and that trial, called Project CLEAR, is ongoing, he said.