Researchers from the Broad Institute are designing a microfluidics-based low-cost sample preparation protocol for next-generation sequencing and have built a prototype chip that consists of around 100 40-nanoliter circular reactors in which all the steps of library construction can be performed.
Paul Blainey, a core member of the Broad Institute and assistant professor of biological engineering at MIT, presented on the technique at last month's Advances in Genome Biology and Technology meeting in Marco Island, Fla.
He became interested in designing such a technology in order to enable a project that would study microbial evolution at the single-cell level. The goal, he said, was to obtain an unbiased, genome-wide mutational assay with single-cell and single-generation resolution.
However, despite the advances in sequencing technology, such an experiment would be too expensive, he said, "not because of the sequencing itself, but because of the sample prep." Using existing technology, he estimated that it would cost a couple hundred dollars to prepare a sample and "we needed to be doing hundreds of cells every day."
Incorporating similar microfluidics techniques published last year in PLoS one with a transposase-based tagmentation library preparation protocol, Blainey's team designed a reusable self-contained chip the size of a microscope slide that could prepare sequencing libraries for 100 samples at once.
However, Blainey said that his chip has some key notable features. For one, "most microfluidic solutions allow you to mix things together and do a reaction, but they don't do clean ups," he said, which has been "a technology barrier for getting library construction working in the microfluidic context."
One microscope slide-sized chip contains around 100 40-nl circular reactors. Sample input can be genomic DNA or cells. All the library construction reactions take place within those nanoliter-sized reactors by using solid-phase reversible immobilization beads, which capture the cells or DNA, and allow the sample to be tracked throughout the construction process. All the steps of library construction can be carried out on the chip, which includes microfabricator pumps for mixing reagents.
At the end, the spent solids can be washed away. As proof of concept, Blainey's lab tested the ability to construct a library from an Eschericia coli sample and also a metagenonmic sample using a transposase-based method that is now sold by Illumina's Epicentre Biotechnologies. The method relies on a hyperactive version of the Tn5 transposase, which simultaneously fragments DNA and adds adaptors. The resulting products are then PCR amplified.
Blainey compared the method to the sequencing results of a high-quality reference E. coli genome sequenced by Illumina. Using his lab's microfluidic technique, the team started with 38 picograms of E. coli DNA and sequenced the resulting libraries using 2x100 bases of paired end sequencing on the Illumina platform. They compared their results to the Illumina-generated dataset, which used 1 nanogram of starting material and 2x250 bp sequencing.
The comparison showed that the microfluidic chip can "produce assembly-grade libraries," Blainey said.
Additionally, he said that barcodes can also be added to each sample while they are on the chip, so that after library construction, samples can be pooled and sequenced. Theoretically, the chip can produce 1,000 libraries per day, although he has not yet demonstrated this capability.
So far, Blainey said that his lab has constructed between 100 and 200 libraries and "even in this early phase testing, we've been getting really reliable production of libraries out of this device."
Currently, he said the lab is focused on workflow integration and validating the performance of the chip. "Ok-ish libraries are not good enough for us. We need top quality libraries," he said. "So we're just cranking through some samples to get a sense of the quality and will then be adding on the extra elements like DNA extraction, so we can take cells as an input."
He added that while demand for such a device is high, his lab is not looking to commercialize the device, although in the future the Broad Institute may decide to offer it as a service. In order to realize the low-cost potential, however, many samples would have to be prepared at once.
Blainey said that the goal is to be able to do library construction at a price of around $1 per sample. "We haven't gotten there in practice yet," he said. Key to hitting this price point will be integrating all the steps of the workflow to eliminate labor. "We've demonstrated all the pieces, but haven't put it all together yet," he said.
Additionally, he said his lab is not the only one working on microfluidics-based library construction. Last year's PLoS One publication was led by Stanford University's Stephen Quake, researchers from Singapore's Agency for Science, Technology and Research, and researchers at Fluidigm, which uses microfluidic technology in its C1 Single-Cell Auto Prep system.
The PLoS One authors described a "similar concept," Blainey said, that also includes the cleanup step on the device. However, they use a different microfluidic design and the method "doesn't really speak to workflow integration." That approach "takes in DNA that's already been sheared and then they do adapter ligation and spit it out onto the chip," so it doesn't include the entire library construction process.