NEW YORK (GenomeWeb) - Scientists from Columbia University have developed a new microwell device that enables solid-phase capture of RNA on either glass plates or beads, which they say could greatly increase throughput and have numerous applications for single-cell biology.
Led by Peter Sims, the researchers described their device and demonstrated its utility in single-cell RNA sequencing in a paper published online earlier this month in Genome Biology.
"There have been three or four really nice papers about conceptually similar techniques," for single-cell biology, Sims told GenomeWeb, including a technique called Drop-seq, which uses nanoliter droplets to prepare and barcode single-cell transcriptomes for sequencing. "I think this will change the field, these new techniques," he said.
Sims' microfluidic device is built on a standard microscope slide, with a thin layer of silicone rubber pocked with microwells and a fluid chamber layered above the wells. The authors wrote that the device enables solid-phase capture of RNA, which has several advantages compared to capture of RNA in solution, including "facile fluid exchange, removal of contaminants, and compatibility with high-resolution imaging."
To capture RNA, the researchers can run cells through the device, using gravity to sink the cells into the microwells, usually with one cell per well. From there, the researcher using the device has two options. "One way to do it is to capture the RNA on a glass surface" printed with hybridization probes, Sims said. "It's kind of like a microarray where each spot is a single-cell transcriptome." In this case, the wells are sealed by the glass.
The other option, which the scientists used to show proof of concept in the paper, is to use beads covered with uniquely barcoded hybridization probes to capture the RNA. In this case, the wells are sealed with oil.
Though the authors chose to demonstrate the use of the device in RNA sequencing, Sims said there are other possible applications for the technology. "We could use fluorescent probes after capturing RNA on the surface of the glass cover slip to quantify certain genes, [or] we could use qPCR. We chose sequencing because it's an inexpensive way to get a genome-wide readout and it's already available."
They saw a benefit in not having to develop readout technology and focus on the device itself, he said. "But going forward we're interested in other ways to do [readouts]," he said.
To access the transcriptomes for sequencing, cells are lysed, the RNA is hybridized on the probes, and reverse transcribed to create cDNA, followed by second-strand synthesis and linear amplification, all of which happen on the chip using the microfluidics
To create the sequencing library, the researchers performed in vitro transcription with RNA polymerase to put the linear amplicons in solution. "In one step we amplified and eluted our pooled library from the bead. That's our way of getting the material off the bead," he said. "Once you have that, all the barcodes from all the cells are linearly amplified."
From there, it's a normal sequencing library preparation process, Sims said. The RNA was pipetted out of the device, enriched with PCR, and sequenced on Illumina NextSeq 500 machines.
"The beads are the easiest way to do sequencing with this approach because there's very easy ways and scalable ways to make uniquely barcoded beads in large quantities," Sims said. "If you wanted to use more of an imaging-based readout, putting the RNA on glass is more advantageous because it's a great imaging substrate."
The scientists wrote in their paper that while the system was cheap, it had several issues that needed to be worked out, especially capture efficiency. The average number of genes detected per cell was lower than the number of genes detected in top-producing cells. In one experiment, they detected only 530 genes, on average, for the total number of about 500 cells, while detecting an average of 1,030 genes in the top half of single-cell profiles.
"We've got room for improvement in terms of capture efficiency," he said.
"Everybody wants to increase capture efficiency, [which is] certainly one of our next steps," though he declined to disclose what strategies they would pursue in trying to achieve that goal.
However, Sims said that even with low efficiency, single-cell sequencing can still produce meaningful results for biological research. "You can make up for low capture by having lots of cells. You don't see every molecule from every cell, but you may see every gene from a few cells, and you can use that info to infer subpopulations [of cells]," he said.
Identifying cell subpopulations is important to research that Sims does with cancer cells, but he said it's also important in basic characterization of healthy tissues. "I think that there's a lot still to be discovered about the cellular composition of complex organs like the brain [and] complex systems like the immune system, just in terms of what are the cell types that are there in a healthy state."
"If you have a complex tissue sample and you want to identify the underlying cell subpopulations, then you don't need very high coverage, but you do need to sequence a large number of cells," he said.
High throughput is something that Sims believes his platform can deliver on, since it's cheap and he's demonstrated that it can process hundreds of cells in parallel. Sims chose to place his device's performance and cost in context by comparing it with Fluidigm's C1 Single-Cell Auto Prep system.
"Currently our capture efficiency is much lower than Fluidigm's, but our library prep technique is much cheaper, so it's in the $.10 to $.20 range [per cell]. Theirs is in the $10 range per cell," he said.
"Our microfluidics system is capturing and barcoding hundreds of cells in parallel. Fluidigm does tens of cells," he added. However, Fluidigm recently announced it would offer a new chip to prepare up to 800 cells for sequencing at a cost of $10.50 per cell.
The cost savings are partially due to the very small volume of reagents needed, as well as the barcoding scheme, which creates sequencing libraries for all of the cells in a single reaction, Sims said. "If you want to sequence 300 cells, you can do one reaction to barcode everything. Now you're doing hundreds of reactions for the price of one," he said. "The more you scale it up, the cheaper it gets."
Sims said that there has been commercial interest in the technology, but declined to provide any details.
He said the next step is to work on improving capture efficiency, develop alternative readout methods, and integrate the transcriptomes the system can produce with other data points.
"That's going to be the next phase of progress in the field. We've got systems for single-cell transcriptome analysis," he said. "Now, can we build in other information that you can't infer from the transcriptome? There's a lot we can't really know just from the transcriptome alone," he said, noting that RNA sequencing can't provide information on post-translational RNA modification or protein abundance in the cell.
"Integrating two types of single-cell measurements, I think that's going to be the next phase of these new tools," said Sims.