NEW YORK (GenomeWeb) — Two research groups at Harvard University have independently developed droplet-based methods for profiling the transcriptomes of thousands of individual cells at low cost and high speed.
Both methods use droplet microfluidic technology developed in David Weitz's lab at the School of Engineering and Applied Sciences at Harvard, but they differ in their approaches for barcoding the mRNA of single cells for next-generation sequencing and other technical aspects.
While the Drop-Seq method, developed by the group of Steven McCarroll, a professor of genetics at Harvard Medical School, can process a larger number of cells in a single run because it uses a more diverse barcode library, the inDrop technique, from the labs of Weitz and Marc Kirschner in the Department of Systems Biology at Harvard Medical School, captures a larger percentage of cells from a given sample, which has advantages if sample is limited.
The two methods are "different attempts to solve a really hard problem," said Allon Klein, an assistant professor of systems biology at Harvard and the lead author of the inDrop article. "Some things we figured out better, other things they did better. In the long term, I hope we can find a method that captures the advantages of both methods," he told GenomeWeb. The techniques are described in detail in separate papers published in Cell today.
By expanding the scale at which single-cell transcriptomes can be studied, the methods open new avenues for studying the diversity of cell types in complex tissues, as well as their functions and relationships. "You start to see all kinds of patterns that would not have been apparent in smaller data sets," McCarroll told GenomeWeb.
In the future, the methods could be expanded to applications beyond single-cell transcriptomics, such as epigenetic profiling, genome sequencing, targeted mutation screening, and antibody-DNA-technology-based proteomics, and might enable multiple assays in parallel, Klein said. In addition to single cells, it could also be used to study organelles or individual chromosomes.
A main advantage of Drop-Seq over existing methods is its low cost, about $.07 per single-cell library, mostly due to enzymes, which is several hundred times less expensive than the Fluidigm C1, according to McCarroll. The technique is also fast — a single scientist can prepare 10,000 single-cell libraries per day.
To make Drop-Seq widely available to other researchers, McCarroll's group has posted a detailed laboratory protocol on its website that has been optimized beyond what is published in the Cell paper.
The technique requires less than $10,000 worth of equipment, he said, including an inverted microscope, syringe pumps, a magnetic mixing system, and a custom-designed microfluidic droplet generator that can be built by any academic or commercial microfluidics facility. Their bead library is available for purchase from a company called Chemgenes.
By growing the user community for Drop-Seq, the scientists hope to optimize the method further over time. "Drop-Seq is a technology in its infancy. We imagine that having a strong community of users who share experiences and solutions to problems will greatly accelerate the pace of technological progress, allowing scientists to tackle many more exciting biological questions," said Evan Macosko, the lead author of the Drop-Seq paper and a postdoc in McCarroll's lab.
Since Macosko presented Drop-Seq at the Advances in Genome Biotechnology and Technology meeting in February, interest in the technique has been high, McCarroll said, and the group has already helped several labs set it up.
It is unclear whether anyone plans to commercialize either group's method, or a combination of the two. Both groups said they have filed patents on aspects of their technologies but did not comment on whether a company has licensed them for commercialization.
"We would like to make the method widely available, and commercialization is a good way to do so, particularly if we can make a 'box' that anyone could use," Klein said. "Commercialization also might have drawbacks, if it stops freely mixing ideas from other technologies in order to get the best possible system for everyone to use," he added.
Klein said his group also plans to put its protocols online within the next few days. Interest from colleagues has been high, "so we are excited to be helping others explore the diversity within their cell populations, be it in the pancreas, the hematopoietic system, neurons," or other tissues.
On the commercial side, the two methods will likely compete with Fluidigm's C1. Earlier this year, Fluidigm announced a new chip for high-throughput single-cell RNA-seq that can profile up to 800 cells and will cut the cost per cell to $10.50.
Also, earlier this year, Cellular Research said that in 2016 it plans to launch a new platform called the Resolve system that will be able to process 5,000 to 10,000 single cells in parallel and have consumables costs of less than $1 per cell. The company's technology uses tiny microtiter plates with picoliter-sized wells and a library of barcoded beads.
In the past, Harvard has licensed microfluidic droplet technology from the Weitz lab to GnuBio, a startup that was developing DNA sequencing technology until it was acquired by Bio-Rad Laboratories a year ago. Bio-Rad already offers droplet digital PCR technology.
"We believe in the importance of droplet-based single-cell RNA-seq technology and we are engaged in its development for our future digital biology platforms," Annette Tumolo, senior vice president and general manager of Bio-Rad's Digital Biology Center, told GenomeWeb in an e-mail.
Drop-Seq starts with single cells in nanoliter-sized droplets, which are combined with droplets containing a single barcoded bead with oligonucleotide primers. The cells are then lysed and the beads capture the mRNAs to form so-called single-cell transcriptomes attached to microparticles, or STAMPs. After breaking the droplet emulsion, the mRNA is reverse transcribed, followed by amplification and sequencing of thousands of STAMPs in a single reaction. From the barcodes, researchers can deduce which cell a transcript originated from.
To generate the bead library, the scientists used a split-pool synthesis approach. Each oligo consists of a constant primer, a cellular barcode that is the same for each bead, a unique molecular identifier (UMI), and an oligo-dT sequence to catch mRNAs. The library has 16 million barcodes, allowing researchers to do experiments with hundreds of thousands of cells in parallel, McCarroll said, and they were able to detect tens of thousands of transcripts per cell.
To gauge the accuracy and sensitivity of their method, the scientists first applied it to a mixture of human and mouse cells, which allowed them to detect instances where libraries derived from more than one cell. Depending on the cell concentration, they detected up to 11 percent cell doublets, where two cells stuck together and their transcriptomes were analyzed as one.
According to McCarroll, it will be important to test other single-cell technologies with mixed-species samples as well. When they ran their mouse-human mixture on the Fluidigm C1, for example, they found that 30 percent of the resulting libraries were species-mixed.
From the expression profiles of about 1,000 mouse and human cells, they were able to infer the cell cycle phase of the cells. Next, they applied Drop-Seq to almost 50,000 single cells from mouse retina, a widely-studied neural tissue with five neuronal classes. Based on their transcription profiles, they found 39 distinct cell populations.
McCarroll is currently using Drop-Seq to study the cerebral cortex and other parts of the brain in order to understand how it changes during development and how it is altered by genetic mutations.
InDrop, short for indexing droplets, also starts by encapsulating cells into droplets and combining them with barcoded oligo primers, followed by lysis and reverse transcription of the mRNA. After breaking the droplets, the cDNA libraries are sequenced.
Instead of beads, the Klein group uses a library of barcoded hydrogel microspheres (BHMs). Each BHM carries covalently coupled, photo-cleavable primers with a barcode. A total of about 150,000 barcodes are available, allowing for random labeling of about 3,000 cells.
After validating their method with a mixture of mouse and human cells, the researchers profiled more than 10,000 mouse embryonic stem cells and sequenced about 3,000 of those cells at greater depth, identifying rare sub-populations that they said would be difficult to find by profiling just a few hundred cells.
Klein said he plans to use inDrop to study stem cells further in order to better understand how they decide to develop one way or another.