NEW YORK (GenomeWeb) – A team of researchers led by the Broad Institute has used microfluidics in combination with a single-cell RNA-seq protocol to look at gene expression of more than 1,700 single primary mouse bone-marrow-derived dendritic cells.
Reporting their work last week in Nature, the group described how they used Fluidigm's C1 system, a transposase-based library prep, and the Smart-seq protocol for single-cell RNA-seq to understand how the gene expression of seemingly identical cells varies in response to stimulation.
The group sequenced the RNA of 1,775 single dendritic cells after the cells were exposed to three different pathogens at five time points and discovered that there was variation not only in the number of cells that expressed a specific transcript, but also variation in the level of expression for a given transcript in a specific cell. In general, they found that a small number of cells showed high expression of a core set of antiviral genes early on after being stimulated, but over time those transcripts became active in all cells.
The study is a continuation of earlier work the group did on just a handful of cells, co-senior author Aviv Regev told IS. The original process involved making over 20 libraries by hand, which is "not very scalable when you want to look at large numbers of cells." Incorporating microfluidics enabled the team to scale, in order to "understand response over different time points and different stimuli," Regev said.
Aside from enabling the team to scale the experiment, the microfluidics device had another important "off-label" use, Regev said. It allowed the team to isolate each cell so that the cells could not communicate with each other after being exposed to the stimulus. This insured that cells were responding to the stimulus alone and not to other cells.
The single-cell RNA-seq protocol within a microfluidic environment "allowed us to interrogate this dynamic immune response at the individual cell level," Andy May, co-senior author of the study and former director of R&D at Fluidigm, told In Sequence. "The modified protocol allowed us to isolate individual cells and stimulate them individually."
Rather than capturing the cells in the microfluidic device and immediately lysing them as part of the sample prep for sequencing, the team first kept the cells isolated in the individual chambers and then added the stimulus. The cells soaked in the stimulus for four hours, but because they were isolated from each other, the cells were responding only to the stimulus and not to each other.
After four hours, the team then continued on with the standard protocol for using the device for sequencing sample prep.
Another slight modification the team made, Regev said, was that they incorporated Nextera's transposase library prep to the Smart-seq protocol. "That's a twist, and definitely increased the throughput substantially," she said.
Researchers at the Karolinska Institute and the Ludwig Institute for Cancer Research in Sweden originally published the Smart-seq protocol for single-cell transcriptome sequencing in 2012. They later revised the protocol, calling it Smart-seq2, to sequence more of the RNA in cells and also to generate longer RNA transcripts.
At the time Regev's lab was doing its study, they were using the original Smart-seq protocol. She said the lab is now using the Smart-seq2 protocol, which is more sensitive and accurate and "less prone to dropping transcripts."
The Smart-seq2 protocol would have also resulted in the capture of more transcripts, she said. In the current Nature study, the detection efficiency was 20 percent, meaning that the protocol might not pick up rarer transcripts.
This 20 percent conversion efficiency is pretty standard for single-cell RNA sequencing protocols, May said. However, the main thrust of the experiment was to compare expression between cells at different time points, rather than to detect transcripts present at very low levels, so the conversion efficiency was not problematic.
Regev said the protocol enables both digital and analog measurements of gene expression — both if a cell is expressing a given transcript and also the level of expression. By contrast, bulk sequencing would not enable this level of discrimination, since it would only give the average expression across all cells.
The study could have important implications for understanding the mechanisms of immune response, she said. For instance, one major finding was that there are a certain number of "precocious" cells that respond early with high levels of expression to a pathogen. One hypothesis Regev said the team would like to follow up on is how controlling the number of these precocious cells affects immune response — whether having too many or too few of these cells results in either too much of an immune response or too little of a response.
She said the team will continue to refine the technique to increase the scale and also to work with more challenging cells.