This story was originally published Aug. 17.
Researchers at the University of Southern California in Los Angeles have developed a method to sequence the transcriptome of single neurons after taking electrophysiology measurements, allowing them to correlate neuronal phenotypes with gene expression.
The scientists, who collaborated with Illumina, published their method last month in Frontiers in Genetics. According to James Knowles, professor of psychiatry and the behavioral sciences at the University of Southern California's Keck School of Medicine and a senior author of the paper, this is the first time researchers have combined the patch clamp methodology with RNA-seq.
The approach requires no sophisticated laboratory equipment, making it broadly accessible, the authors noted.
The new method reflects the growing interest in single-cell genomic analysis. At the Biology of Genomes meeting this spring, for example, several researchers presented methods for sequencing the genomes of individual cells for cancer, genetic mapping, and other studies (IS 5/15/2012). Also, last month, Swedish researchers published a study where they sequenced the transcriptomes of single circulating melanoma cells (CSN 7/25/2012).
Previous studies have used microarrays to measure gene expression in single neurons, but according to the USC authors, sequence-based transcriptome profiling offers a number of advantages, such as an extended linear detection range, high accuracy, binary readings, and independence from a reference genome.
"Our study differs from previous studies in that we developed protocols specifically for assaying neurons, and our method can be directly coupled with electrophysiology studies on individual neurons, thus enabling the correlation of single-cell cellular phenotypes with single-cell expression phenotypes," they wrote.
For their study, the scientists analyzed five cultured mouse embryonic hippocampal neurons as well as three layer-5 pyramidal neurons from a fresh mouse brain slice.
Using a patch electrode, they first measured membrane and electrophysiological properties of a single neuron and then sucked out its contents for sequence analysis.
They prepared RNA-seq libraries using the SMARTer Ultra Low input RNA kit from Clontech, which can take as little as 100 picograms of RNA input, and Illumina's DNA sample prep kit, and sequenced the library on the Illumina GAIIx. On average, they generated 22 million 50-base-pair reads per neuron, which they aligned to the mouse genome and analyzed to quantify expression levels for each transcript and gene. Knowles said the method should also be adaptable to other sequencing platforms.
The scientists found that gene expression differed markedly between individual cells, but more between the cells from the brain slice than between the cultured cells, suggesting that neurons in the brain are more heterogeneous than in vitro. While they did not perform technical replicates, they wrote that "our previous experience showed that correlations between technical replicates are typically >0.95, so that noises from the sequencing run per se are unlikely to have major influence."
Using patch-clamping to obtain the RNA from single cells has advantages over other methods, Knowles said. Dissociating cells and separating them via a cell sorter, for example, might change gene expression. And using laser capture microscopy, it is difficult to hit just one cell without getting parts of another cell as well. On the other hand, his own method might leave some of the cell content behind, he said.
Contamination is a concern, and minute amounts of nonspecific DNA or RNA "ruin your prep," Knowles said. In fact, in their paper, the researchers noted that their first set of experiments failed because their sample was contaminated with DNA from pollen, since the work was done on an open lab bench in the springtime. To avoid this, they have since switched to a laminar flow hood.
Another possible way to avoid contamination is to use a microfluidic device as a barrier. Knowles said his group has been using NuGen Technologies' Mondrian system (IS 4/17/2012) to generate DNA sequencing libraries and plans to adapt it for single-cell RNA-seq libraries as well. They also recently started to use NuGen's Ovation Ultralow Library System kit for RNA-seq and have seen "good performance" with that as well.
As a next step, Knowles and his team plan to use the method on neurons in hippocampal sections that have been stimulated with excitatory molecules in order to see how gene expression changes at the single-cell level.
Knowles said the method could be further improved by using an array of micropipettes, which a computer could move through brain tissue, sucking out neurons as they are hit by the pipettes. That way, a three-dimensional tissue transcription map could be constructed. "That's something we could envision but is not currently funded," he said.
Also, he said, as the technology improves, it might become possible to do RNA-seq not only on single cells but on subcellular compartments, for example dendrites and axons of neurons. There is some evidence, he said, that different isoforms are directed to different regions of a neuron.