A research team led by Carl Hansen of the University of British Columbia's Centre for High-Throughput Biology has developed an integrated microfluidic device for high-throughput digital PCR analysis of single cells.
The device combines two technologies that the group previously developed — highly parallel microfluidic single-cell sample preparation and digital PCR based on surface tension partitioning — and could enable a wide array of difficult-to-perform single-cell research applications such as quantifying short transcripts like microRNAs, elucidating RNA editing, or assessing allelic imbalances between single cells, according to the researchers.
In addition, Fludigm has licensed the surrounding intellectual property, although it does not have any immediate plans to develop a product based on the technology, a spokesperson said. However, this does not preclude the company from developing such a product in the future, and in fact, Fluidigm's commercially available C1 single-cell autoprep system already uses some aspects of the technology.
Hansen and colleagues described the new device in a paper published this week in Analytical Chemistry.
The group had previously developed a microfluidic device that performs reverse-transcriptase qPCR at a throughput of 300 single cells per run by sequentially transferring reagents through three chambers for the steps of cell lysis, reverse transcription, and PCR, resulting in PCR amplification of each cell product in single 50-nL chambers — work that was described in a 2011 PNAS paper (PCR Insider, 8/4/2011).
In addition, Hansen and colleagues also previously developed a digital PCR chip that uses oil-based surface tension partitioning to achieve planar dPCR densities of up to 400,000 reaction volumes per cm2 — so-called "megapixel" dPCR, which was described in a different paper published in 2011 in Nature Methods (PCR Insider, 7/14/2011).
"It's basically the ability to trap, process, and amplify nucleic acids from single cells integrated on a device similar to what Fluidigm has commercialized as the C1, and then putting digital PCR arrays on the end of that," Hansen told PCR Insider this week.
As described in the recently published Analytical Chemistry paper, the integrated platform, which has a physical footprint of 10 cm2, distributes cDNA from individual cells into a dedicated dPCR array consisting of 1,020 chambers, each with a volume of 25 picoliters. The high density of the dPCR format — 118,900 chambers per cm2 — enables the analysis of 200 single cells per run, for a total of 204,000 PCR reactions.
The researchers tested the ability of their device to conduct a variety of single-cell measurements. First, they measured GAPDH transcripts from single K562 cells. GAPDH encodes for glyceraldehyde 3-phosphate dehydrogenase and is a commonly used high-copy number endogenous control for RT-qPCR experiments.
This experiment revealed a log-normal distribution of GAPDH across single cells with an average copy number of 1,563 transcripts per cell. In three replicate measurements, using separate devices and different cell cultures, the mean GAPDH expression was very reproducible, ranging from 1,412 to 1,741 copies per cell, and was in good agreement with estimates of 1,761 copies per cell previously determined by normalizing single-cell microfluidic RT-qPCR measurements to a standard curve calibrated by dPCR.
Next, Hansen and colleagues measured the expression of BCR-ABL transcripts, a fusion gene resulting from a reciprocal translocation of chromosomes 9 and 22 that is associated with chronic myelogenous leukemia and is also present in K562 cells — although in relatively low abundance.
Three separate experiments reliably detected the BCR-ABL fusion transcript in every cell analyzed, with a mean expression of 33 copies per cell, a range spanning 25-fold in relative expression, and a standard deviation of 18.9 copies per cell. These numbers were consistent with an independent study based on RT-qPCR that measured BCR-ABL mRNA in single K562 cells ranging from 2 to 262 copies per cell with the majority of cells containing about 40 copies.
The researchers also created a two-step RT-PCR protocol to demonstrate the use of their technology to absolutely measure microRNA levels in single cells. The importance of miRNAs as post-transcriptional regulators and disease markers has been well-established, but their short length has stymied previous efforts at single-molecular quantification.
Using commercially available miRNA-specific stem-loop reverse transcription primers from Life Technologies, Hansen and colleagues measured the expression and variability of miR-16, an miRNA expressed at medium levels across a broad range of tissues, in K562 cells. Over three replicates, the cells were found to have a mean copy number of 675 per cell — in close agreement with dPCR measurements performed on bulk samples instead of single cells.
Lastly, the group used its device to measure the extent of single nucleotide RNA editing. Editing of RNA molecules by nucleoside deaminases has recently gained attention as a mechanism by which transcripts can be modified away from the genomic code, with potential implications for transcript stability, alternative splicing, translation efficiency, and protein sequence, the researchers wrote in their paper.
Specifically, they measured adenosine-to-inosine editing of position chr16:22296860 (hg19) in the mRNA coding for single EEF2K cells. This edit, they explained, was initially identified in lobular breast cancer and found to be edited at a frequency of about 0.33 in RNA-seq data. Using their device, Hansen and colleagues found that 71 of 221 cells (32 percent) expressed both wildtype and edited EEF2K transcripts; 12 of 221 (5 percent) expressed only edited transcripts; 103 of 221 (47 percent) expressed only wildtype transcripts; and 35 of 221 (16 percent) did not show expression of either form of the transcript. Thus, the population-averaged editing frequency was found to be about 0.19 from single-cell measurements, consistent with measurements made on dilutions of purified RNA.
In general, Hansen believes that the experiments demonstrate the potential of the new technology as a general tool for research and understanding the fundamental biology of transcription.
"There is a range of things … people are doing generally to measure single-cell transcription and gene expression," Hansen said. "People are doing a lot of RNA-seq … but currently you're not going to do that on thousands of cells — it's just too expensive and challenging. You may do it on a few cells to find candidate genes."
In addition, commercial technologies such as Fluidigm's BioMark and C1 "let you take panels of genes you know to be interesting and look across reasonable numbers of cells for expression … and it's reasonably quantitative."
However, he added, "digital PCR provides much higher accuracy, and the ability to really get down to very low copy numbers with absolute certainty, and to look at the expression of transcripts that are highly homologous — single-nucleotide variants — and quantify the difference between the two. It's really the precision instrument, and I see it being particularly useful for things like RNA editing … allelic imbalance, [where there are] two alleles with a different SNP, and you want to know noise and expression, or fundamental studies about noise and allelic imbalance and RNA processing."
Hansen also said that he can "imagine scenarios where this might be useful in diagnostics, but that is quite far-reaching."
Hansen is a longstanding member of Fluidigm's scientific advisory board, and the company has a broad IP license with his lab.
In fact, the technology described in the latest paper is also described in some claims of a patent that Fluidigm has already licensed in order to use aspects of the single-cell sample prep in its C1 system. Fluidigm also has a license to the megapixel dPCR technology, and has developed prototypes of such a device, but has not indicated that it is producing a commercial product based on it, Hansen said.
This week, Fluidigm spokesperson Howard High told PCR Insider that although the latest effort from the Hansen lab is intriguing, the company is not planning to move forward with a related commercial product.
"Fluidigm has worked with Carl Hansen on several microfluidic chips and approaches," High said. "We find his experimentation innovative and this latest effort to combine single-cell and dPCR into one structure is such an example. We didn’t work on this project because it didn’t fit into our current commercial priorities for single-cell, but share a common belief that innovative microfluidic-based solutions will provide the best platform for single-cell research and innovation."
Even though Fluidigm doesn't have a commercial interest in this particular product, "if we did want to [commercialize it] in the future, through this license we could use the IP that [Hansen] developed in creating this chip," High said. "In addition, we continue to work with Carl on other versions of single cell chips that might become commercial products or that have elements we might adopt in commercial products we design in the future."
From Hansen's perspective, Fluidigm's decision is not technology-based, and makes sense from a business standpoint.
"For market analysis reasons, they've decided not to make a product out of this," he said. "I can completely understand that — they've got the C1, and they've got [the BioMark] digital PCR system. I think it's a very interesting capability, and people will want to do it, but it's niche enough that I don't think it makes good business sense for them right now to do that."
Nevertheless, Hansen sees his group's technology as adding a formidable component to the ever-expanding toolbox for single-cell genomics studies.
"For now I think this is a scientific tool when you have a transcript you're interested in, and you need the precision and specificity and certainty that you don't get with qPCR," he said. "Despite the fact that single-cell qPCR on devices like these is very good, it's not perfect, and digital PCR does a better job. There's something very satisfying about being able to actually count the molecules, and know you've got [for instance] 15, and it wasn't just background."