Researchers at the University of California, Irvine, have created a microfluidic droplet array that can process 1 million individual droplets for digital PCR — a 100-fold increase in the number of on-chip, non-serial reaction volumes over other digital PCR methods, according to the developers.
The increased droplet number as well as other design details — such as a three-dimensional stacking design, on-chip thermocycling, and wide-field fluorescence imaging — are meant to overcome limitations in dynamic range, throughput, and imaging ability of current digital PCR technologies, the group reported in a paper published this month in the journal Lab on a Chip.
The authors demonstrate in the paper that their system can generate more than a million monodisperse, 50-picoliter droplets in two to seven minutes. The droplets then self-assemble into a three-dimensional packing configuration, where they can undergo on-chip PCR amplification and fluorescence detection with a low-cost 21-megapixel digital camera.
The entire PCR process can be completed in two to four hours, the authors report, including up to 1.5 hours each for thermocycling and image processing.
Abraham Lee, director of the Micro/Nano Fluidics Fundamentals Focus, or MF3, Center at UC Irvine and a senior author on the paper, told PCR Insider last week that his team set out to achieve a 1-million-droplet device originally with input from Beckman Coulter (now part of Danaher), which was an industrial member of the MF3 Center at the time the project was started, but is no longer directly involved.
"Through some discussion [with the company] we found an area that had some challenges in the field in terms of microfluidics for some of the PCR needs, and that's a combination of digital PCR, emulsion PCR, and real-time PCR in a high-throughput fashion," he said.
Digital PCR requires a large number of reactors to increase dynamic range, which is reduced by the low sample concentrations necessary to partition a sample into single molecules for quantification. "So that's one thing we set out to do — is to have a large number — and we set one million as a reasonable and practical target to achieve," Lee said.
While digital PCR technologies from companies like Bio-Rad's QuantaLife can generate individual reaction droplets numbering in the millions, they do not allow imaging for quantitative real-time data collection, according to Lee. RainDance Technologies also has a similar picoliter droplet technology but does not yet have a commercial digital PCR product.
"They usually do a serial process, meaning they do PCR with a flow-through, or somehow they do PCR and then run it through to do detection," he said. "That means that after the reaction or during the reaction, you look at one reactor at a time, so you lose the temporal information."
"People can make a million droplets, but if you want to fill a chamber around a few centimeters square with 50 microliters of sample, which is a commercial-level sample size, [that's not been done]," he said.
On the other hand, technologies like the Digital Array from Fluidigm and the Open Array from Life Technologies can't partition nearly as many individual discrete reaction volumes, but do allow imaging and real-time data collection, Martin Pieprzyk, director of marketing at Fluidigm, told PCR Insider last week.
Pieprzyk said that the "form factor in which [the UC Irvine] technology makes sense is if they can image [the reactions] while thermocycling to obtain the real-time data," Pieprzyk said. "But once you have that real-time data, how useful it is would depend on the type of assay and application being run."
Pieprzyk added that most of Fluidigm's customers using Digital Arrays for digital PCR experiments – where they are only trying to determine if reaction volumes with single molecules have been amplified or not – end up not using the real-time data even though it is readily available. "At this point in time very few digital PCR applications require real-time data," he said.
Pieprzyk added that the UC Irvine system is "an interesting but not groundbreaking technology. The major issue is that the market is still looking for the killer digital PCR application."
In a follow-up e-mail, Lee told PCR Insider that a practical benefit of being able to perform real-time PCR on digital PCR reaction volumes is the ability to observe PCR efficiency. "You can check each molecule being thermocycled and detect after each cycle, since with our imaging platform you can see all reactors at once over time," he said.
The UC Irvine group's platform design uses a droplet splitter that divides a parent droplet eight times into 256 "daughter droplets." In a paper earlier this year, also in Lab on a Chip, the group discussed its "tunable" droplet packing method, in which the height of the imaging chamber is tweaked so that droplets arrange in three dimensions, rather than spreading out in only one layer.
"We found that … you don't have to spread them out in a two-dimensional way like any other platform so far has demonstrated. We can actually stack the droplets because they are semitransparent, so as long as you stack in a two- or three-layer fashion, you can still resolve all the droplets in terms of their fluorescence through our imaging platform, which is a very inexpensive, low-cost setup using basically a commercial camera and a lens to resolve at least 20 pixels per reactor."
In the more recent paper, the group reported that for the purposes of visibility, they tested a two-layer "lattice-like" arrangement within an 11-square-cm array. Imaging of this arrangement and concentration was possible with a 21-megapixel consumer dSLR camera, the authors reported.
"The [droplet] stacking allows you to [pack] more reactors in the same 2D area. For a square centimeter, you can get two or three times more reactors imaged because they are stacked up," Lee said.
For droplet numbers higher than one million, the authors reported, higher pixel ratio numbers will be required to maintain the 15 to 20 pixels per droplet necessary to resolve each independent droplet's fluorescent signal.
The group tested the platform by performing on-chip PCR thermocycling of the reactor droplets, which required a total of 65 minutes. Droplets were able to undergo 40 or more cycles with less than 5 percent coalescence, the authors reported.
The authors wrote that due to the "deformability of droplets and PDMS," and "variations in actual droplet size," the lattice orientation in their test frequently combined both square packing and hexagonal packing. This could cause errors in determining the droplet number for digital PCR, but they compensated for this by measuring actual droplet size and density in the array image.
Testing a variety of PCR dilutions, from 20 to 100,000 copies in 50 microliter samples, the researchers found that total DNA concentrations measured by their platform closely matched the predicted outcomes with average errors of less than 15 percent, a "more than advantageous level of performance for digital PCR," according to their report.
"Compared to other works on digital PCR, this system of 1 million droplets yields a 100-fold increase in non-serial digital PCR reactor number, and is the first to integrate a 1-million droplet array throughput with real-time imaging ability on a single microfluidic device," they wrote.
Lee said that the platform demonstrates higher dynamic range than currently available systems.
"At the time we started and then demonstrated our platform, we were kind of limited [to] 105 in dynamic range at best due to some imaging difficulties," Lee said. "It was still the best based on reported numbers we saw. Since then, however, others have pushed the limits of their dynamic range to similar or even better dynamic range based on end-point PCR. Therefore in the paper we did not over stress the dynamic range claim."
However, Lee added that his group's technology can combine digital PCR for low target concentrations with real-time PCR for high target concentrations to further increase the dynamic range. "Theoretically, endpoint PCR [has] an upper limit whereas real-time PCR can continue to resolve higher concentrations," he said.
In addition, "we're not claiming high speed here, but these are small reactors and you can optimize that to some degree," he said. "Compared to other technologies [like QuantaLife and RainDance], our platform offers a larger reactor number with real-time PCR capability. We felt that this is novel and complementary."
Lee said that Beckman Coulter was originally interested in commercializing the platform, but the company ended up separating from the M3 center.
"We continued to hack along even after the initial work together ended," said Lee, adding that the group is now "very comfortable that we have the superior technology."
At this point Beckman Coulter still has "remote interest," according to Lee, though he noted that the team has also been approached by "the usual suspects" in the digital PCR field, including RainDance and QuantaLife.
"Honestly I don't know where it's going to end up in terms of commercial prospects," Lee said. "We have the IP and Quantalife and RainDance all think it’s a great technology. But they're all busy with their own technologies, so it's hard to say where it's going to go … but I think it's valuable."
— Ben Butkus contributed additional reporting to this article.
Have topics you'd like to see covered in PCR Insider? Contact the editor at bbutkus [at] genomeweb [.] com.