Method Special: New trends in digital PCR

All or Nothing
by Steven Buckingham, Labtimes 01/2017

The concept of digital PCR is not new – it actually goes back to the early nineties, when different labs experimented with the then “single molecule” or “limiting dilution PCR”. But digital PCR has really taken off in recent years, with the advent of droplet-based digital PCR techniques and platforms.

PCR has been with us for over 20 years and when talking about the impact it has had on biology, it is difficult to avoid superlatives. It is not too much of an exaggeration to say that it has changed everything – not only what we do but the way, in which we think about living structures.

But has PCR changed much itself in the 20 years or so it has been with us? On the face of it, it would be surprising if something so powerful and so versatile should not have changed. Actually, the fundamental process underpinning PCR has remained the same – mixing together a template with the polymerase enzyme and the four nucleotides found in nature, then cycling the cocktail between three temperatures: one to denature the DNA into its two strands, one to anneal primers onto the template and one for the polymerase enzyme to extend along the template. But what has changed, is the introduction of new technical methods of running the reaction and ever more imaginative applications. Sometimes, these changes have taken place through a series of leaps, sometimes by gradual, slow evolution.

Researchers at the Southern California Coastal Water Research Project are testing the prototype of a newly-developed portable droplet digital PCR instrument, designed for field studies. Photo: SCCWRP

As for the leaps, the biggest step changes can be summarised in two words: ‘digital’ and ‘droplet’, the two components in the hottest field of PCR today – droplet digital PCR (ddPCR).

Femtolitre droplets for fast cycles

Let's start with the droplet part first. Back in your undergraduate days, you probably did your first PCR reaction in 1.5 ml tubes, dropping the tubes one-by-one into a Thermocycler (or dipping them from one water bath to another, if you are “of a certain age”) and waiting for the reaction to finish. But doing PCR in tiny droplets makes a lot more sense than doing it in tubes. For one thing, by cycling your mix as femtolitre-sized droplets, you can get really fast temperature shifts.

That means you can run through PCR cycles much more quickly. And what is more, the small size of the droplets means that temperatures are more uniform throughout the reaction. Droplets also mean that good mixing comes for free. At that scale, diffusion is fast enough to do the job.

So, how have researchers been doing that? Nearly all methods of droplet PCR have involved creating an emulsion of droplets of the aqueous reaction mix, dispersed in oil. This lends itself nicely to microfluidics, where the droplets can be created, sorted, merged or split. In 2009, Kelly Frazer and team at the University of California, San Diego, published a paper describing a microfluidic device that simultaneously amplified 4,000 sequences (Nature Biotechnology 27: 1025).

Chaoyong James Yang's team at Xiamen University, China, found that you can embed a PCR reaction in agarose droplets, using an emulsion generator that created 500 droplets per second. The PCR reaction takes place with the agarose mix in the liquid phase but when the reaction is over, you can just cool the beads, so they solidify and then store them for later analysis (Lab Chip, 10: 2841-43). Other labs have taken advantage of the fact that, using droplets in a microfluidic device, you can inject liquid at any time into water-in-oil droplets. The progress of the reaction can be monitored using fluorescent markers and lasers can be used to heat the mix (Opt Express 17: 218). Another alternative to droplets in oil, is to print the mix onto an oil-covered surface using a microfluidic robot (Scientific Reports 5, 9551).

So much for droplets, where does digital come in?

Digital PCR arose to meet the need for better quantitative PCR. On the face of it, you would not expect PCR to be very good for quantifying the amount of DNA in a sample. It is, after all, a chain reaction – the DNA gets amplified in an explosive way as the reaction proceeds. The amount of product increases exponentially (at least in the main part of the reaction). How much of a specific sequence is there, when you have finished the reaction, is not a good guide to how much was there in the first place. So, quantitative PCR (qPCR) relies on watching for when the exponential phase starts and estimating roughly when it hits its mid range. The idea is that the exponential phase only really takes off when the product reaches a certain density and it would take rarer molecules longer to get to that point. But qPCR is hard to do well and, because it relies on making a measurement of a runaway process, it is liable to many errors.

Digital PCR takes a completely different approach. Here, the sample is diluted to a low concentration and then aliquoted out into separate reactions. If you get the dilution just right, the vast majority of reactions will contain zero or just one DNA molecule. To a first approximation, the number of DNA molecules will follow a Poisson distribution. You then amplify up each separate reaction mix and you should end up with an all-or-none signal.

Digital PCR has a number of advantages that spring directly from the fact that only one molecule is in the starter mix. First of all, the quantification step is all-or-none. At the end of the reaction, there is either a lot of DNA or none at all. Hence the name, ‘digital’. You work out the starting concentration backwards, using the terms of the Poisson distribution, the volume of the reaction and the dilution factor.

But it is when you put digital and droplet together that some truly amazing possibilities open up. By doing so, the accuracy of quantification, offered by digital PCR, combines with the superior reaction conditions and ease of handling offered by droplets. No wonder, then, there has been an exponential growth in papers citing the technique.

Microfluidic control of ddPCR

Running PCR in droplets, whether it is digital or otherwise, has another huge advantage – it makes it easier to do it using microfluidics. Sure, you can do PCR without droplets. Just guide the reaction slowly through a maze of microfluidic capillaries that pass through the different temperatures needed to drive the reaction. But this method, known as ‘single phase’ microfluidic PCR, has some big drawbacks. First, there is continual contact with the vessel wall. That means the danger of material getting adsorbed to the surface, causing all kinds of contamination mischief. Dragging along the vessel wall also means slowing the progress of the fluid, which means you get uneven progress through the vessels and possibly uneven temperature changes.

Microfluidic control of droplet PCR solves a lot of the problems that come with single-phase approaches. First, there is no contact between the mix and the vessel wall. Quite the opposite – each reaction runs in its own self-contained environment. That means you can do different things with different droplets.

Several labs have published details of microfluidic chambers that allow droplets to be herded, merged, split and even re-injected with fresh material in the middle of the reaction. For instance, Yonghao Zhang's group in the Centre for Microfluidics and Microsystems Modelling at Cheshire, UK's CCLRC Daresbury Laboratory, describes a chip that sits on top of two heaters. The fluidic channel zig-zags between the two heaters, so that the fluid, and the droplets it bears, is rapidly alternated between the two temperatures (Microfluidics and Nanofluidics 3 (5): 611-21).

Meanwhile, whilst ddPCR is a revolution in the technique, there is some quiet evolution going on, as well.

Benny Chain is happy that a lot of routine PCR has been taken over by robots. Photo: Weizmann UK

For one thing, Robotics has steadily been taking over the hard work of PCR. Whether it is droplets or eppendorfs, digital or analogue, anyone who has done a lot of PCR knows how tedious it can be. “One of the biggest changes in the world of PCR is that a lot of routine PCR is being run by robots,” comments Benny Chain, Professor of Immunology at University College London. “Putting lots of small amounts into little tubes is tedious, to say the least.”

Robots are making a big impact on Chain's lab. “It is a big game-changer for us. We are interested in getting a snapshot of the adaptive immune response by sequencing all the receptors we can get hold of in a sample. We have to work out the sequences of literally millions of slightly different DNA strands, to give us an overview of what is going on in the immune system. We'll need robotic machines even more in the future. We want to look at the immune response to lung cancer and that means gearing up to sequencing thousands of samples. Doing PCR on that scale with pipettes and eppendorfs would be tiresome and boring in the extreme.”

There have also been evolutionary changes in the performance of PCR equipment.

“We did a thorough analysis of the PCR market”, Chain says, “and the outcome is clear – aside from the advances in ddPCR and the like, it is a story of steady, incremental progress. Thermocyclers get faster, better, smaller and cheaper.”

Smaller and smarter instruments

How cheap? Well, how does US$ 600 sound to you? That is the current price of ‘openPCR’ (, the PCR machine you can build yourself in three hours, using a couple of tools (so the website claims). And it works – so the website also claims. Chain is thinking of using it as a sort of team-building exercise in his lab.

According to Chain, PCR has followed two general trends. “There is, on the one hand, a move to [the] more robust, cheap and lightweight PCR and, on the other hand, a move towards scaling up with high-throughput, fully-automated systems. Scaling up and scaling down.”

In the coming years, we can expect to see ddPCR applied in ever more imaginative ways, and the confluence of ddPCR with microfluidics providing more flexible and higher turnover applications. At the same time, we will see the development of smaller, more robust systems for point-of-care or in-field testing.

ddPCR means a lot of reactions can be done on small amounts of starting material – and done very quickly. This will greatly expand the uses PCR will be put to. It is very powerful at detecting very low frequency sequences, so will prove powerful in detecting low levels of pathogens or rare ­alleles. It also performs better than other PCR techniques in measuring differences in copy number variants.

The ease, with which a large number of reactions can be performed in parallel and very quickly, means that systematically testing the performance of gene-editing tools, such as CRISPR, is now possible, where it can be used to hone in on the optimal experimental conditions for editing and for checking for off-site effects, two problems that have dogged these exciting advances.

Last Changed: 11.02.2017