“Where PCR is really going,” says Olivier Harismendy at the University of California, San Diego, “is parallelization and miniaturization.” Indeed, researchers are making use of a wide variety of materials and applications toward these goals.

Miniaturizing PCR protocols offers a range of benefits, says Bruce Gale, who directs the Utah State Center of Excellence for Biomedical Microfluidics. “As soon as you go to microfluidics, you can bump up the speed and precision because you use small volumes,” he says. Other advantages include portability plus the ability to work with smaller samples and fewer reagents. Growing enthusiasm for shrinking volumes is being felt in the industry, says Jeremy Gillespie, group product manager at Thermo Fisher Scientific. Scientists are increasingly interested in buying smaller amounts of reagents because they are using smaller volumes in their assays, he says.

Most microfluidics chips conduct PCR using minute volumes of highly dilute samples, relying on fluorescence (or melting-point temperatures) to determine the presence and quantities of a genetic sequence in a sample. Many applications involve finding some version of a needle inside some version of a haystack: rare mutations among lots of wild-type DNA or fetal DNA in a maternal blood sample, for example. Some applications look at relative amounts of DNA sequences, such as copy-number variation, allele ratios or microbe sampling. Newer PCR applications use microfluidics techniques as a preparative technology for next-generation sequencing. These sequester DNA in tiny reaction vessels for PCR amplification, then collect the amplicons for subsequent analysis.

The Access Array chip from Fluidigm can amplify 48 genomic regions from 48 samples. Credit: Fluidigm Corporation

Microfluidics for next-generation sequencing

With each run of a sequencer potentially generating data for 100 gigabases, “you've got to think about how to design experiments so as not to waste data, because it's costly,” says Daniel J. Turner, head of sequencing technology development at the Wellcome Trust Sanger Institute. “The two things that affect how much sequencing you have to do to see what you need to see are how specific your sequence enrichment is and how uniform your data are,” explains Turner.

Sequencing capacity is wasted when some genomic regions are amplified more than others. “When you say you want 20-fold, you want 20-fold everywhere,” says Harismendy. “If you want to sequence all the exons at 20-fold coverage, you don't want some at 100-fold or at fivefold,” he adds. If some exons in a study are only present at fivefold coverage, then an analysis can only claim fivefold coverage. Compared with other technologies for massively parallel PCR, he says, microfluidics platforms excel at uniformity because the products of PCRs do not compete with each other.

Fluidigm launched one such microfluidic platform, its Access Array, in the fall of 2009. The chip uses a matrix that guides samples and reagents into tiny chambers. “You take 48 samples and 48 primer pairs, and it makes every combination and does PCR, and then you can pump the samples back out,” explains Ken Livak, technology developer at Fluidigm Corporation. During the PCR a DNA barcode is added to each sample so that the 2,304 amplified products can be pooled and shuttled into any next-generation sequencer. Separate chips can provide unique sets of barcodes so that PCR products from multiple chips can be pooled and sequenced together. The Access Array Chip excels in producing uniform numbers of the desired amplicons, says Livak. “You get a 'tight' range so that you don't have to go way deep [into the sequence] without missing anything,” he adds.

The stand-alone thermocycler from Fluidigm conducts PCRs on a microfluidics chip. Credit: Fluidigm Corporation

Johan den Dunnen heads the Genome Technology Center at Leiden University Medical Center in the Netherlands, where his team uses Fluidigm's products as well as an approach called hybridization capture (Box 1). The larger or more complex the targeted regions are, he says, the more likely he is to go with hybridization, but the Fluidigm PCR approach offers several advantages. Even without any optimization, he says, it has “nice uniform enrichment” compared to hybridization techniques. Plus, with hybridization capture, researchers only find out that targeted regions were missed after a sequencing run has been completed. In contrast, Fluidigm's arrays can show whether a target region was enriched before sequencing, allowing his team to decide whether or not to run a sample.

RainDance Technologies has commercialized an instrument that uses microdroplets to efficiently prepare samples for next-generation sequencing. The core of RainDance's technology is the creation and combination of two large sets of tiny, precisely aliquoted drops, each about the size of a typical eukaryotic cell. In one set, each drop contains a specific pair of PCR primers designed to target a region of interest. In the other set, each drop contains a large portion of genomic DNA to be amplified along with the enzymes and other reagents necessary for PCR. A specially designed chip merges the drops at a rate of about 3,000 per second, all the while ensuring that the drops combine in a constant one to one ratio.

Harismendy recently completed a collaboration with RainDance showing how the technology can be used for target capture. He and colleagues reported using this approach to examine 435 exons in samples from six people1. They chose regions that included repetitive elements, varying amounts of G+C content and other sequence features that often cause variation in amplification. The results showed accuracy and coverage equivalent to what they would have obtained if they ran all the PCRs in 1.5 million separate 20-microliter tubes, but they used a fraction of the reagents and disposables. “It was so obvious that it was a big savings and it's also much easier,” says Harismendy, who is not affiliated with RainDance. “There's little handling. You just put your [primer] library on one side and your template [DNA] on the other side,” he adds. The droplets for each sample are then transferred into a single PCR tube that is placed into a thermocycler to allow PCRs to run their course; after a sufficient number of amplification cycles, the oil-water emulsion that keeps droplets separate is broken and barcodes can be ligated to PCR products, which are then ready to go into any of the next-generation sequencers.

Samples loaded into the Access Array Integrated Fluidic Circuit are amplified, barcoded and pumped out again for analysis. Credit: Fluidigm Corporation

An advantage of microfluidics systems is that they can work reliably with whole genome–amplified DNA without becoming biased to certain alleles. This could be particularly useful for researchers who may only have tiny amounts of starting DNA, says Darren R. Link, a co-founder of RainDance. “They can do the [whole-genome amplification], use RainDance technology and still have sample remaining,” he says. RainDance's platform currently allows about 4,000 primer pairs to be used at one time, but improved efficiencies to be launched this summer will expand the instrument's range to around 20,000 primer pairs.

Nonetheless, researchers like Wellcome's Turner are currently using multiplexing rather than microfluidics approaches because the former are capable of pulling down tens of thousands of genomics regions at once (Table 1). When RainDance expands its primer capacity in the summer of 2010, the technology will start to become competitive with multiplexing, Turner says, though he thinks the projects he is running will likely require coverage of many more regions. He is quick to point out that the decision about which preparative technology works best depends on the numbers of samples as well as the number and type of sequences to be studied.

Table 1 Preparative DNA technologies for next-generation sequencing
RainDance creates tiny droplets as vessels for PCR amplification. Credit: RainDance Technologies

Little solutions

Still, there is more to genotyping than sequencing. In addition to its gene expression products, Fluidigm already offers microfluidics chips that can analyze copy-number variations and single-nucleotide polymorphisms. Both analyses run on the BioMark instruments also used for gene expression analyses. These chips have an optically clear top through which fluorescence signals corresponding to DNA content can be read. Because the chip is designed to interface with standard laboratory equipment, researchers can load the chip with whatever primers and probes they want. “The chip works like a microtiter plate,” explains Robert Jones, executive vice president of research and development at Fluidigm. “You can buy assays from a number of vendors, you can put them in the part of the chip just like you would a plate, you put the samples in, and the system does all the PCR,” says Jones.

RainDance plans to launch several additional applications for its microdroplets: deep resequencing and methylation analysis and, in the future, single cell analysis. Other technologists are also creating applications based on the ability to make droplets with precisely controlled volumes. Researchers led by Richard Mathies at University of California, Berkeley recently reported a microfluidic technique that can be used to detect one dangerous Escherichia coli O157 pathogen in a background of 100,000 harmless bacterial cells2. First, forward primers specific to each cell type to be analyzed are attached to thousands of 34-micrometer beads, so that every bead can bind DNA from the cell types expected in the sample. These beads, along with the cells to be analyzed, are diluted into nanoliter droplets produced by a microfabricated emulsion generator array. Each droplet contains PCR reagents, including reverse primers labeled with a unique fluorescent dye for each cell type. Researchers can readily control the creation, loading and transport of intact droplets into standard tubes and the thermocycling used for PCR, explains lead author Yong Zeng. After a sufficient number of PCR cycles, the beads are analyzed by flow cytometry. The fluorescence signal allows researchers to track the distribution of the rare cells as well as count the droplets containing no, one or multiple cells.

Most of these 'digital PCR' applications rely on having just one analyte per sample and thus require excessive dilutions, but the ability to make droplets with extremely precise volumes circumvents that requirement. As all the droplets are the same size, explains coauthor Richard Novak, statistical analyses can be used to calculate absolute numbers and frequencies of rare cells in a population. Novak and Zeng believe perhaps as many as 20 separate fluorophores could be used in future versions of this technique, allowing the detection of rare mutations in clinical samples and perhaps even the co-occurrence of mutations in a single cell.

RDT 1000 instrument conducts massively parallel PCRs in picoliter droplets. Credit: RainDance Technologies

It is too early to consider commercialization of this device, says Zeng, but the droplet protocol should translate readily to other laboratories, provided they have access to microfabrication facilities. “The current limitation for translating microfluidics to biological labs is the cost of fabrication and the expertise.”

Much of the cost of a microfluidics chip depends on sophisticated lithography used to create nanoscale pumps and valves. This means that the chips can simultaneously conduct dozens of assays on dozens of samples in little time, so companies can charge hundreds of dollars for a chip. Gale, however, hopes to take microfluidics to the other end of the market. This year he reported the use of small plastic disposable disks for certain types of genotyping3. Each disk contains a thousand nanoliter-sized wells that dilute material from a single sample. The sample is brought into chambers not with the standard pumps and valves, but by spinning the disk on its axis, filling wells by centrifugal force. In a prototype, Gale's research team showed each PCR cycle could be completed in about half a minute. The amount of amplification in the wells is detected by fluorescence and indicates how much of a particular DNA sequence a sample contained. Analysis took just over half an hour, including image analysis. Though the device is still being optimized, Gale imagines that versions of it could be used to find rare cell types or mutations in both environmental samples and bodily fluids. Too young to even have a website, start-up company Espira aims to commercialize these disks. The goal is to pair a $20 camera with PCR chips that cost just a few pennies.

A microfabricated emulsion generator array from the Richard Mathies lab at the University of California, Berkeley uses PCR to find rare cells. Credit: Yong Zeng

The scope of other PCR-based applications is tremendous, Gale says. The hope is that the advance of new technologies can work with microfluidic PCR amplifications in much the same way as it has worked in sequencing. In other words, as capacity goes up, costs and size come down. See Table 2

Table 2 Suppliers Guide: Companies offering products for PCR, target enrichment and next-generation sequencing analysis