EVENTS | VIEW CALENDAR
PCR turns 30 (part two)
(To return to part one of this special report, click here)
Increasing standards and pushing the envelope
Until recently, the standard for measuring DNA changes has been qPCR, which is a comparative technology, where you have to normalize your answers according to known standards, and you have to have multiple replicates in order to accurately confirm your input as well, points out Dr. Michael Samuels, the principal research scientist and scientific liaison at RainDance Technologies in Billerica, Mass.
"There is also a theoretical limit of a twofold change, representing a single PCR cycle threshold," Samuels notes.
Digital PCR removes the limitations seen in qPCR by using the power of partitions. No longer are interesting genes compared to a known standard, but rather, they are given absolute values. When an assay has been optimized, a single well can provide absolute quantitation in total copies or copies per microliter. By partitioning into droplets, rare events of interest can have the chance to be found, copies can be counted accurately, very small amounts of virus can be detected, cancer may be found by looking for tumor markers in blood—and much more. Researchers are able to multiplex up to 10 genes at the same time by simply using varying concentrations of just two probes in a single well. The partitions remove most of the competition, allowing the target genes to be amplified and visualized on a 2D plot. Here are applications RainDance lists where dPCR has been successfully demonstrated:
Digital PCR measurements are performed by dividing the sample and qPCR assay mixture into a very large number of separate small-volume reactions, such that there is either no or one target molecule present in any individual reaction. This is the fundamental concept for making "digital" measurements, RainDrop's Samuels points out.
When operating in the digital range (where all reactions contain either no or one target molecule), it is possible to multiplex qPCR assays without concern for competition or cross reactivity, as each single-molecule target-containing reaction will proceed with the target binding to its specific primers and probe, whereas no amplification will occur in reactions without targets. Having each single-target molecule in a separate reaction allows both highly abundant and low-abundant targets to be counted in the same experiment without concern for "swamping out" the low abundant target (since each separate reaction has at most one target, independent of its concentration in the average sample volume).
RainDance Technologies offers the RainDrop dPCR System for digital PCR, utilizing the company's patented picodroplet technology to generate up to 10 million droplets in a single well. In approximately 30 minutes' time in eight separate wells, 80 million droplets can be created. This allows researchers to create up to one billion droplets per day, with each droplet encapsulating a single molecule. With a standard detection capability of one mutant in 250,000 wild-type genes and a dynamic range of six logs, the RainDrop dPCR System can accurately quantitate extremely rare alleles, such as circulating tumor biomarkers like IDH1 for glioma patients, or KRAS for colorectal cancer patients.
"Researchers who desire to push the envelope," Samuels says, "and go for even rarer mutants will find that the RainDrop dPCR System has a lower limit of detection of one mutant in one million."
"Not only does the RainDrop dPCR System provide unrivaled sensitivity," Samuels states, "it allows researchers to go beyond the typical multiplexing capabilities of qPCR. Using two probes with varying intensities, researchers are able to get up to a 10-plex assay done in a single sample. Researchers can find multiple SNPs, housekeeping genes and other potential genes of interest in a single well, unleashing true digital potential."
PCR: The tough part
In his Noble Prize acceptance speech, Dr. Kary Banks Mullis apologized to his audience for the complexity of the story explaining his invention. In his words (edited for length):
Cetus hired me in the fall of 1979. Within two years, there was a machine in my lab from Biosearch of San Rafael, Calif., turning out oligonucleotides much faster than the molecular biologists at Cetus could use them. I started playing with the oligonucleotides to find out what they could do.
I started thinking about doing some experiments wherein an oligonucleotide hybridized to a specific site could be extended by DNA polymerase in the presence of only dideoxynucleoside triphosphates. I reasoned that if one of the dideoxynucleoside triphosphates in each of four aliquots of a reaction was radioactive, then an analysis of the aliquots on a gel could indicate which of the dideoxynucleoside triphosphates had added to the hybridized oligonucleotide, and therefore which base was adjacent to the three prime end of the oligonucleotide. It would be like doing Sanger sequencing at a single-base pair.
On human DNA, it would not have worked, because the oligonucleotide would not have specifically bound to a single site. On a DNA as complex as human DNA, it would have bound to hundreds or thousands of sites depending on the sequence involved and the conditions used. What I needed to make this work was some method of raising the relative concentration of the specific site of interest (emphasis added). What I needed was PCR, but I had not considered that possibility. I knew the difference numerically between 5,000 base pairs as in a plasmid, and 3 billion base pairs as in the human genome, but somehow, it didn't strike me as sharply as it should have. My ignorance served me well. I kept on thinking about my experiment without realizing that it would never work. And it turned into PCR.
One Friday night, I was driving, as was my custom, from Berkeley up to Mendocino, where I had a cabin far away from everything off in the woods. I was thinking. Since oligonucleotides were not that hard to make anymore, wouldn't it be simple enough to put two of them into the reaction, instead of only one such that one of them would bind to the upper strand, and the other to the lower strand with their three prime ends adjacent to the opposing bases of the base pair in question? If one were made longer than the other, then their single-base extension products could be separated on a gel, and one could act as a control for the other. What I would hope to see is that one of them would pick up one radioactive nucleotide, and the other would pick up its complement. Other combinations would indicate that something had gone wrong. It was not a perfect control, but it would not require a lot of effort. It was about to lead me to PCR.
Encouraged by my progress, I continued to think about it, and about things that could possibly go wrong. What if there were deoxynucleoside triphosphates in the DNA sample, for instance? What would happen? I needed a way to insure that the sample was free from contamination from deoxynucleoside triphosphates. I could treat the sample before the extension reaction with bacterial alkaline phosphatases (BAP). The enzyme would degrade any triphosphates present down to nucleosides, which would not interfere with the main reaction, but then I would need to 'deactivate the phosphatase' before adding the dideoxynucleoside triphosphates—and everyone knew at that time that BAP, as we called it, was not irreversibly denaturable by heat. The reason we knew this was that the renaturation of heat-denatured BAP had been demonstrated in classic experiments that had shown that a protein's shape was dictated by its sequence. In the classical experiments, the renaturation had been performed in a buffer containing lots of zinc. What had not occurred to me or apparently many others was that BAP could be irreversibly denatured if zinc was omitted from the buffer, and that zinc was not necessary in the buffer if the enzyme was only going to be used for a short time and had its own tightly bound zinc to begin with. There was a product on the market at the time called matBAP, wherein the enzyme was attached to an insoluble matrix, which could be filtered out of a solution after it had been used. The product sold because people were of the impression that you could not irreversibly denature BAP. We'd all heard about, but not read, the classic papers.
This says something about the arbitrary way that many scientific facts get established—but for this story, its only importance is that, had I known then that BAP could be heat-denatured irreversibly, I may have missed PCR. As it was, I decided against using BAP, and tried to think of another way to get rid of deoxynucleoside triphosphates. How about this, I thought? What if I leave out the radioactive dideoxynucleoside triphosphates, mix the DNA sample with the oligonucleotides, drop in the polymerase and wait? The polymerase should use up all the deoxynucleoside triphosphates by adding them to the hybridized oligonucleotides. After this was complete, I could heat the mixture, causing the extended oligonucleotides to be removed from the target, then cool the mixture, allowing new, unextended oligonucleotides to hybridize. The extended oligonucleotides would be far outnumbered by the vast excess of unextended oligonucleotides and therefore would not rehybridize to the target to any great extent. Then I would add the dideoxynucleoside triphosphate mixtures, and another aliquot of polymerase. And now, things would work.
But what if the oligonucleotides in the original extension reaction had been extended so far, they could now hybridize to unextended oligonucleotides of the opposite polarity in this second round? The sequence which they had been extended into would permit that. What would happen?
EUREKA! The result would be exactly the same only the signal strength would be doubled.
EUREKA, again! I could do it intentionally, adding my own deoxynucleoside triphosphates.
And again, EUREKA! I could do it over and over again. Every time I did it, I would double the signal.
For those of you who got lost, we're back! I stopped the car at mile-marker 46.7 on Hwy. 128. In the glove compartment, I found some paper and a pen. I confirmed that two to the 10th power was about 1,000, and that two to the 20th power was about a million, and that two to the 30th power was around a billion, close to the number of base pairs in the human genome.
I drove on down the road. In about a mile, it occurred to me that the oligonucleotides could be placed at some arbitrary distance from each other, not just flanking a base pair, and that I could make an arbitrarily large number of copies of any sequence I chose—and what's more, most of the copies after a few cycles would be the same size. That size would be up to me. They would look like restriction fragments on a gel. I stopped the car again.
I had solved the most annoying problems in DNA chemistry in a single lightning bolt: abundance and distinction. With two oligonucleotides, DNA polymerase and the four nucleosidetriphosphates, I could make as much of a DNA sequence as I wanted, and I could make it on a fragment of a specific size that I could distinguish easily.
I celebrated my victory with Fred Faloona, a young mathematician and a wizard of many talents whom I had hired as a technician. Fred had helped me that afternoon set up this first successful PCR reaction, and I stopped by his house on the way home. As he had learned all the biochemistry he knew directly from me, he wasn't certain whether or not to believe me when I informed him that we had just changed the rules in molecular biology.
'Okay, Doc, if you say so.'
—From Dr. Kary Banks Mullis' Nobel Lecture: The Polymerase Chain Reaction, 1993
(To return to part one of this special report, click here)