Individual polymerase molecules immobilized within ZMWs in an aluminum substrate generate detectable pulses as fluorescently labeled bases associate with the enzyme before their addition to the nascent DNA chain. Credit: Pacific Biosciences

Scientists would be hard-pressed to engineer a reagent for DNA synthesis that rivals the speed and efficiency of DNA polymerase, and a platform that fully exploits this enzyme's potential as a driver for fast and accurate real-time sequencing would be a powerful asset. However, developing such a system has posed a daunting technical challenge, requiring the capability to directly monitor the activity of individual enzymes without impeding their function.

Steve Turner and Jonas Korlach first started collaborating to tackle this task over a decade ago, as graduate students at Cornell University. Korlach had initially approached Turner, hoping to benefit from his expertise in nanostructure design, to develop a system for imaging individual polymerases. “I was originally interested in trying to find a way to see these machines in real time and understand their dynamics,” recalls Korlach. “And then very quickly, we realized that if you could do that and identify which base is incorporated at any time, then you could have a potentially very powerful DNA sequencing technology.”

Now, Korlach and Turner—who subsequently founded a company, Pacific Biosciences—have transformed their thought exercise into a working instrument. Their system makes use of nanostructures called zero-mode waveguides (ZMWs), tiny wells in which individual polymerase molecules have been immobilized in fixed orientation. These ZMWs establish small volumes in which each enzyme can be readily imaged with minimal background noise in the presence of appropriately high concentrations of fluorescently tagged nucleotides.

These latter also represent an important advance: conventionally labeled nucleotides generally impede polymerase processing or terminate chain synthesis, so Turner and Korlach's team developed nucleotides labeled at the terminal phosphate, which is released as bases are appended to the nascent DNA chain.

In the final system, each ZMW contains a polymerase molecule, synthesizing a complementary chain to the template molecule. As each labeled nucleotide enters the polymerase, it generates a transient fluorescent pulse that becomes visible in the ZMW, then vanishes once the new base is attached. The entire process is simultaneously monitored in every ZMW in real time with a specially designed multiplexed confocal imaging instrument.

Despite the many technical challenges that needed to be overcome, the instrument performs strongly, delivering accurate real-time sequences for both circular and linear templates at a rate of 2–4 bases per second. Although errors are a routine problem for individual reads, the multiplexed nature of the system makes it simple to achieve an accurate consensus, and the team is continuing to optimize the platform. “Whereas in the paper we required 15-fold coverage for a consensus with 99.3% accuracy, we've now gotten a consensus of 99.97% with 11-fold coverage,” says Turner, “and this is still improving.” Notably, the system can also achieve longer individual reads than existing sequencing-by-synthesis methods—up to 4 kilobases, in this demonstration.

The authors anticipate that their first-generation commercial instrument—slated for release in 2010—will dramatically streamline the sequencing process but believe that future versions will be able to tackle a host of other applications as well. “If in half an hour you can exhaustively sequence all the nucleic acids in a sample, you can emulate a microarray platform or a quantitative PCR or other things,” says Turner. “We plan on reducing a whole series of problems in biology to essentially a software problem.”