Microfluidics

SILICON VALLEY MEETS BIOLOGY

Chow's comparison to the computer industry is apt, as the initial idea for microfluidics came from integrated circuit technology. Thirty or so years ago, Stanford University researchers began applying semiconductor fabrication techniques to bioanalytical devices to improve their sensitivity, producing a miniature gas chromatograph in 1979.¹ But, this instrument chilled rather than stimulated the emerging field. "It didn't work very well," says Ramsey.

"It instigated a lot of thought, but it also put up a barrier," he explains; so much so that when he and others, nearly 20 years later, proposed putting fluids through microchannels, they encountered resistance. "I tried this idea out on some of my colleagues and they kind of snickered at me," Ramsey recalls. His first grant proposal was rejected, but by 1992 he obtained funding, and two years later produced a capillary electrophoresis chip.

Though Ramsey's patent searches turned up nothing resembling his work, he wasn't alone in the field. Others, including Andreas Manz and Jed Harrison at Ciba-Geigy in Basel, Switzerland, were also working on the idea, as was Allen Northrup, of Pleasanton, Calif.-based Microfluidic Systems.

Northrup was using microreaction chambers produced at the University of California, Berkeley's state-of-the-art fabrication facility, when "it dawned on me that the volumes in PCR are in microliters," he says. He went to Cetus, where PCR was largely invented, and proposed the idea of putting the reaction on a microfluidics device. "They were pretty excited about the idea," he recalls. Collaborations ensued between Berkeley, Cetus, and Lawrence Livermore National Laboratory, where Northrup did his postdoctoral work. Launching the technology was mostly a matter of "establishing collaborations with people on the leading edge," he says.

DOLLARS AND SENSE

Once these pioneers had laid the initial groundwork, the field developed rapidly. Commercialization followed closely with the founding of Caliper in 1995. According to Chow, the company chose to start with capillary electrophoresis ("low-hanging fruit," as she calls it), for faster time-to-market and earlier revenue potential. The strategy seems to have paid off: Caliper today is the only lab-on-a-chip company with significant revenues from product sales rather than grants and collaborations, according to Marlene Bourne, an industry analyst with EmTech Research, Ann Arbor, Mich.

Chris Luft is one of four assay developers at Amphora Discovery Corp. of Research Triangle Park, NC, a company whose scientific strategy hinges on Caliper technology. Together, these four scientists have developed 150 protease, kinase, and phosphatase assays for drug screening in only two years, due to the speed and data quality microfluidics provides, he says. Amphora researchers can test an entire 130,000-compound library in a week, generally using just one vial of enzyme, for $5,000 to $7,000 per screen. "For a minimal amount, you can get a bucket load of information," he points out.

But those savings must be balanced against the cost of the hardware. Amphora has 24 Caliper instruments, which cost nearly half a million dollars each. The chips themselves cost on the order of $100 apiece. Chow says this cost will drop as more companies adopt the technology, predicting economies of scale. "It's a chicken and egg phenomenon," she notes. "Most new products are expensive [inhibiting adoption], but you must get to high volume to make them cheaper."

In the meantime, "just putting a Caliper machine in your lab is not a guarantee of success," Luft says. Amphora's engineers worked for nearly a year to optimize the instruments' operating parameters and software to achieve the necessary data quality.

Several companies have followed Caliper's lead, placing other bioanalytical separation techniques onto microfluidics chips. Nanostream of Pasadena, Calif., and Eksigent of Livermore, Calif., have developed chip-based high-performance liquid chromatography (HPLC) systems, primarily for use in adsorption, metabolism, and toxicology studies of drug candidates. Palo Alto-based Agilent Technologies has introduced an integrated HPLC-mass spectroscopy chip for proteomics applications. And a Dutch company named Concept to Volume, has developed a new miniaturized gas chromatograph for remote, rapid detection of specific biothreats.

Also pursuing "homeland security" applications are Microfluidic Systems, which is developing automated PCR systems for forensics and biothreat detection, and Cepheid of Sunnyvale, Calif. "Extremely fast PCR really requires microfluidics," says Northrup, owing to the unique physical properties of very small volumes. Cepheid makes PCR-based threat-detection devices installed at mail processing facilities around the United States. But the company also has its sights set on clinical diagnostics, hoping to develop chips that can distinguish among different types of cancers or infectious agents.

Meanwhile, Biotrove of Woburn, Mass., has developed a high-density, PCR-based chip for genotyping and RNA analysis, designed for use in pharmacogenomics. Costing $300 to $350 each, these chips use 64-fold less polymerase and three- to four-times less sample than conventional methods, says chief technology officer Colin Brenan, and are currently being used for biomarker identification at Johnson & Johnson Pharmaceutical Research and Development in Rahway, NJ.

THE NEXT GENERATION

Though borne of the semiconductor industry, the industry's traditional materials of glass and silicon are not always the best choice for lab-on-a-chip devices. Fluid valves require a soft seat to seal, like the rubber washer in a faucet. In 1998, Harvard University chemist George Whitesides introduced "soft lithography," a technique for fabricating microchannels in rubber-like materials such as polydimethylsiloxane. Besides providing better seals, these soft materials can be fabricated faster, more cheaply, and in smaller dimensions than their silicon and glass counterparts.

Steve Quake, a California Institute of Technology bioengineer, refined Whitesides' method by introducing "multilayer soft lithography," which enables design of very small micromechanical valves. Quake patented his method in 1999 and founded Fluidigm. Since then, company researchers have designed chips with ever-higher valve densities, like the dynamic array chips Radich is testing. The newest chip allows testing of 48 samples under 48 different reaction conditions, for a total 2,304 experiments. Such an experiment would be highly impractical using microplates, says Marc Unger, a Fluidigm scientist.

Such massive parallellization will also have an impact on the production of drug candidates, says Jensen, who is developing microfluidics-based chemical reactors at MIT. Jensen's chips allow difficult, expensive syntheses to be tried under many conditions in parallel, on an otherwise impractical scale, while their use of very tiny volumes improves safety and reduces hazardous waste. "It puts the chemist back into the role of discovery scientist rather than spending a lot of time synthesizing chemicals. It has the potential to revolutionize drug discovery," he says.

And it makes once-unthinkable experiments possible. Ramsey, for instance, is using labs-on-a-chip to profile single cells. Cell populations that appear uniform on a macroscopic scale can be quite heterogeneous, confounding analysis of signaling pathways, for example. He and Nancy Allbritton of University of California, Irvine, are collaborating on high-throughput chips that can analyze the individual contents of hundreds of cells per minute.² Among other applications, such chips could facilitate detection of extremely rare abnormal cells for cancer diagnosis.

But beyond the benefits of microscopic scale and massive parallelization, researchers point to integration as a key asset of future lab-on-a-chip devices. "My guess is that it will very much follow the way the MEMS field is going ... more integration of multiple technologies," says David Beebe, a biomedical engineer at the University of Wisconsin-Madison. Caliper's Chow agrees. "It is the integration of different processes on one platform that has made integrated circuits so powerful." Such processing power will be needed to take full advantage of genomics and offer personalized medicine in the future, she says. Amphora founder, Bill Janzen, adds: "We are in the forefront of a new paradigm for drug discovery."

Microfluidics Milestones