By Keay Davidson
Illustration by Terry Allen
A few miles from the garages and office buildings where the microchip revolution began, Angela Wu is studying leukemia in mouse cells.
A generation ago, she would have used the traditional tools of laboratory research such as microscopes, test tubes and petri dishes.
Nowadays, though, the 24-year-old Stanford graduate student of bioengineering is using what some see as biology’s equivalent of a microchip: a microfluidic chip. She’s learning how to shrink a lab onto a rubbery, translucent chip the size of a postage stamp. If it works, this “lab on a chip” will automatically conduct a complex experiment on her cell samples — all within a surface area of a few square inches. Via traditional means, the same experiment would have consumed much more time, sample material, lab space, personnel and money.
In recent years, these chips have begun living up to their potential: “We are just now seeing the first wave of applications, after a decade of basic research in the field,” says Wu’s advisor, Stephen Quake, PhD, an international star of microfluidics, specializing in minute plumbing projects of many sorts. In the biomedical world, “big science” is becoming “little science.” Microfluidic chips are quietly revolutionizing biomedical research — and Stanford is a leader of the revolution.
“We think of these chips as the integrated circuit for biology,” says Quake, a professor of bioengineering. He expects them to transform life much as microelectronics chips have since the 1960s. With a microfluidic chip, one or a few researchers can achieve what once would have required a small army of workers.
“That,” says Quake, “is a major effort in our lab: trying to do big science with skeleton crews” — and it’s no small consideration as the global financial crisis forces the world’s scientists to pinch pennies. “It is a shame to waste money,” he says. He’s already used the chip as a tiny production plant, manufacturing — directly on the chip — a radiopharmaceutical tracer used in PET (positron emission tomography) scanners.
Futuristic though it seems, in one sense microfluidics is old-fashioned. There’s no laboratory “division of labor” here: The scientist is also an instrument maker and lab technician. Wu can design, manufacture and test her chips all by herself — a skill she learned in Quake’s 3-year-old microfluidics foundry in the basement of the James H. Clark Center.
A typical microfluidic chip is made of two or three layers of the silicon-based organic polymer polydimethylsiloxane. It’s covered with the microscopic equivalent of test tubes, pipettes, petri dishes, sieves, filters, reaction chambers — all linked by elegant meandering and sometimes zigzagging highways of microchannels that resemble diamond-cut etchings on glass.
Valves and gates, akin to the locks in a canal network, control fluid flow. When changing fluid pressures in a region of one layer cause it to protrude into a channel in an adjacent layer, the valve closes. When fluid pressure causes it to withdraw, a gate opens. By regulating flow, the chip carries out experiments.
Chip-making is as much of an art as a science: “It takes a couple of tries to get everything right,” Wu admits with a smile. “It’s almost like baking — some recipes require more care than others.” The foundry’s clean room resembles those in Silicon Valley microchip factories, with workers wearing masks and white suits. And for good reason: Dust and other contaminants damage and obstruct the chip’s channels (thinner than a human hair). “We could make these devices in the open lab,” recalls Quake. “But only about 40 percent of the devices actually worked.”
Researchers design their chips on a computer screen. “For example,” Wu explains, pointing at an image on her computer screen that resembles a schematic, multicolored map of a subway system, “if you want to draw a channel, you just draw a rectangle and then join it to another rectangle.” Chip design requires patience and care. In the clean room, “when you’re building the structures on the silicon wafer, you wash off the parts that you don’t want. Early on, I made small mistakes, like doing a wash for maybe a few seconds too long — I washed off the parts that I wanted to keep. So I had to start over.” A chip designer eventually learns certain rules of thumb. For example, don’t make the microchannel ceiling too low (it might collapse) or put adjacent channels too close together (they might leak into each other). “It can be a pretty delicate process.”
The physics rules of the macro scale — the everyday experiential world of birds and trees, cars and coffee houses — turn weird at the micro scale as one approaches the subatomic and quantum realms. Somewhat similarly, microfluidics research confronts scientists with surprising phenomena, observes Juan Santiago, PhD, associate professor of mechanical engineering and a microfluidics investigator since the mid-1990s. At the macro scale, a fluid flows easily in a large, smooth pipe, but inside a micro-pipe it flows sluggishly, like syrup. The same fluid doesn’t easily become turbulent: hence different ingredients don’t mix easily. Disruptive “shock waves” form at the intersection of a microchannel and a much narrower “nanochannel.”
Researchers have struggled to find better ways to force liquids through the tiny pipes ever since microfluidics was pioneered in the early 1990s. Air pressure is one way. Another involves electrical-driven flow. Each method has its drawbacks. To overcome such problems, scientists design channels as cunningly as possible to encourage whatever fluid behaviors are needed for their planned experiments. For example, they etch zigzag channels into the chip, which encourage fluid mixing.
Sometimes it helps to think counterintuitively. While at MIT, Martin Bazant, PhD, who recently spent a sabbatical at Stanford, developed a way to speed microfluid flow. He did so by placing electrodes in the way of the flow, rather than keeping the microchannel clear. One would expect the electrodes, as obstacles, to block the flow. Instead, they generate eddy currents that hasten it.
Meanwhile, practical applications of microfluidics are emerging. Many of them involve teams dominated by Stanford researchers, especially (judging by his numerous publications) the ubiquitous Quake. Among the field’s first fruits:
Working last year with Stanford gastroenterologist and virologist Jeffrey Glenn, MD, PhD, Quake and colleagues made the first microfluidic-based discovery of a new drug treatment — a new use for an old drug. Using microfluidic chips to analyze 1,200 candidate drugs, the scientists identified a once-popular anti-itching powder, clemizole hydrochloride, as a potential weapon against hepatitis C. “There would have been no way that I would have predicted” that clemizole would be singled out, says Glenn, a professor of gastroenterology and hepatology and co-senior author with Quake of the article on the work, in the September 2008 Nature Biotechnology. A better treatment for hepatitis C — which afflicts 150 million people worldwide — is badly needed, partly because of the serious drawbacks of traditional treatments such as interferon and ribavirin. “Many patients cannot tolerate them,” says Glenn, who is director of Stanford’s Center for Hepatitis and Liver Tissue Engineering. “Most people taking these treatments feel like they have the flu all the time. And if they can tolerate them, the drugs still don’t get rid of all of the virus in many patients.”
Says Quake: “There’s been great interest in using microfluidics for drug discovery, but this is the first time it has actually happened. It would have taken years to do this same thing before — not even practical. The whole point is these microfluidics tools, which operate on such small volumes and such small amounts of material, let you do things that you can’t otherwise do.”Detecting Down Syndrome
Popular news media tend to overuse the phrase “medical breakthrough.” Still, it’s hard to think of a better description for the microfluidics-based method that speeds detection of the gene for Down syndrome in pregnant women. Normally in such cases, an expectant mother’s cells are cultured in dishes for two weeks. But a microfluidics chip greatly accelerates the process of detecting the extra chromosome 21 associated with the disorder. Stanford obstetrics and gynecology fellow Yair Blumenfeld, MD, reported this in January at the annual meeting of the Society for Maternal-Fetal Medicine, confirming Stanford researchers’ published findings. Blumenfeld, Quake and colleagues used a commercially available chip manufactured by a microfluidics-chip firm, Fluidigm, cofounded by Quake in 1999. The chip requires far smaller samples than are normally needed for mass prenatal testing — “thousands of times less, in nanoliters,” says lead author Christina Fan, a graduate student of bioengineering in Quake’s lab. The technique detects the extra chromosome 21 within a few hours, compared with the two-week wait required by amniocentesis, a traditional prenatal test.
Our mouths are full of bacteria — they’re microbial jungles with upward of 700 species of micro-organisms. Quake and colleagues, including professor of infectious diseases David Relman, MD, have shown how a microfluidic chip they developed can help resolve mysteries about the nature of many of these microbes, which are difficult to grow in the large quantities normally needed for study. The scientists used the method to genotype rod-like bacteria from human tooth plaque. Their microfluidic chip enabled the team to steer single bacteria to chambers where the cells were broken open, then their individual genomes were “amplified,” in other words mass-copied. “Kind of like xeroxing them,” Quake says, who reported the work in the July 17, 2007, Proceedings of the National Academy of Sciences. They linked the bacteria to the so-called TM7 phylum of oral microbiota and identified more than 1,000 genes, most of which “are only distantly related to genes found in other organisms.” Thanks to such future genetic analyses using microfluidic chips, the authors forecast, “a much richer tapestry of microbial evolution will emerge.”
Word is out that microfluidics can enable seemingly impossible feats. So it’s no surprise that a microfluidics laboratory will be included in Stanford’s new, 200,000-square-foot Lorry I. Lokey Stem Cell Research Building, scheduled for completion in mid-2010. “Stem cells are quite rare,” Quake explains, “and therefore there has been quite a bit of interest in using the phenomenal sensitivity and small-volume control of microfluidics to analyze them.”
Microfluidics investigator Santiago is developing portable microfluidic-chip devices to detect toxic industrial chemicals and chemical weapons. “We hope to make a toxin detector that you can carry in your pocket,” he says. DARPA is funding the research for use by soldiers. But he thinks civilians — for example, environmental watch groups — also might use the devices, perhaps to monitor stream quality.
And Paul Blainey, PhD, a postdoctoral scholar in Quake’s lab, plans to use the chips to collect and study microbial samples from the wild. A possible collection site is the hot springs at Yellowstone National Park. Yellowstone “has a zillion little hot springs, each of which has its own distinct chemistry and microbial populations,” he says. “Nature has evolved all these genetic sequences, yet we’ve only scratched the surface of the diversity that’s out there. This is the 21st-century version of going into the rain forest and discovering new types of mammals.”
The advent of microfluidics could be truly game-changing for science and medicine. Imagine this futuristic scenario: A parent wakes up at night with a sick child. Should she race the child to the hospital? To find out, she places a drop of the child’s saliva on a microfluidic chip in a handheld device. The user interface lets her know if she can stay home — or if that rush to the doctor is needed.
In the same way, medics working far from advanced medical facilities could more confidently diagnose their patients’ ailments. And conceivably, a chip implanted in a patient’s body could monitor the bloodstream for warning signs — say, of dangerous fluctuations in insulin levels.
The potential that microfluidics holds for improving lives drives Quake to push deeper in the micro-realm. Along the way, he has started two companies, Fluidigm and Helicos Biosciences, which makes a high-speed DNA sequencer. Such entrepreneurship is vital to ensure that good ideas don’t remain trapped on paper, he says. Trained as a physicist, he’s comfortable tinkering with gadgets, but many bioscientists aren’t. “It’s not the culture of the biological field to build things,” he says. “Most biologists won’t use something unless they can buy it in a box. To see your technology get used, you can’t just publish it, you have to find ways to make sure it is commercialized.”
But it’s more than such practical concerns that draw these researchers to work with the chips. There’s another motive — one that’s hard to argue with. They find the chips irresistibly cool, and sometimes downright beautiful.
When colored dyes run through a chip’s channels, it resembles a Mondrian-style abstract painting, Quake says. “I love looking at them,” he admits.