Tiny Bubbles
Stephen Quake has the full lips and startling blue eyes of a cherub in a Renaissance painting, and the only sign that he is old enough to teach at a top academic institution is his corona of thinning curls. Wearing an olive-green golf shirt, loosely fitting khaki pants and sandals--the equivalent of formal attire among Caltech’s lab rats--he ushers his visitor into an office where papers are stacked precariously high. The 34-year-old physicist exudes the restless energy of a teenager, and hardly looks like a man at the center--make that epicenter--of a revolution.
Quake whizzed through a bachelor’s in physics and a master’s in math in four years at Stanford University before arriving at Caltech at a mere 27. He was focusing on DNA research but was frustrated by the time-consuming laboratory processes in which huge, expensive machines were required to separate cells for study.
He had a feeling that a new technology called microfluidics might help. Microfluidics can process a sample of liquid thousands of times smaller than a drop of water through minuscule laboratory plumbing composed of hundreds of channels--each about the width of a strand of hair--and mixing chambers the size of a few cells. This mini experiment all takes place on a rubbery chip as small as a pat of butter. Like computer microprocessors, the ultimate goal is to perform hundreds, even thousands, of experiments at the same time on one tiny chip, and thereby accelerate the pace of medical research and testing.
But when it came time to have a company involved in microfluidics actually produce the equipment for Quake’s Caltech operation, “we realized the field wasn’t as far advanced as we thought, and we ended up having to go back and invent a whole new set of technology,” he says, fidgeting in his chair.
So they did. For more than a decade, scientists had been trying to use glass and silicon, the brittle crystal used in computers, to shrink cumbersome laboratory equipment down to one-inch-square chips and perform relatively simple functions, such as snipping strands of DNA. But Quake and his research team found these materials were difficult to work with, expensive and slowly produced. The costly silicon, which needs to be used in a sterile room, was too stiff for making tiny devices, and the valves that controlled the fluid flow had to be sealed with rubber so they wouldn’t leak.
After some trial and error, Quake’s group in 1997 found a refreshingly easy solution. Inspired by Harvard chemist George Whitesides’ soft lithography, a technique used to mold and stamp tiny structures on pliable materials such as silicone, they hit upon the idea of using a piece of silicone--the same inexpensive rubber used to caulk bathtubs and augment breasts--to fashion their own microchips.
For the chips, silicone was poured onto a mold that formed channels in the material, a sort of lab “piping,” and another sheet of silicone was put on top to create an enclosed device. But there were problems. “We really couldn’t use it because it didn’t have any valves, so there was no capacity to turn things on and off,” recalls chemist Marc A. Unger, one of Quake’s first postdoctoral researchers at Caltech. “It’s like having all the showers and faucets in your house running all the time.”
Still, Quake felt he was on to something. In 1999 he co-founded a company, now called Fluidigm, a South San Francisco biotech, to commercialize his microfluidic systems.
Shortly after, Unger and his team came up with an ingenious solution to the valves problem. They made a two-level structure where the channels crossed at right angles, and pumped pressurized water or air through the top layer and fluids through the bottom. When the water on the top pushes down on the lower channels, it closes them, like stepping on a garden hose.
“We slapped the two pieces together and it worked better than I had any right to expect,” Unger says. “Now we had a method of controlling the fluids.”
Quake was doubtful, though, whether the larger scientific community would recognize the significance of this elegantly simple breakthrough. He bet Unger $20 that their paper outlining the discovery wouldn’t be published by the prestigious journal Science. “Sometimes you’re ahead of the curve and sometimes you’re behind it,” Quake says. “I thought we were ahead of the curve.”
Science accepted the paper. Once the daunting technical problems were resolved, he says, “the technology really took off.”
As a child, Stephen Quake witnessed a similar breakthrough in scientific technology, one he thinks will be mirrored by microfluidics. The son of an early software pioneer, Quake grew up in New Canaan, Conn., a bucolic enclave of Yankee affluence, immersed in computer technology. He wrote his first program using a stack of punch cards at age 11, and he earned extra money in high school running a computer programming camp at his parents’ home for the neighborhood kids.
From that vantage, he watched computers evolve from the lumbering mainframes of the 1970s to the lithe, pocket-sized devices of today. He believes microfluidics will follow a similar trajectory. After all, he points out, the transistor was invented in 1946, and the first microprocessor came along quite a while after the transistor and integrated circuit because it took some time for people to figure out what to do with the new technology. It’s not much of a conceptual leap to see the similarities. In the 1950s, battalions of scientists operated vacuum-tube computers that took up entire rooms to do complex mathematical computations. In a similar fashion, today’s biological experiments can be sheer drudgery, often requiring hundreds of steps and costly equipment to decipher genetic structures, isolate crucial chemicals or probe the life cycle of cells.
“Right now, we’re using the equivalent of vacuum tubes to do biomedical research,” Quake says. Microfluidics will profit from the same benefits of miniaturization as microprocessors, he adds, allowing you to do things “faster, cheaper and better.” And perhaps open the door to a day when instant at-home diagnostic tests can pinpoint the exact moment when a cell turns cancerous. Or make it possible for medications to be tailored on the spot for a patient’s individual chemical makeup and genetic quirks. And make possible genetic blueprints as easy to obtain as fingerprints.
That Quake jumped from one field to the top of another with warp speed seems a product of precocious talent coupled with an ability to capitalize on opportunity. After his whirlwind education at Stanford he earned a doctorate in physics from Oxford University, which he attended on a scholarship. He then apprenticed briefly in the Stanford lab of Nobel laureate Steven Chu before arriving at Caltech. “Steve was always sprinkled with stardust,” says Gajus Worthington, a Stanford classmate who is now Quake’s business partner.
Quake has since received a lot of glowing ink from the scientific and financial press, including Popular Science, MIT’s Technology Review and Scientific American (in which Fluidigm was named a 2003 top business leader). “Stephen’s a real visionary who’s done breakthrough research,” says Todd Thorsen, a bioengineer and microfluidics expert at the Massachusetts Institute of Technology in Cambridge, Mass. “Brilliant and driven, Steve is very insightful and knows how to bring different talents together. But what is really key to his success is that he takes a very hands-on approach, and spends a lot of time working side by side at the laboratory bench with his students.”
Venture capitalists looking for the next best thing in science regularly troop through Quake’s sprawling 18-person lab, and rewarded his efforts with a $21-million infusion of capital earlier this year. At the lab--a mini-United Nations, with handpicked grad and postdoctoral students from Bulgaria, Korea, Uganda, Israel and Canada--12-hour days and seven-day workweeks are a given, though colleagues say Quake has mellowed since becoming a father in 2002 and now actually takes off Sundays.
Since producing that first chip in 1999, Quake’s lab has churned out a series of increasingly complex micro-scale chips that are replacing conventional full-size machines. One can sort cells, an essential piece of lab equipment that makes it easier to track the results of each experiment, while another chip can replicate snippets of DNA, giving scientists more genetic material to work with. A third device, called a flux stabilizer, is a prototype for an implanted drug-delivery system that would continuously dispense medication deep inside the body, and eliminate the need for injections or remembering to take pills.
Currently the lab is working on a DNA sequencer on a chip that would replace today’s $300,000 dishwasher-sized machines. Also under development is a method of automating the growing of cell cultures, an essential part of biological research. This would “liberate researchers from the tyranny of pipetting,” says Quake, a time-consuming process akin to feeding a litter of kittens with an eyedropper around the clock.
Yet another system, which is now being sold commercially through his company, grows protein crystals, a critical and technically challenging first step in figuring out a protein’s structure. Proteins are the body’s versatile workhorses, and scientists need to know their structures to diagnose diseases and to develop drugs that precisely target their vulnerabilities. This chip whittles down a $100,000 robotic workstation to a small device that sells for less than half that price. “This technique is just fundamentally better--it grows crystals faster,” says Carl Hansen, a grad student in Quake’s lab who helped devise the protein crystallization system.
But it was the device called the multiplexer, which was showcased in an October 2002 cover story in Science, that catapulted his team ahead of established rivals. Quake and his collaborators unveiled a chip that some experts consider a rudimentary version of a biological microprocessor. The one-inch-square grid is a gleaming triple-layered matrix containing 2,056 microvalves and 256 chambers to mix chemicals, and it can run tens of thousands of chemical reactions every hour.
Now his graduate students can produce a prototype chip virtually overnight. They carve a design on a piece of glass, shine an ultraviolet light on the plate to make the mold, slather on silicone, stick it in a convection oven, then peel it off the glass--and voila. The chips are so easy and inexpensive to make that the lab is littered with rejects, which are given as souvenirs to visitors.
“A lot of people had made chips with single pumps, and valves, but Steve put it all together and made systems that could do really complicated processes,” says David J. Beebe, a microfluidics expert and biomedical engineer at the University of Wisconsin in Madison. “He is really the leader in taking this silicone technology and matching it with appropriate applications, and demonstrating that this could work. Before, microfluidics was limited to electrical engineers who knew how to use silicon and glass. But his work using more flexible materials really opened up the field.”
Not everyone is convinced that microfluidics will usher in such sweeping changes.
“The field is in its infancy,” cautions J. Michael Ramsey, a scientist at the Oak Ridge National Laboratory in Tennessee. “My gut feeling is that there is pay dirt out there. It’ll be interesting to see what happens 10 to 20 years down the road--whether it advances at an exponential pace, like microelectronics, or fizzles out.”
The big limitation with chips made out of silicone is that they can only use water to flow through the channels. The soft rubber can’t tolerate the harsh solvents or chemical reagents used in some experiments. Quake’s group has “built some of the most complex circuits, but it’s not clear what they’re useful for yet,” Ramsey says. “If they can overcome the solvent obstacle, it would be a significant breakthrough.”
This isn’t just sour grapes from an upstaged rival: Ramsey is the godfather of the field. In the late 1980s, he pioneered the concept of microfluidics and the notion of miniaturizing the scale of laboratory experiments. Initially his ideas were dismissed by skeptics who thought they were impossible to execute. He eventually proved the doubters wrong by figuring out how to overcome the technical obstacles and constructing minuscule networks of pipes using glass and silicon.
“We thought about using plastics back then,” he says. “Making inexpensive devices seemed like the way to go. But there’s a whole number of experiments that are chemically incompatible with this material, and we preferred not to paint ourselves in a corner.”
By the mid-’90s, Ramsey says, “the field was getting traction.” More academic researchers had jumped in and spun off biotechs to commercialize their new inventions.
In the years since, a handful of companies such as Fluidigm, Caliper Technologies in Mountain View and i-STAT Corp. in New Jersey have devised viable products that are used for biomedical research, and in the case of i-STAT, for doing blood tests at a patient’s bedside.
But the field is undergoing a shakeout. The technology was over-hyped to the business community, and scientists failed to hit the kind of home runs that generated enough profits to justify the investment of millions in seed money. Some of the original companies are struggling or going out of business, while others are settling into more modest niches.
Still, Fluidigm “seems to be doing everything right,” says Marlene Bourne, a senior analyst with In-Stat/MDR, a market research firm in Scottsdale, Ariz. “With any emerging technology, you have the initial promise of what it can do. But the first-, second- or even third-generation devices aren’t even going to come close to capturing that promise. It’s a matter of evolution, and it’s going to take a little more time for the market to really embrace microfluidics.”
Microfluidics, like the kitchen appliances that liberated the housewife, has already removed some of the drudgery from biological testing. But where it can go next moves from the mundane to the more speculative. Take, for example, the battle against cancer. In a scenario that microfluidics might make possible in our lifetime, a woman with a history of breast cancer in her family could give herself a weekly blood test, smearing a blood sample on the surface of a portable device about the size of a credit card.
Should she see the telltale signs of pre-cancerous cells, she could then visit her doctor, who would take a quick DNA blueprint using a microprocessor developed by a microfluidics company. Using that information, he could then use the same microprocessor to concoct a cocktail of DNA fragments specifically programmed to latch onto her cancer cells, thereby preventing them from reproducing. The customized medicine would be injected into her bloodstream using the flux stabilizer, the microscopic drip-irrigation system powered by blood pressure that dispenses drugs around the clock.
Because such small quantities of liquids are used, scientists for the first time could study single cells. This would enable them to track the exact moment when cells in the body turn cancerous, rather than detect tumors years after they’ve gained a deadly foothold and are far more difficult to eradicate. The implications are breathtaking, but Quake takes it in stride.
“It’s humbling to be on the ground floor of the development of what could be a lifesaving technology,” says Quake. “It’s also been tremendously exciting to have some ideas, and to be in a position to see these ideas realized in an entirely new technology.
“At this point, I’m pretty confident we can do anything we want in the biological area. We’ve got enough pieces in place, and now it’s just a question of putting the pieces together to solve interesting problems.”