By Robert M. Frederickson
February 10, 2003 | NEARLY A CENTURY ago, the engineers of macroscopically assembled systems found themselves facing a problem commonly referred to as the "tyranny of numbers" — the practical limit to the complexity of assembled systems. Nowhere was this problem more evident than in the design of intricate electronic circuits, which were composed of multiple bulky, power-hungry vacuum tubes. The invention of the transistor provided a smaller, less expensive alternative to tubes, but it meant having to connect increasingly complex circuits comprising hundreds or thousands of transistors.
Then came a revolutionary solution in the late 1950s: the invention of the integrated circuit. All of its components — capacitors, transistors, resistors, interconnections — were created in situ, obviating any need for assembly.
The tyranny of numbers problem has arisen again of late with the advent of biochips — effectively, miniature laboratories on glass slides or microchips. Most biological assays are performed in solution, and the challenge has been to develop integrated devices that can perform complex biological assays in a massively parallel, miniaturized format using microfluidic technology.
The impetus for these new tools is the desire of researchers to boost throughput and to reduce costs of expensive or rare substrates and reagents. Once more, assembly is an acute problem, for as biochips become closer in size to cells and molecules, so too must the channels, pumps, and valves that manipulate the solutions of reagents and substrates.
San Francisco-based Fluidigm Corp. has come up with a solution to this problem by co-opting the concepts behind the integrated circuit: It fabricates the pumps and valves directly into the microfluidic circuitry of the biochips. Using a process known as multilayer soft lithography, Fluidigm spins out a soft silicone elastomer — a pliable, gas- permeable, rubbery material — onto a computer-designed mold. The mold is a flat surface with raised channels that is created with traditional photolithography. Once cured, two or more layers can be fused and placed on a glass slide to create a 3-D chip. A sample layer contains channels through which substrates and reagents flow, under the influence of pressure changes. (A demonstration of Fluidigm's chip design can be seen at www.fluidigm.com/tech.htm.)
The first killer application of this technology, according to Fluidigm CEO Gajus Worthington, is the Topaz chip for protein crystallization. "We had a unique technology platform," Worthington explains, "but we needed a product that both addressed a real problem in biology and that would be difficult for competitors to replicate."
With so many genomes now sequenced, the task at hand for both drug developers and academic scientists is deciphering the structures of myriad gene products, which is both costly and cumbersome.
|Micromixer: The Topaz chip integrates channels, pumps, and valves for handling fluids.|
"We have developed a microprocessor that crystallizes proteins in situ, using one-hundredth of the sample used by current macroscale techniques and making use of integrated microfluidics," Worthington says. The device exploits a process known as free interface diffusion, which involves exposing a protein sample to a concentration continuum of a crystallization reagent. The physics of the chip duplicates this process by controlling the mixing of the components with simple diffusion, enabling the user to sample a wide spectrum of crystallization conditions using minute amounts of material.
Several proteins resistant to more conventional techniques have already been crystallized, according to University of California at Berkeley's James Berger, a co-inventor of the chip and a Fluidigm advisor. "A key advantage afforded by this approach is the ability to tackle proteins and protein complexes that simply don't exist in the quantities needed for more traditional techniques," Berger says. "Just as significant, however, is the ease of use of the Topaz setup, which should bring the technology of protein crystallization out of the domain of the experts and into the hands of a greater number of biochemists."
The analogies with semiconductor chips are not just physical, however, but conceptual as well. In a recent publication in Science (thebigone.caltech.edu/quake/publications/ scienceoct02.pdf), California Institute of Technology investigators Stephen Quake and Todd Thorsen described using the Fluidigm technology to create a microfluidic processor chip. The key is the microfluidic multiplexor — essentially, a combinatorial array of binary valve patterns that enables complex fluid manipulations with a minimum of elements. A prototype "massive sample partitioning microprocessor" has already been designed that can partition a complex sample into as many as 80,000 spatially addressable and retrievable parts. Applications include genetic testing and assays for proteins, biomolecules, and even infectious agents. "This platform has huge implications for chronically underfunded fields ... by making diagnostics cheaper and portable," Thorsen says.
Quake and Worthington say they're convinced they have yet to scratch the surface of the potential of their microfluidic technology. "We foresee fabricating devices with up to 100,000 functioning valves before not too long," Worthington says.
At that level of integration, they envision the elimination of robots in many clinical and research labs. "We plan to end the tyranny of pipetting," Quake says.
Robert M. Frederickson is a science writer based in Seattle. He can be reached via e-mail at email@example.com.