SCIENCE
Sorting cells The tiny world of microfluidics still has some big hurdles, and sorting cells is one of them. Tried-and-true methods such as centrifugation and flow cytometry are not easily reduced, so many researchers have turned to dielectrophoresis, which sorts cells by moving them through a nonuniform ac electric field. Now, Xiao-Bo Wang and colleagues at the University of Texas' M. D. Anderson Cancer Center have taken dielectrophoretic (DEP) separation to new heights. In this issue of Analytical Chemistry (p. 911)) the researchers report a new DEP/gravitational field-flow fractionation (DEP/G-FFF) technique that separates cells using levitation. The researchers use a narrow chamber (0.42-mm high by 25-mm wide by 300-mm long) that has interdigitated microelectrodes along the bottom. The chamber is filled with a homogeneous solution, and the cells are loaded into the inlet region. When an electric field is applied, the DEP force generated by the electrodes pushes the cells up through the solution, levitating them. At the same time, gravity is pulling the cells down, and they settle at the point where the two forces balance. Different kinds of cells rise to different heights because the equilibrium point depends on the dielectrophoretic properties of the cells. Unlike electrophoresis, dielectrophoresis does not require the cells to be charged to be moved. Instead, the applied electric field induces polarization charges. Depending on the relative polarizabilities of the cells, they move toward strong or weak regions in the electric field and, in this case they eventually levitate to different heights. This dielectrophoretic separation is enhanced when a fluid flow is started in the chamber, because fluid drag makes the liquid at the bottom of the chamber move more slowly—the same way that water near the bottom of a river is slower than water closer to the surface. Because the liquid farther from the bottom of the chamber moves more quickly, it exits sooner. (This is true up to the half-height of the chamber.) That is field-flow fractionation. II nractical terms, it means that the cells that rise higher in the chamber are eluted earlier. Wang and colleagues used the technique to separate MDA-435 breast cancer cells from normal blood cells. Efficient separation of the two cell types could be achieved in ~5 min, with the cancer cells
eluting in the first peak, indicating a higher equilibrium height. This difference was partially explained by the cancer cells' lighter density, but these cells also had a larger dielectric factor (a DFI .), and, unlike the blood cells, an otnFP that varied with Schematic drawing of DEP/G-FFF. The cells levitate to the point the frequency of the where the DEP levitation forces (FDEPJ balance with the electric field. sedimentation force (Fgmv). Because the velocity further from the The researchers bottom (VFFF2) is faster, cells levitated higher are eluted sooner. (This is true up to the half-height of the chamber.) believe the differences in dielectric properties of the cells are related to the membrane erwise, all of the cells would migrate tocharacteristics. "We think various cell ward the same region or would not migrate types have different membrane capaciat all. Similarly, the dielectrophoretic affintances because the cell surfaces have disity method—in which the desired cells similar morphologies and membrane comstick to electrodes while the other cells are positions," Wang says. 'That is why differwashed away—requires large differences ent cells can be separated." to ensure that the cells do not all stick to the electrodes or get washed away. Because the separation depends on the physical properties of the cells, the techDEP/G-FFF has the opposite problem, nique is noninvasive, says Wang. In consays Wang. "Our method is extremely sentrast, techniques such as magneticsitive," he says, "and because of that, we activated cell sorting (MACS) and antibod} fractionate cells of the same type because binding require either adding magnetic the dielectric properties are not identical particles or binding to receptors. "Antibodcell to cell." Although this was initially a ies may activate cells or otherwise modify criticism of the group's work, Wang says them," he says. "By separating the cells, they are now looking more closely at such you may be changing them." cells to see if there is a genetic or biological basis for the fractionation. DEP/G-FFF is also more general than such techniques, says Wang. It does not At this point, the system can sort —500 require specific antibodies, which are not cells per second—about a factor of 10 always available. DEP/G-FFF also has inslower than a high-performance flow cyherent flexibility because cell separation tometer, says Wang. "But I think this numdepends on electric forces—voltage and ber can increase by a factor of 10-100 by frequency—that are easily controlled and scaling up the system and refining the automated, he says. And, unlike flow cytechnology," he adds. tometry, which can sort two to three popuWang also expects elution times to imlations of cells, DEP/G-FFF can sort many prove. Miniaturizing the technology will populations, he adds. reduce the sample volume, he says, so even if the sorting rate remains the same, Other dielectrophoretic methods share some of the advantages of DEP/G-FFF, but the elution times should drop to 1-2 min. That is important for developing "point-ofWang says the new technique is more senservice" diagnostic systems, which is the sitive. "With our method, the differences ir the dielectric properties can be very small," long-term goal, he says. he says. 'The other methods require big The focus now is on detecting metastatic differences in the dielectric properties." In cells in the bloodstream, Wang says. "The dielectrophoretic migration, he explains, problem there is mainly one of sensitivity some cells migrate toward the strong rebecause there are very few cancer cells in a gion of the field while others migrate toblood sample," he says. "But we think we ward the weak region. The differences in may be able to do that in the long run." dielectric properties have to be large; othElizabeth Zubritsky Analytical Chemistry News & Features, March 1, 1999 171 A
