These protein markers, called PSA for prostate-specific antigen, are found at elevated levels in the blood of men with prostate cancer, which is the number two killer of men in this country.

"The technique is sensitive enough to detect levels 20 times lower than the clinically relevant threshold," said lead author Arun Majumdar, professor of mechanical engineering at the University of California, Berkeley. "This is currently as good as and potentially better than the ELISA assay, which is the standard today for detecting protein markers like PSA."

The microcantilever technique has far broader applications, however. Any disease, ranging from breast cancer to AIDS, characterized by protein or DNA markers in blood or urine could conceivably be assayed by arrays of these microcantilevers. A microcantilever array would be one of the first "protein chips," analogous to the DNA chip used broadly today in research labs and the biotechnology industry to conduct hundreds of DNA analyses simultaneously.

"This offers the possibility of a common platform for high-throughput detection of proteins, DNA and/or RNA, in areas ranging from disease diagnosis to drug discovery," said Majumdar, a member of UC Berkeley's Health Sciences Initiative. "This could lead to fast screening and molecular profiling for many diseases and a possible cancer chip for detecting cancer."

"A big advantage of this technology is that one could look at multiple markers in a single reaction, whereas currently available assays require a separate reaction for each analyte," said colleague Richard J. Cote, MD, professor of pathology and urology at the Keck School of Medicine of the University of Southern California and the USC/Norris Comprehensive Cancer Center. "So the cost of performing a cantilever assay as opposed to a typical ELISA assay is potentially much, much lower."

Another advantage this technique has over current assays such as ELISA - enzyme-linked immunosorbent assay - is that there is no need to attach fluorescent tags to molecules. This could prove very useful in testing how well candidate drugs bind to their targets, since drugs typically are so small that attaching fluorescent labels interferes with their binding to a protein or enzyme.

Cote notes, in particular, that the technique is ideal for detecting single base-pair variations in DNA, the most common form of biological diversity. Thanks to the human genome project, scientists are finding more and more of these single nucleotide variations - called single nucleotide polymorphisms, or SNPs - and expect them to have major significance in biomedical research.

"From a discovery point of view, this is a very, very important advance," Cote said.

Majumdar and his colleagues, including UC Berkeley graduate student Guanghua Wu; Cote and Ram H. Datar, PhD, of USC in Los Angeles; and Karolyn M. Hansen and Thomas Thundat of Oak Ridge National Laboratory in Tennessee, report their findings in the Sept. 1 issue of Nature Biotechnology.

Wu fabricated the cantilevers from silicon nitride using techniques identical to those employed by the semiconductor industry to make microprocessors. He worked closely with his Oak Ridge
colleagues, who perfected a way to coat the top surfaces of the cantilevers with antibodies. When proteins bind to these antibodies, they elbow one another apart and force the lever to bend downward. The cantilevers can also be coated with single-strand DNA for binding to complementary DNA.

The higher the concentration of the protein or DNA being measured, the greater the deflection of the cantilever, so that the chips not only detect the presence of the protein, but also its amount. Majumdar and his colleagues measured the deflection with a laser.

The cantilevers themselves are about 50 microns wide - half the width of a human hair - 200 microns long (a fifth of a millimeter) and half a micron thick. When molecules bind to the surface, the cantilever moves only about 10-20 nanometers - the diameter of 100-200 hydrogen atoms. Lasers can detect a deflection as small as a fraction of a nanometer, however. That is the equivalent of a cantilever the length of a football field bending the mere width of a quarter, Thundat said.

"The primary advantage of the microcantilever method originates from its sensitivity, based on the ability to detect cantilever motion with sub-nanometer precision, as well as the ease with which it may be fabricated into a multi-element sensor array," he said. "No other sensor technology offers such versatility."

Majumdar and his UC Berkeley colleagues have found a way to put several hundred cantilevers onto a single silicon chip, and have developed a way to measure the deflection of all simultaneously with a single low-power laser or light emitting diode.

"It's not trivial to go from one cantilever to hundreds of them on a chip, a millimeter apart, detecting hundreds of different biomolecules," Majumdar said. "But that's what we need to do low-cost, high-throughput, label-free assays."

Thundat first reported in 1994 that MEMS (microelectomechanical systems) cantilevers bend when molecules glom onto them. Following similar reports in 1999 by researchers in Germany and Switzerland, various groups began experimenting with different proteins and DNA to see how the cantilevers perform and whether they could be used to detect many different chemicals and biological molecules.

What had been missing was an explanation for the phenomenon and proof that the deflection is a quantitative measure of the amount of a specific protein that binds to the cantilever.

"This present study represents the first and most quantitative investigation of a clinical application," Thundat said.

Majumdar and his colleagues explained the origin of cantilever deflection in a February 13, 2001, paper in the Proceedings of the National Academy of Sciences, describing why and how the binding of antigen and antibody or two complementary strands of DNA alters the surface stress and bends the cantilever.

The current paper shows that cantilever deflection accurately reflects the amount of protein binding. Majumdar said that, in practice, three or four cantilevers will have to be used in parallel for each molecule detected, so that a comparison can be made between the deflection of bound and unbound cantilevers. This is necessary because local conditions make the microscopic levers flutter, bend and sway.

Nevertheless, cheap arrays of cantilevers would allow enough "controls" to make for accurate assays.

"In just two years, this has gone from a possible way of detecting biological reactions to a highly sensitive method for detecting PSA, comparable to the best assays currently available," Cote said. "In a very short period of time, we've really made enormous advances. I'd be surprised if this doesn't become a viable assay system within the next three to five years."