Proteins lack DNA's copying ability and do not readily undergo amplification, making separation and fractionation more important — especially for small amounts of proteins. And the inherent complexity and diversity of proteins makes a viable protein array an even more difficult goal. But the need to process proteins en masse is so urgent that heroic efforts are under way to develop a workable protein chip.

Leading the field at present are designs based on antibodies tethered to a solid surface. Large Scale Biology in Germantown, Maryland, and Biosite Diagnostics in San Diego, California, are developing an array of antibodies against 2,000–5,000 protein targets from the former's human protein index database. Biosite will use its omniclonal phage display technology to generate high-affinity antibodies against the targets. The companies hope the system will be available in the second half of 2002.

But an inherent drawback of antibody chips — or any protein chip, for that matter — is the destructive effect of proteases that may be lurking in the analyte mixture. “You have to use protease inhibitors if you're sampling microdissected tissue,” says Lance Liotta of the US National Cancer Institute's Center for Cancer Research, who invents tools for proteomics and has surveyed the existing technology. “Process the tissue, lyse it, stain it and pray that these manipulations don't affect the 3D state of the protein.”

Perhaps the biggest challenge is the accurate quantification of low-abundance protein. The faint signal of a protein of interest may easily be swamped by the much higher concentrations of other surrounding proteins.

Ciphergen in Fremont, California, is selling a device that helps scientists to detect low-abundance proteins. The company's chip uses specific surface chemistries to affinity-capture minute quantities of proteins. “A peak in one sample but not the other says a variation exists, but you still have to figure out what it is,” says Mike Baldwin, a chemist at the University of California, San Francisco. “It's an interesting approach, but not mainstream proteomics — at least, not yet.”

Another recent quantitative protein-expression and -identification technique using mass spectrometry is isotope-coded affinity tagging (ICAT), a kind of labelling invented by Ruedi Aebersold at the Institute for Systems Biology in Seattle. The start-up company Sense Proteomic, based in Cambridge, UK, is trying to use smaller numbers of mounted proteins to assay for suspected protein–protein interactions such as those known to play a role in toxicity.

Other chip approaches towards proteomics include atomic-force microscopy, aptamer libraries and biosensors.