"Our primary goal is to take control of the cell surface," says Bertozzi. To do that, she says, "we appropriate the cell's natural metabolic machinery for assembling tailor-made oligosaccharides." All cell surfaces are decorated with oligosaccharides—complex structures strung together inside the cell from a few simple sugars. Chemically unique, specific to different cells, different stages of development, and different environments, each oligosaccharide imparts to the cell a unique surface for interaction with the outside world.

Bertozzi reasoned that if a properly designed synthetic sugar with novel chemical properties was ingested by the cell, the sugar might be incorporated in an oligosaccharide and delivered to the surface, giving the cell new surface properties. Bertozzi, graduate student Lara Mahal, and postdoctoral fellow Kevin Yarema engineered an unnatural sugar related to sialic acid, which in its natural form is often found in the oligosaccharides of human cells.

"We chose an analogue of sialic acid that carries an unnatural functional group, the ketone group," says Bertozzi. "We hoped that if the cells ate the analogue's precursor—without noticing, so to speak—they would install it in oligosaccharides and decorate themselves with these unnatural markers." Although the ketone group isn't normally found on cell surfaces, it isn't harmful to the cell, and it reacts strongly with other functional groups such as the hydrazide group. Bertozzi's team hoped this reactivity would give ketone-labeled cells a selective affinity for materials that had been outfitted with the hydrazide group, such as ceramics, organic thin films, and metals.

When fed with the artificial precursor, cells manufactured oligosaccharides with ketones and expressed them in copious amounts—over a million copies on the surfaces of most cells. The researchers found they could precisely control the degree of ketone labeling by adjusting the relative amounts of natural and unnatural precursors they fed to the cells.

While Bertozzi, Mahal, and Yarema were developing the technology, they realized that "many human cancer cells, including colon, breast, and prostate cancers and certain leukemias, have aberrant patterns of oligosaccharides," Bertozzi says. "For one thing, they show extremely high levels of sialic acid. The possibilities were obvious."

Bertozzi and her colleagues showed that ketone-labeled cancer cells, otherwise robust, could be made uniquely vulnerable to a derivative of the natural plant toxin ricin, if it was armed with the reactive hydrazide group.

"It worked," says Bertozzi. "We killed 'em."