A novel approach to the fabled "magic bullet" has been invented and has proven itself unusually effective at killing cancer cells in a test tube, medical researchers reported in last week's Science.

A magic bullet -- a term coined by Paul Ehrlich, the German medical pioneer who died in 1915 -- is an ideal, and perhaps impossible, drug or other therapeutic agent that goes straight to its target in the body, sparing healthy tissues.

The bullet in this case is the "killer T cell," a class of normal immune-system cells that drift in the blood and can detect and kill cancer cells. Killer cells produce enzymes that eat holes in the cancer cells' membranes, causing them to leak to death.

Many cancer researchers suspect that cells turn cancerous frequently but that constant surveillance by killer cells normally blocks their development into tumors. Tumors arise from the rare cells that escape detection.

Scientists at Harvard Medical School and the Massachusetts Institute of Technology have come up with a way to improve the killer cell's ability to find cancer cells.

It consists of a laboratory-made molecule with two parts. One side of the molecule can bind to specific receptors on the outer surface of killer cells, while the other binds to another kind of receptor on the cancer cell. A killer cell that acquired the molecule on its surface would gain a chemical affinity for a cancer cell.

When cancer cells were put in a test tube with plain killer cells, the killers failed to kill them. But when the intermediary two-part molecule was added, it helped bring the two kinds of cells close enough for the killers to destroy the cancer cells.

Similar approaches have been tried before, but they used intermediary molecules tailored to bind to receptors that exist on most, but not all, cancer cells. The cancer cells that escape killer cells, however, may be those that lack this receptor. Thus the Harvard-MIT group made their intermediary molecule to bind to a receptor for a growth hormone, something all fast-growing cells -- such as cancer cells -- should have.

The research was done by Margaret Ann Liu of Harvard and MIT, Samuel R. Nussbaum of Harvard and Herman N. Eisen of MIT.