There is growing evidence that small clusters of a protein called amyloid beta are what cause brain damage in Alzheimer's disease.
Scientists at the annual meeting of the Society for Neuroscience in Washington, D.C., say clusters of just two or three amyloid beta protein molecules can damage synapses, the connections between brain cells.
New research also shows that plaques of amyloid beta — much larger groupings made up of millions of cells — are not nearly as harmful as the small clusters.
Until the past couple of years, amyloid beta plaques were considered the most likely cause of Alzheimer's brain damage.
But Dennis Selkoe, a professor of neurology at Harvard Medical School, said the new findings suggest that hypothesis is wrong.
He presented research using various forms of amyloid beta protein taken from the brains of people with Alzheimer's disease. The proteins were injected into the brains of healthy rats.
Selkoe said entire plaques did not affect synapses. Neither did single proteins.
"But you could inject the doublets and triplets into the brain of a healthy rat and show that the rat got forgetful," Selkoe said.
Selkoe's results were bolstered by a Swedish study that found high levels of even larger clusters in the brains of people with Alzheimer's disease who had died.
Frida Ekholm Pettersson of Uppsala University in Sweden says her team found no clusters in healthy brains.
Also at the meeting, researchers from Duke University presented new evidence that it's possible to reverse Alzheimer's disease by removing amyloid beta protein from the brain.
Donna Wilcock described an experiment using a new mouse model of Alzheimer's created by a colleague at Duke. The mice develop a memory problem very much the way people with Alzheimer's do.
But when the mice received a vaccine against amyloid beta, their memory improved.
Wilcock says the approach may work because it causes the body's immune system to attack all forms of amyloid beta — including small clusters.
Scientists are trying a similar approach in people.