WASHINGTON – A doughnut created in a lab and made of silk on the outside and collagen gel where the jelly ought to be can mimic the basic function of brain tissue, scientists have found.

Bioengineers produced a kind of rudimentary gray matter and white matter in a dish, along with rat neurons that signaled one another across the doughnut’s center. When the scientists dropped weights on the material to simulate traumatic injury, the neurons in the three-dimensional brain model emitted chemical and electrical signals similar to those in the brains of injured animals.

It is the first time scientists have been able to so closely imitate brain function in the laboratory, experts said.

If researchers can replicate it with human neurons and enhance it to reflect other neurological functions, it could be used for studying how disease, trauma and medical treatments affect the brain – without the expense and ethical challenges of clinical trials on people.

“In terms of mechanical similarity to the brain, it’s a pretty good mimic,” said James J. Hickman, a professor of nanoscience technology at the University of Central Florida, who was not involved in the research.

“They’ve been able to repeat the highest level of function of neurons. It’s the best model I’ve seen,” he said.

The research, led by David Kaplan, the chairman of the biomedical engineering department at Tufts University, and published Monday in the journal PNAS, is the latest example of biomedical engineering being used to make realistic models of organs such as the heart, lungs and liver.

Most studies of human brain development rely on animals or on slices of brains taken after death; both are useful but have limitations.

Brain models have been mostly two-dimensional or made with neurons grown in a three-dimensional gel, said Rosemarie Hunziker, program director of tissue engineering and biomaterial at the National Institute of Biomedical Imaging and Bioengineering, which funded Kaplan’s research.

None of those systems replicate the brain’s gray or white matter, or how neurons communicate, Hunziker said.

“Even if you get cells to live in there, they don’t do much,” she said.

“It is spectacularly difficult to do this with the brain.”

Kaplan’s team found that a spongy silk material coated with a positively charged polymer could culture rat neurons, a stand-in for gray matter. By itself, though, the silk material did not encourage neurons to produce axons, branches that transmit electrical pulses to other neurons.

The researchers formed the silk material into a doughnut and added collagen gel to the center. Axons grew from the ring through the gel – the white matter substitute – and sent signals to neurons across the circle.

They got “these neurons talking to each other,” Hunziker said. “No one’s really shown that before.”

By adding nutrients and growth factors, scientists kept the brainlike tissue alive in an incubator for two months, at which point they experimented on it.

Adding a neurotoxin essentially killed the neurons, as it would in a real brain. To simulate traumatic brain injury, they dropped disklike weights from different heights.

That was a key experiment, Hickman said, because “if you take a real brain and you start whacking it, you should get the same forces and the same stretching.”

Kaplan said the brain-in-a-dish “didn’t go splat,” but reacted like “a kitchen sponge, and it would compress down and then partially spring back up.”

He said measurements of glutamate, a neurotransmitter that surges in brain injury, showed that “the more severe the damage, the higher the spike” in glutamate.

Gordana Vunjak-Novakovic, a biomedical engineering professor at Columbia who has collaborated with Kaplan on other studies, described the model as a kind of “Lego approach,” a “modular structure” that can be expanded and made more complex.