For decades, scientists have proposed that learning occurs and memories are stored when connections among nerve cells are weakened or strengthened, but there's been no direct way to prove it.

Now, a Johns Hopkins study using mouse cells reveals what seems to be the very last step that occurs as nerve cells temporarily weaken their connections. In the June 13 issue of Science, the Johns Hopkins team also reports that blocking this step prevents connections from weakening without affecting anything else, making it possible--finally--to see if weakening connections really do contribute to learning and memory.

"Our finding defines what is happening during this process, called long term depression, and offers the first opportunity to engineer a mouse that will allow us to really examine carefully its role in learning motor skills," said Richard Huganir, a Howard Hughes Medical Institute investigator and a professor of neuroscience at the School of Medicine. "There is a lot of controversy about whether this and a related process really underlie learning and memory because there's been no way to test the idea directly."

The crucial last step revealed in the Johns Hopkins work is a single, tiny modification of a protein called GluR2, which helps brain cells detect the chemical glutamate. By preventing that modification, weakening didn't occur, and the mouse neurons stayed in touch with their neighbors when they shouldn't have.

"The beauty of having the nitty-gritty detail is that now we can create a mouse with just that single change and see what happens to its behavior," said co-author David Linden, professor of neuroscience.

Nerve cells "talk" to one another with the help of chemicals like glutamate. One neuron produces the chemical and sends it across to a neighboring cell where it latches on and creates a cascade of events inside the cell. In response to certain patterns of stimulation, a nerve cell will pull the binding spots for glutamate into the cell, weakening its connections and ending the "conversation."

For 40 years, enzymes known as protein kinases, which modify other proteins, have been known to be involved in nerve cells' ability to "unplug the phone"--long term depression, or LTD--and the opposite effect, long term potentiation, or LTP. In those early experiments, nerve cells missing certain protein kinases failed to respond properly in the lab, and mice missing the same enzymes couldn't learn or remember normally. But those experiments aren't good enough to link LTD and LTP to learning, said the researchers.

"The problem is that protein kinases are not specific to LTD and LTP--they literally modify thousands of proteins and affect many processes," said Linden, whose research focuses on the molecular goings-on of unusual, sea-fan-shaped neurons called Purkinje (pur-KIN-jee) cells. "The holy grail of this science is to link a molecule to a behavior and understand all the steps in between. We're not there yet, but we're a whole lot closer."

GluR2 firmly connects protein kinases to LTD in Purkinje cells, but it will take a genetically engineered mouse to make the final connection between GluR2 and behavior--or not. Developing a mouse with the appropriately mutated GluR2 may take a year or more, noted Huganir, whose M.D.-Ph.D. student Jordan Steinberg has taken on the project.

In their current study, graduate student Hee Jung Chung and Steinberg created genetic instructions to build two mutant versions of GluR2 that prevented a specific modification (phosphorylation) by a protein kinase. The scientists then coated microscopic beads of gold with the genetic instructions for mutant GluR2 and green fluorescent protein. Using a "gene gun," the scientists shot the beads into neurons from mice missing GluR2.

Linden tested the reactions of glowing neurons, which also had the mutant GluR2, to stimulations known to induce LTD. "LTD was completely, absolutely, 100 percent gone with either change to GluR2," he said, noting that the GluR2 pathway is likely just one of many that allow nerve cells to undergo LTD.

If GluR2 modification does play a role in learning and memory, it is most likely to do so for simple motor skills, since the Purkinje neurons used in this study originate in the cerebellum, the home of motor skills in the brain.

Authors on the report are Chung, Steinberg, Huganir and Linden. Chung is now at the University of California at San Francisco. The work was funded by the Howard Hughes Medical Institute, the National Institute of Mental Health and the Develbiss Fund.