 |
PAGE 2 OF 2
High magnification microscopy allows visualization of individual spines on a neuronal dendrite.
Now it was time to do some reverse engineering to find out why spines were "filtering," or impeding, incoming nerve signals. Why, after all, would evolution select an architecture that diminished incoming signals at the very point where they were received? It occurred to them that "maybe the logic of the design is to enable the nerve cell to add arithmetically, just as you would teach a child to do," Yuste explains. "The spine neck could serve to electrically isolate inputs, thus enabling the soma to add every input without interference."
They then found that, when two spines on the same dendrite were simultaneously stimulated, the voltage they conveyed to the soma was precisely the sum of their signals. At the same time, when two regions devoid of spines were stimulated, they interfered with each other, and the resulting voltage was much smaller than the sum of the two inputs. Thus, not only do spines compartmentalize calcium to regulate synaptic strength, but they also help the neuron accurately add inputs.
The implication, Yuste hypothesizes, is that dendritic spines enable cortical neurons to work in a linear fashion and serve as adding machines. In a brain marked by a "distributed" circuitry, in which neurons sample incoming information over the widest possible area, a mechanism that accurately sums many signals would gather all possible information.
|
|
|
|
|
 |
|
|
|
|
|
HHMI investigator Rafael Yuste is driven by more than a theoretical interest in the structure and function of the cerebral cortex. As a physician-scientist, he also wants to see what happens when cortical circuits malfunction and what clinicians can do about it.
When he started his lab at Columbia, Yuste, took an interest in epilepsy, a major disease involving the cortex. The first to use calcium imaging to visualize a circuit, Yuste and graduate student Tudor Badea "were able to image the spread of an epileptic seizure, neuron by neuron, for the very first time," Yuste says. More recently, Yuste and postdoc Andrew Trevelyan (also a physician-scientist) carried out experiments aimed at discovering how epileptic seizures—or at least their analogs in mice—spread through cortical territory. Their findings, reported over the past year in the Journal of Neuroscience, showed that seizures advance in a series of small steps controlled by a specialized type of inhibitory neuron.
"When they are working properly and firing, these inhibitory neurons prevent epilepsy from spreading," says Yuste. "You can think of them as policemen, each responsible for a 'neighborhood' of cortical neurons." Three or four officers might patrol a beat encompassing some 1,000 neurons. When they detect an approaching epileptiform wave, the inhibitory neurons fire, discharging γ-aminobutyric acid (GABA), a neurotransmitter, directly into the cell bodies of neighborhood neurons.
So long as they have a reserve of GABA, the neuronal police can stop the wave. A person with epilepsy might experience this as a seizure whose effects are localized. But if the GABA-releasing neurons exhaust their supply, the wave passes through, recruits nearby neurons, and prepares to invade the next territory—leading to a full-blown seizure.
While cautioning that the work to date involves mouse cells, Yuste believes it could eventually lead to new targets for better treatments in people. "Unfortunately, in 2007 the only 'cure' for severe epilepsy is still neurosurgery," he says. "But instead of cutting out a whole section of the cortex, we might find a way of stimulating this specific class of inhibitory cells so that they can stop progression of a seizure."
—P.T.
|
|
|
|
|
|
|
|
|
Yuste's speculation goes further. Perhaps, based on simple addition, nature found algorithms that could be used to build a diverse set of mental functions. Yuste aims to demonstrate this notion in human tissue. If he succeeds, linear summation would be shown to be the elegant operation that renders the cortex a kind of universal computer, able to mobilize elementary math to accomplish such complex (and, paradoxically, nonlinear) ends as thinking, remembering, and imagining. 
Photo: Yuste lab
|
|
|
Back
