New findings support a controversial hypothesis about the biological role of sleep: Snoozing may be a way for the brain to clear clutter accumulated after a hard day of synapse forming and strengthening. Two Science studies published today suggest that the brains of sleeping Drosophila undergo an overall depression in synaptic strength and number, eliminating some minor neuronal connections while merely weakening stronger ones.

Expression of synaptic markers is low after sleep (left) and high after waking (right) in most regions of the fly brain Image courtesy of Chiara Cirelli

"Essentially you're reducing the signal to noise ratio," Paul Shaw, a Washington University neurobiologist and lead author on one of the papers, told The Scientist.

The hypothesis was first proposed about five years ago by Chiara Cirelli and Giulio Tononi, neurobiologists based at the University of Wisconsin-Madison and coauthors on the other Science paper. (You can read more about their work in the April issue of The Scientist.)

"The idea is that when we are awake, we are always learning. Synapses all over the brain get bigger and stronger," Cirelli said. "The system can get saturated by learning quite easily." Cirelli explained that the brain must pare down the strength and number of all synaptic connections, weak and strong, because of limited space and resources in the organ. "We think that a very fundamental function of sleep is this downscaling."

"Their paper probably provides the best support for that hypothesis that I've seen yet from their lab," Marcos Frank, a University of Pennsylvania neurobiologist not involved with either study, told The Scientist.

Both the University of Wisconsin team and the Washington University team studied sleep-deprived fruit flies that had been subjected to enhanced social environments, and found that less rested flies showed less synaptic downscaling.

Shaw's lab bred fruit flies--which, like humans, tend to sleep more when subjected to socially enriched environments--that contained mutant versions of three genes required for learning and memory. The mutant flies did not exhibit the typical increase in sleep after social enrichment (a complex environment that included several sibling flies), while flies in which the function of the mutant genes were rescued got sleepier. The mutants may not have needed the sleep, explained Shaw, because they were unable to form robust synaptic connections in the sensory-rich environment.

Shaw's group also showed that the brains of socially isolated flies contained fewer synaptic terminals than flies subjected to social enrichment, and that the number of terminals decreased in flies that were allowed to sleep. "I think our data shows the first signs of real structural changes," he said.

"[Shaw's] paper is particularly striking because he's isolating, circuit by circuit, cell by cell, gene by gene, the mechanism involved in the increase in sleep that happens after you learn something," Frank said.

In their paper, Cirelli, Tononi and coauthor Giorgio Gilestro showed increases in five proteins central to synapse formation across the fly brain during waking, and decreases in the proteins during sleep; sleep-deprived flies' synaptic protein levels remained high. The results echoed similar findings in rat brains that the group reported last year.

Shaw's team found newly-formed synapses in the brains of fruit fly's kept in a stimulus rich environment. The cells are highlighted on the upper right side of the image. Image courtesy of Washington University School of Medicine

"I think this is the most convincing evidence that a correlate of sleep is the deemphasizing of these synaptic markers," Frank said. "Whether this is something that is really central about sleep," he added, "I'm not so sure."

Jan Born, a neuroscientist at the University of Lübeck in Germany who was not involved with either study, pointed out that some synapses are actually upscaled and that activation levels for some genes related to memory formation increase during sleep. "In fact," he said, "the more weakly encoded behaviors benefit more from sleep than those encoded more strongly," he said. "Behaviorally, you don't have any evidence for a forgetting during sleep."

Both Shaw and Cirelli concede that many more questions remain to be answered with regard to how their observations relate to behavioral states. The mechanism behind synaptic downscaling during sleep is also unclear.

Cirelli and Tononi previously proposed that the slow electrical waves, or oscillations, that sweep over the sleeping brain of mammals might drive such downscaling, but in Drosophila, such "slow wave sleep" has not been concretely identified. Perhaps, Cirelli suggested, the process is instead regulated by fluctuating levels of neural chemicals. Then again, the sleep-related fluctuations seen by both research teams could be the result of passive processes, according to Shaw.

Shaw and his colleagues plan to devise a way to tag individual synapses to track what happens to them during wakefulness and sleep. Cirelli and her team will further explore the morphological changes in Drosophila brain regions beyond the ventral lateral neurons that Shaw explored.

Ravi Allada, a neurobiologist at Northwestern University who studies circadian clock genes in Drosophila brains, said that understanding why sleep is biologically necessary may be a long way off, but working in simple model organisms is a good place to start. "I think we're going to understand why fruit flies sleep in the next ten years," he told The Scientist. "That's going to tell us a lot about why other things sleep."


Related stories:

Disappearing Before Dawn
[April 2009]

The Gears of the Sleep Clock
[April 2009]

Unlocking the clock
[June 2008]