Her attacks of acute sleepiness could occur anywhere: in the car (she no longer took the wheel), at the dinner table, in the kitchen. And if Katherine felt any strong emotion—if she was angry or if someone said something so funny that she laughed out loud—she might have an episode of cataplexy: She would lose control of all her muscles and collapse on the floor, still awake but seemingly paralyzed.

Not surprisingly, people with narcolepsy (about 150,000 in the United States alone) have a terrible time finding or keeping jobs; falling asleep at one's desk is not likely to impress employers. But they have other problems, too. Unless bolstered by strong medications, which present their own difficulties, many patients won't carry anything breakable lest they drop it during a sleep attack. They dare not stand on a chair to change a light bulb, for fear of breaking their own necks. They don't take part in sports because the excitement of winning or losing might trigger an attack of cataplexy at the height of the game. Even fishing may be dangerous, because they could fall in the water while asleep or paralyzed. And public speaking is usually out of the question.

Until very recently, scientists who sought to understand this intriguing and unique condition—narcolepsy is the only neurological disorder that specifically affects sleep—found its cause elusive. But two separate teams of researchers, using totally different methods, have now uncovered the roots of the ailment and, incidentally, galvanized the whole field of sleep research.

One team was led by HHMI investigator Emmanuel Mignot at the Stanford University School of Medicine in Stanford, California; the other was headed by HHMI investigator Masashi Yanagisawa at the University of Texas Southwestern Medical Center at Dallas. Their discoveries, which complement each other like a lock and key, involve the genes that, when mutated, cause narcolepsy.

NARCOLEPTIC DOBERMANS
Mignot, a French physician-scientist, moved from Paris to Stanford in 1986 "because they had this unique colony of narcoleptic dogs," he says—dozens of Doberman pinschers that the famous sleep researcher William C. Dement had started breeding in 1973. Unlike most human narcoleptics, these dogs inherited the disorder from their parents, yet their symptoms were quite human-like. The Dobermans inexplicably fell asleep in the middle of activities. They also lost muscle control and flopped down in cataplectic attacks when they became too excited. This made for unusual problems when the scientists tried to mate the narcoleptic dogs: The dogs tended to collapse at the critical moments. "It took much longer than with normal dogs," Mignot recalls, "but finally the dogs succeeded."

In his early experiments, Mignot gave the dogs various drugs and examined their effect on the animals' brains and behavior. "But I soon realized these drugs were not going to lead me to the cause of narcolepsy," he says. "For that I needed to isolate the gene involved."

Tracking down a gene mutation in dogs in the 1980s was enormously difficult. Only a few human disease genes had been identified by positional cloning (solely through the gene's position on a chromosome, without knowing what protein the gene encoded), as Mignot planned to do. To make matters worse, little was known about the genes of dogs. But he persevered for more than a decade, propelled by one important fact: Because the Dobermans inherited narcolepsy directly from their parents (as a recessive trait, meaning that a dog had to receive a bad copy of the gene from both parents), Mignot could be sure he was looking for a single gene. "I knew I'd eventually find this gene in dogs," Mignot says, "but I thought it would take a maximum of 5 years—not 10!"

After many ups and downs, his team isolated a marker, a section of DNA that was found only in the narcoleptic dogs and therefore had to contain the faulty gene. "It was a huge portion of dog chromosome 12," he recalls. The researchers slowly made their way along this stretch of DNA, seeking the guilty gene with the help of several hundred DNA-sequencing machines. Finally, in 1999, they zeroed in on a gene that, when normal, encodes a receptor that sits on the surface of nerve cells and brings news of incoming small peptide hormones, called neuropeptides. When mutated, the gene (called Hypocretin receptor 2, or Hcrtr2) was clearly responsible for the dogs' narcolepsy.

PATHS MERGE
A year earlier, while working with mice, Masashi Yanagisawa had discovered the gene for a neuropeptide that binds this receptor—but he did not know it had anything to do with narcolepsy. Both groups were astounded at what Mignot calls "this amazing convergence." It was especially surprising because Yanagisawa had not even planned to study sleep.

Yanagisawa's goal was to find previously unknown chemical messengers in the body. He was particularly interested in the neuropeptides, which regulate such essential functions as circulation, eating and sleep, and he had figured out an ingenious way to seek them out. First he looked through the newly sequenced human genome and picked out several hundred genes whose DNA followed a particular pattern—that of the G protein–coupled receptors, which usually play important roles in the brain—but whose specific functions were still a mystery. Then he inserted these genes into mouse cells, signaling the cells to produce the corresponding receptor proteins. The newly made receptors attracted neuropeptides, which then bound to them.

It was while "fishing" for neuropeptides in this way that Yanagisawa discovered a pair of them that seemed to control appetite. When he injected these neuropeptides into the brains of rats, "this caused them to eat much more," he recalls. So he named these agents "orexins A and B," after the Greek word for appetite.

His group then generated mice whose orexin genes had been knocked out. Nothing dramatic happened to the mice during the day, which is when mice sleep. But when the researchers used infrared video to record the activities of these mice at night, when the animals normally feed and play, they noticed that the mice would unaccountably collapse, seemingly asleep for a minute or so, many times during the night.

"At first we thought it was some kind of seizure disorder," says Yanagisawa. "Could it be petit mal epilepsy without the shaking? Then we realized these mice actually had a sleep disorder that looked like narcolepsy."

Despite their different names, Yanagisawa's orexins and Mignot's hypocretins soon proved to be one and the same. "We're very glad that we contributed to discovering the fundamental molecular cause of narcolepsy," says Yanagisawa. "Now we can try to develop a fundamental cure for it."

BETTER TREATMENTS
The standard treatment for narcolepsy used to be stimulant drugs such as Dexedrine, a type of amphetamine. These stimulants did keep people awake, but they also produced jitteriness and interfered with nighttime sleep. As a result, patients often took sleeping pills, and many of them ended up addicted to both kinds of drugs. A more recent stimulant drug, modafinil, works better, but its molecular targets are unknown. Furthermore, modafinil does not prevent attacks of cataplexy, for which patients must take tricyclic antidepressants (which have their own side effects).

Orexin-like drugs would be far more specific and effective, Yanagisawa believes. To test the power of orexin, his team injected some of it directly into the brains of knockout mice that could not make their own. "The mice stayed awake continuously and had no cataplexy," he reports.

The ideal cure would be to swallow some orexin/hypocretin and have it travel to the brain, Yanagisawa explains. But like other neuropeptides, orexin would be broken down by the body before reaching the bloodstream—and even if injected intravenously, it would not cross the blood/brain barrier. "What we need is a small molecule that would mimic the activity of orexin but would have a different chemical structure so as to overcome both of these obstacles," he suggests, adding that drug companies are very good at finding such molecules.

Meanwhile, Mignot and his associates discovered that about 90 percent of people with narcolepsy have no orexin/hypocretin at all in their cerebrospinal fluid. "We were stunned," he says, "particularly because we couldn't find any mutations in the patients' hypocretin genes." Autopsies on the brains of deceased narcoleptic patients told a similar tale: Nearly all the hypocretin-secreting cells were missing. "This suggests that the cells that normally secrete hypocretin in the brain have been selectively killed or have degenerated," Mignot says. "So we may be able to treat patients by replacing the hypocretin molecule with a drug, but the patients would have to keep on taking this drug. Longer term, we need to figure out what is killing these cells and then try to prevent it from happening."

AN AUTOIMMUNE DISORDER?
In humans, the disorder seems to strike randomly. "The large majority of my patients have no relatives with narcolepsy," says Mignot, who sees patients regularly at the Stanford Center for Narcolepsy. "There may be an inherited predisposition, but it is very small. And when one identical twin has narcolepsy, the other one usually does not. So something else must be involved."

That "something else," Mignot speculates, may be "an autoimmune reaction." A decade before Mignot began his research on narcolepsy, a Japanese group found a close association between cases of narcolepsy and a particular variant, or allele, of HLA (histocompatibility antigen, which can bring on a transplant rejection when the donor and recipient are mismatched) called HLA-DR2. This was discovered "by pure luck," Mignot says, though it did not lead anywhere at the time.

Taking it one step further, Mignot and his associates studied 420 patients and 1,082 controls from three ethnic groups (white Americans, African Americans and Japanese). The research confirmed that people with narcolepsy are much more likely to carry the HLA-DR2 variant than those who don't. The most striking association, however, was with another HLA variant located close to the first one on chromosome 6, HLA-DQB1*0602. People who inherit this allele from both parents are two to five times more likely to develop narcolepsy than those who inherit it from only one parent.

These findings resemble what is known about several autoimmune diseases such as multiple sclerosis and type 1 diabetes, which also are associated with particular variants of HLA. As Mignot points out, all these diseases start in adolescence or early adulthood, and all involve the selective destruction of particular classes of cells. "If the immune system is involved in narcolepsy, we need to find out how it works," he says, "so we can find a way to stop the autoimmune process before it is too late."

Mignot has also started experiments with zebrafish, the tiny multicolored fish that have become a prized model organism for geneticists. Zebrafish have two advantages over dogs: Their genome has been almost entirely sequenced, and it is easy to produce specific mutations in them. They also have a big advantage over mice: "They produce more progeny," Mignot says. He found that zebrafish have normal receptors for hypocretin/orexin, as well as 20 cells that secrete it. He is trying to create single-gene mutants in which these 20 cells are somehow killed, which might enable him to locate the source of the killing.

ALTERED STATES
Yanagisawa, for his part, has become fascinated with the different stages of sleep. He says that the discovery of orexins has raised new questions about what he calls the three different states of behavior: wakefulness, during which one is alert and can move muscles at will; non-REM (rapid eye movement) sleep, when the sensory system is damped down and muscles are resting; and REM sleep, the dreaming phase that occurs about five or six times a night. This is the time when the brain is most isolated from the environment and muscles are completely limp, as if paralyzed.

Yanagisawa would like to know, for example, why people with narcolepsy fall into REM sleep almost immediately. "The normal progression is to doze from wakefulness into non-REM sleep and to stay there for at least an hour and a half before going into REM sleep," he says. "You never go directly to REM sleep—unless you have narcolepsy."

He recently discovered different roles for the two orexin receptors: One seems to regulate mainly non-REM sleep, while the other controls the onset of REM sleep. "One immediate implication is that if we can find drugs to block or activate these two receptors separately, it may lead to some interesting—and I hope useful—controls of sleep," Yanagisawa says. "What if we could turn REM on, make it last longer, cut it down or suppress it at some point, at will?"

Insomniacs who depend on sleeping pills are deprived of normal sleep, he points out. "The majority of currently available sleeping pills sedate the entire brain, nonselectively, so there is no REM sleep at all. Yet REM sleep seems to be very important." Exactly why it's important remains a mystery, however. "We need to find out what would happen if you specifically increased or decreased the REM state," Yanagisawa suggests. Research of this type "might lead to a totally new class of sleeping pills."

At the same time, research on orexin-like drugs might lead "not only to an instant cure for narcolepsy but to a new family of wake-up pills," he says, that could be lifesavers for airline pilots on extended flights, soldiers on the front lines and tired drivers.

ALTERED AGENDAS
"Why do we need to sleep at all?" Yanagisawa asks. "What is replenished by sleep? That's a big question in biology. A Nobel Prize is waiting for whoever finds the answer. But we do know that sleep is essential for sustaining life in mammals—for example, rats that are totally deprived of sleep die within 10 days or two weeks. It's also one of our most time-consuming activities, as one-third of our lives is spent on sleep."

It is probably no coincidence, he adds, that we spend another third of our lives seeking food or eating it. Some of his experiments have revealed that hungry mice produce more orexin, which "impels them to run around to find food," he says. By contrast, knockout mice that produce no orexin display no such drive, even when they really need to eat.

This brings Yanagisawa to an interesting hypothesis: "Mammals must have some mechanism to switch between sleep and feeding, which are mutually exclusive," he says. "Our theory is that orexin may be this switch, coordinating nutritional status and the sleep/wake cycle."

This could have major implications for research and development. Until now, narcolepsy has been a so-called "orphan disease," generally ignored by the drug companies. "It's way too small a market," Yanagisawa says. But the latest studies could lead to drugs for a much wider audience—all the people who suffer from some kind of sleep disorder. Such medications might help us "fine-tune" our state of alertness, he suggests, and eventually provide ways to remain awake and productive for days on end—on demand.

In effect, we will learn to control the sleep/wake cycle, Yanagisawa predicts, while remedying some basic problems. The sleep-deprived could enjoy real, REM-filled sleep whenever they want to, and those with narcolepsy could lead normal lives.

Download this story in Acrobat PDF format.
(requires Acrobat Reader)

Photo: William Duke

Reprinted from the HHMI Bulletin,
March 2003, pages 8–13.
©2003 Howard Hughes Medical Institute