When Anna’s response to flumazenil appeared to be sustained, Rye decided he needed to pivot away from CART. An opportunity came through the Obama administration’s 2009 economic stimulus package. The American Recovery and Reinvestment Act included a $10 billion boost for the National Institutes of Health.3 A buzzword in Washington, DC, at the time was “shovel-ready,” meaning that a construction project would generate jobs quickly if funded. In the research realm, the concept was applied to projects expected to yield fast results. If scientists had already gathered a large number of samples and wanted extra funding to analyze them, those proposals were given priority. Rye’s successful grant application fit into that framework. The funding climate grew colder after the stimulus package, and he would not receive federal support for hypersomnia research for another six years.

  A SHADOWY SUBSTANCE

  Jenkins and two colleagues at Emory, the pathologist Jim Ritchie and the pharmacologist Mike Owens, had already been probing what the somnogenic substance could be. When they centrifuged CSF from hypersomnia patients through membranes, what came through was still active in the patch clamp assay, indicating that it was smaller than the holes in the membranes. In addition, when CSF was subjected to the enzyme trypsin, which chews up proteins, its patch clamp activity was mostly wiped away.

  Together, these clues suggested that the mysterious substance was a peptide, or small protein. The prospect of a somnogenic peptide resonated with a long line of research in the field, stretching back decades. As one example, celebrated experiments with sleep-deprived goats had identified a sleep-promoting peptide called “Factor S” in their CSF, although it didn’t look like the hypersomnia substance was the same thing. In contrast, before Rye and Jenkins, several groups had gone searching for abnormalities in CSF among people with idiopathic hypersomnia. Some looked for changes in dopamine metabolites; others looked for alterations in the neurotransmitter histamine. This line of research had yielded few returns.

  Over the next few years, Jenkins was guarded about concluding whether CSF from people with hypersomnia had more of an otherwise ubiquitous GABA-enhancing peptide. An alternative was that one of those peptides might be altered and more potent than usual in people with hypersomnia. He and others have described CSF as “the sewer of the nervous system,” since they were analyzing waste products, without knowing their precise origin. CSF components can change when someone has an infection or other brain injury and can also vary depending on the site where the fluid is removed.

  One experiment hinted that Rye and Jenkins’s research might be applicable beyond people with rare sleep disorders. The 2009 patent includes data from rhesus monkeys that Rye and Keating had drawn CSF from at various times of the day: early morning, afternoon, and late evening. (Rye’s work on Parkinson’s had used monkeys; at the time, he was running out of money to maintain them.) After four monkeys had been kept awake and would normally be asleep, the CSF GABA-enhancing activity in two of them increased above the threshold for hypersomnia in human patients. Although it was never published, this preliminary data reveals how Rye was thinking. Part of the patent speculates: “It appears that this substance waxes and wanes in animals as it does in humans during the normal day-night cycle. These promising results indicate that under sleep-deprived conditions, humans may also benefit from flumazenil or other GABAergic therapy to relieve the symptoms of fatigue they experience from the accumulation of this somnogenic compound.”

  Identifying the somnogen in hypersomnia patients’ CSF was a classic biochemistry problem: purifying a protein out of a complex mixture. When Jenkins started to separate human CSF into fractions biochemically, he glimpsed peptides that were present only in Anna’s CSF. The peptides’ sequences were not in biomedical databases, suggesting that what they were hunting had not been seen before.

  At the outset, Jenkins had to establish whether the “sleepy stuff” was stable over time and would not decay. Could he let a sample of CSF sit around at room temperature for a few hours or even overnight? What if he froze it and rethawed it, or dried it in a vacuum? Both were OK. Were measurements the same between experimenters? With patch clamp assays, it was only possible to measure a small number of CSF samples per day.

  Early on, Jenkins compared results with Garcia, and later with his postdoc Amanda Freeman, a former graduate student of Rye’s who was asked to help a couple days per week. Accustomed to painstaking electrophysiology work, Freeman found she could get through measurements of three or four patients per day, if everything was set up correctly. “Working on the rig all day—for me, that was a good time,” she said.

  In contrast to measurements of GABA-enhancing activity, the hunt for the peptides responsible ran into difficulties. After Jenkins retooled his rigs to reduce the flow rate, less CSF was needed per measurement, but CSF from patients was still scarce and valuable. The 2009 grant application included a proposal to measure GABA-receptor activity with fluorescent dyes instead of with patch clamps, which could allow experimenters to process samples more quickly. In the years to come, Jenkins had a tough time convincing graduate students to work on the hypersomnia project; they gravitated toward anesthesia-related projects with more predictable returns.

  Colleagues peppered Jenkins with suggestions, like obtaining large amounts of CSF from animals or trying particular candidate peptides whose attributes seemed relevant. At some point, the decision was made to pool CSF samples from hypersomnia patients. That gave experimenters more volume to work with, but the decision implied the active “sleepy stuff” ingredients in each person were the same. Both Rye and Jenkins would later regard this choice as a mistake.

  Imagine police detectives on television. Someone witnesses a crime but doesn’t know the perpetrator’s identity. So the police ask the witness to describe the person she saw to a sketch artist. How tall was he? What was he wearing? Like the sketch artist, Jenkins and his colleagues had a few clues: the GABA-enhancing substance’s approximate size, what it was made of, and its sensitivity to flumazenil. But they didn’t know what it was. The perpetrator, or perpetrators, had escaped into the crowd.

  CUSTOMIZED EQUIPMENT

  A few years later, when Rye gave a seminar at Emory, the puzzle captured my imagination. From my own biochemistry experience, I guessed the process could be a hard slog. Patch clamping has a reputation for difficulty. Much lore and special tricks go along with the technique. For many years, it resisted being automated, unlike other processes in molecular biology, such as DNA sequencing. A graduate student in California engaged in patch clamp wrote in 2013: “Patch-clamp bugs require a personality with the highest tolerance for frustration I’ve ever encountered.”4

  Practitioners of the technique have retained some elements of craft (figures 6.1, 6.2). The glass tubes needed for patch clamping are melted and stretched in the lab, rather than being manufactured and delivered, because of fragility and the possibility of introducing dust. A manual on patch clamping advises scientists to try different locations in a building, seeking the place where the floor vibrates the least.5 Sometimes, the experimenter surrounds the entire setup with a metal Faraday cage to screen out electrical influences from other pieces of equipment. Some patch clampers fire-polish their pipettes or apply bleach to the silver wire that measures current, aiming to leave just the right coating of chloride on the silver.6

  FIGURE 6.1. Patch clamp rig in Jenkins’s laboratory at Emory, 2018.

  FIGURE 6.2. Patch clamp rig in Jenkins’s laboratory at Emory, 2018.

  I asked Jenkins to visit his lab to see what he and others had been dealing with. In the corner of a large research building, his lab had two active rigs, with components held in place with aquarium glue and glass pipettes resting on modeling clay in covered dishes. Several Post-it notes next to the rigs compiled a list of procedural reminders. A rig included a microscope, a micromanipulator for bringing the glass pipette in contact with the target cell, electrical signal amplifiers, and a thicket of plastic syringes and flexible tubing. Adding up all the components, each rig cost about $100,000.7 “A rig is a customized piece of equipment, which is constantly being torn down and rebuilt to do different experiments,” Jenkins told me.

  I was surprised to learn that patch clampers use their mouths to control suction through the glass pipette. In my graduate school years, fledgling scientists were admonished to avoid “mouth pipetting,” because of the risk of ingesting some harmful chemical. Patch clampers use their mouths because their hands are busy positioning the glass tip and managing electrical equipment. Jenkins chuckled when asked about this, saying that the cheeks have more sensory nerve endings and provide better control. “It’s also partly personal preference,” he said. “I never had success with a syringe in my hand. Using my mouth just gives a better feel.”

  Olivia Moody, who would earn her doctorate working with Jenkins, showed me how she established contact between the glass pipette and the cell membrane, applied suction, then waited for the electrical resistance to climb to gauge whether the seal is good enough. When I asked about the difficulty of patch clamping, she shrugged, smiled, and said: “Mice don’t always do what you want either.”

  SLEEPY PUPPIES

  A century before, scientists in France and Japan conducted experiments with sleep-deprived animals, believing that a sleep-inducing substance, or “hypnotoxin,” accumulated in their bodies. They put their subjects into extreme, cruel situations, outside the rhythms of day and night. In a way, they were following Koch’s postulates: viewing sleepiness like a disease-causing microbe that could be isolated and transferred to another animal. In several countries, authorities overseeing animal welfare would discourage such experiments today.8

  Starting in 1907, the Japanese physiologist Kuniomi Ishimori and a host of medical students in Nagoya kept puppies awake for up to 113 hours, “by whipping the animals via ropes or by other coercive means.”9 Ishimori’s group proposed that the puppies’ brains contained substances that induced sleep when injected into fresh animals. Their experiments had limitations, since their whole-brain extracts also provoked shivering, salivation, and tear production in recipients. Still, only the sleep-deprived puppies’ brains, and not those of animals allowed to rest, contained the sleep-inducing factor, they reported.

  A few years later in France, the physiologists Henri Pieron and Rene Legendre had dogs constrained by leashes so they were unable to lie down. Pieron and Legendre isolated various body fluids from the dogs and discovered that CSF from sleep-deprived dogs contained a sleep-promoting activity. The effect could be seen when the CSF was injected directly into the brains of rested dogs. Pieron and Legendre concluded: “A prolonged waking state produces in the organism, probably as a waste product of cerebral metabolism, the formation of a toxin that causes local cellular lesions and a pressing need for sleep, phenomena which are obviously related.”

  Nathaniel Kleitman, a pioneer of sleep research in the United States, cited Pieron’s work as an inspiration. In another groundbreaking study, Pieron tracked the body temperature rhythms of nurses working the night shift at an asylum. His 1913 book Le probleme physiologique du sommeil was considered one of the twentieth century’s most influential scientific studies of sleep, because its focus was distinct from philosophical introspection or questions of the soul.10

  Nobody today knows what the hypnotoxin was. Pieron speculated that it might be responsible for narcolepsy, which is unlikely, given what we know now. Techniques for biochemically characterizing sleep-inducing substances were then primitive, and Ishimori’s and Pieron’s were probably different. But the concept was influential. Previous experimenters had kept animals awake with exercise, making it difficult to separate fatigue from sleepiness. While Ishimori was little noticed in Europe or the United States, Pieron’s experiments with dogs spurred a series of efforts to repeat and extend their results.

  If we look at the history of scientists’ search for molecules connected with sleep, quests like Pieron’s have led to unexpected insights—or dead ends. Some of the molecules now considered the most important for regulating sleep were hiding in plain sight or were discovered by scientists who were not primarily investigating sleep.

  Like their predecessors, Rye and Jenkins were making a bet that hypersomnia patients had something inside them embodying their sleepiness—and they could extract it. However, what the Emory investigators were doing was somewhat different conceptually. Rye and his team weren’t deliberately putting people with narcolepsy and IH under the stress of sleep deprivation. They thought that what they were looking for was more fixed and long-lasting. Don Bliwise put the distinction this way: “It depends on whether you see sleepiness as a temporary state or a permanent trait. If you see overall need for sleep as a lasting trait, it makes a lot of sense to probe people with hypersomnia and figure out what distinguishes them.”

  PAPER CLIPS AND RUBBER BANDS

  In 1939, researchers at Northwestern University in Chicago reported they had confirmed Pieron’s observations regarding sleep deprivation in dogs. However, they argued that “the normal picture of sleep is not produced,” because manipulations of CSF produced changes in both body temperature and intracranial pressure. The challenge emerged of untangling the processes that induce healthy sleep from those producing sleepiness driven by fever or illness.

  Physiologists at Harvard led by John Pappenheimer followed Pieron in the 1960s, using goats instead of dogs.11 The shape and thickness of bone at the rear of goats’ skulls allowed the Harvard scientists to implant nylon tubes and repeatedly withdraw CSF without anesthesia. The goats were conditioned, with electric shocks initially, to avoid lying down in the laboratory and thereby kept awake for forty-eight to seventy-two hours.12

  Pappenheimer wrote in Scientific American that his lab’s initial results, obtained by injecting goats’ CSF into cats, “were so striking and of such potential physiological significance that we put other plans aside in order to devote full time to the systematic exploration of the Pieron phenomenon.”13 When extracted CSF was injected into animals’ brains, it produced something close to natural sleep, with EEG signatures resembling those observed after the animals were sleep deprived.

  Pappenheimer and his colleagues began trying to concentrate and purify what they dubbed “Factor S.” An old-fashioned scientist, Pappenheimer preferred bicycles to cars and didn’t like making long-distance phone calls. He was “parsimonious in quite an English way in that he wanted to do things with paper clips and rubber bands.”14 “Each assay required 6 hours of recording on two or more rabbits fitted with implanted guide tubes and EEG electrodes,” Pappenheimer said in a 1982 lecture. “This part of the research took several years, and there was very little to show for it until there was enough pure product to analyze chemically.”

  Like Rye and Jenkins, the Harvard group could determine the substance’s size and that it was a peptide. Over three years, they collected about six liters of CSF from a colony of goats, but this wasn’t enough. They moved on to extracting whole brains of goats, sheep, and rabbits but estimated they would need thousands of sleep-deprived animals. Eventually they turned to human urine, which was collectable at a sufficient scale.15

  Once they had enough and could confirm that urinary Factor S behaved chemically the same as CSF Factor S, they teased Factor S apart and found it contained surprising components: muramyl peptides, which come from bacterial cell walls and not from mammalian cells. Pappenheimer suggested that the bacterial components “may be regarded as akin to any of the essential amino acids or vitamins which cannot be synthesized by mammalian cells.” “There was a lot of skepticism,” said James Krueger, who began working with Pappenheimer at Harvard in the 1970s. “I’d go to meetings, and people would be shocked at the idea that something from gut bacteria could be acting in the brain. They thought it was contamination. Grant applications came back rejected for the same reason.”16

  How the microbes that live in our intestines, on our skin, and in other parts of our bodies contribute to our metabolism and brain function was underappreciated at the time. Since they are bits of bacteria, muramyl peptides are highly stimulatory for the immune system and have been used as vaccine-enhancing adjuvants. Krueger proposed that the bits were produced through processing by macrophages, white blood cells that engulf and digest bacteria. In animal experiments, muramyl peptides brought on fever, which could be alleviated with acetaminophen, without blocking their sleep-promoting effects.

  Muramyl peptides were potent, but they weren’t as fast as cytokines—inflammatory messengers that arouse the immune system—at inducing sleep when injected into the brain. So the bacterial bits were thought to act indirectly. Throughout the 1980s and 1990s, more players entered into Factor S’s proposed sleep-triggering mechanism. “When I started with Pappenheimer, we were working with a simple idea,” Krueger said. “He thought there was a kind of sleep hormone, a single sleep factor. Now it has become clear that there are many molecules involved.”

  IMPLICATIONS FOR IH

  The most straightforward interpretation of Factor S was that it represented sleepiness associated with infection or inflammation. This is consistent with everyday experience; a mild infection or fever generally enhances sleepiness, pushing down REM sleep in favor of non-REM, but a more intense fever can disrupt someone’s sleep. The lethal effects of sleep deprivation in animal experiments may have been caused by bacterial invasion of internal organs.17 Yet Krueger and his colleagues have accumulated evidence that cytokines induced by Factor S also have an everyday physiological role, outside of fever or extreme sleep deprivation. Their levels vary rhythmically, rising at night, and tweaking their levels in mice can modulate how much the animals sleep.18